Cellpose API Guide

Cellpose class

class cellpose.models.Cellpose(gpu=False, model_type='cyto3', device=None)[source]

Main model which combines SizeModel and CellposeModel.

Parameters
  • gpu (bool, optional) – Whether or not to use GPU, will check if GPU available. Defaults to False.

  • model_type (str, optional) – Model type. “cyto”=cytoplasm model; “nuclei”=nucleus model; “cyto2”=cytoplasm model with additional user images; “cyto3”=super-generalist model; Defaults to “cyto3”.

  • device (torch device, optional) – Device used for model running / training. Overrides gpu input. Recommended if you want to use a specific GPU (e.g. torch.device(“cuda:1”)). Defaults to None.

device

Device used for model running / training.

Type

torch device

gpu

Flag indicating if GPU is used.

Type

bool

diam_mean

Mean diameter for cytoplasm model.

Type

float

cp

CellposeModel instance.

Type

CellposeModel

pretrained_size

Pretrained size model path.

Type

str

sz

SizeModel instance.

Type

SizeModel

eval(x, batch_size=8, channels=[0, 0], channel_axis=None, invert=False, normalize=True, diameter=30.0, do_3D=False, **kwargs)[source]

Run cellpose size model and mask model and get masks.

Parameters
  • x (list or array) – List or array of images. Can be list of 2D/3D images, or array of 2D/3D images, or 4D image array.

  • batch_size (int, optional) – Number of 224x224 patches to run simultaneously on the GPU. Can make smaller or bigger depending on GPU memory usage. Defaults to 8.

  • channels (list, optional) – List of channels, either of length 2 or of length number of images by 2. First element of list is the channel to segment (0=grayscale, 1=red, 2=green, 3=blue). Second element of list is the optional nuclear channel (0=none, 1=red, 2=green, 3=blue). For instance, to segment grayscale images, input [0,0]. To segment images with cells in green and nuclei in blue, input [2,3]. To segment one grayscale image and one image with cells in green and nuclei in blue, input [[0,0], [2,3]]. Defaults to [0,0].

  • channel_axis (int, optional) – If None, channels dimension is attempted to be automatically determined. Defaults to None.

  • invert (bool, optional) – Invert image pixel intensity before running network (if True, image is also normalized). Defaults to False.

  • normalize (bool, optional) – If True, normalize data so 0.0=1st percentile and 1.0=99th percentile of image intensities in each channel; can also pass dictionary of parameters (see CellposeModel for details). Defaults to True.

  • diameter (float, optional) – If set to None, then diameter is automatically estimated if size model is loaded. Defaults to 30..

  • do_3D (bool, optional) – Set to True to run 3D segmentation on 4D image input. Defaults to False.

Returns

tuple containing
  • masks (list of 2D arrays or single 3D array): Labelled image, where 0=no masks; 1,2,…=mask labels.

  • flows (list of lists 2D arrays or list of 3D arrays):
    • flows[k][0] = XY flow in HSV 0-255

    • flows[k][1] = XY flows at each pixel

    • flows[k][2] = cell probability (if > cellprob_threshold, pixel used for dynamics)

    • flows[k][3] = final pixel locations after Euler integration

  • styles (list of 1D arrays of length 256 or single 1D array): Style vector summarizing each image, also used to estimate size of objects in image.

  • diams (list of diameters or float): List of diameters or float (if do_3D=True).

CellposeModel

class cellpose.models.CellposeModel(gpu=False, pretrained_model=False, model_type=None, diam_mean=30.0, device=None, nchan=2)[source]

Class representing a Cellpose model.

diam_mean

Mean “diameter” value for the model.

Type

float

builtin

Whether the model is a built-in model or not.

Type

bool

device

Device used for model running / training.

Type

torch device

mkldnn

MKLDNN flag for the model.

Type

None or bool

nchan

Number of channels used as input to the network.

Type

int

nclasses

Number of classes in the model.

Type

int

nbase

List of base values for the model.

Type

list

net

Cellpose network.

Type

CPnet

pretrained_model

Full path to pretrained cellpose model(s).

Type

str or list of strings

diam_labels

Diameter labels of the model.

Type

numpy array

net_type

Type of the network.

Type

str

__init__(self, gpu=False, pretrained_model=False, model_type=None, diam_mean=30., device=None, nchan=2)[source]

Initialize the CellposeModel.

eval(self, x, batch_size=8, resample=True, channels=None, channel_axis=None, z_axis=None, normalize=True, invert=False, rescale=None, diameter=None, flow_threshold=0.4, cellprob_threshold=0.0, do_3D=False, anisotropy=None, stitch_threshold=0.0, min_size=15, niter=None, augment=False, tile=True, tile_overlap=0.1, bsize=224, interp=True, compute_masks=True, progress=None)[source]

Segment list of images x, or 4D array - Z x nchan x Y x X.

eval(x, batch_size=8, resample=True, channels=None, channel_axis=None, z_axis=None, normalize=True, invert=False, rescale=None, diameter=None, flow_threshold=0.4, cellprob_threshold=0.0, do_3D=False, anisotropy=None, stitch_threshold=0.0, min_size=15, niter=None, augment=False, tile=True, tile_overlap=0.1, bsize=224, interp=True, compute_masks=True, progress=None)[source]

segment list of images x, or 4D array - Z x nchan x Y x X

Parameters
  • x (list, np.ndarry) – can be list of 2D/3D/4D images, or array of 2D/3D/4D images

  • batch_size (int, optional) – number of 224x224 patches to run simultaneously on the GPU (can make smaller or bigger depending on GPU memory usage). Defaults to 8.

  • resample (bool, optional) – run dynamics at original image size (will be slower but create more accurate boundaries). Defaults to True.

  • channels (list, optional) – list of channels, either of length 2 or of length number of images by 2. First element of list is the channel to segment (0=grayscale, 1=red, 2=green, 3=blue). Second element of list is the optional nuclear channel (0=none, 1=red, 2=green, 3=blue). For instance, to segment grayscale images, input [0,0]. To segment images with cells in green and nuclei in blue, input [2,3]. To segment one grayscale image and one image with cells in green and nuclei in blue, input [[0,0], [2,3]]. Defaults to None.

  • channel_axis (int, optional) – channel axis in element of list x, or of np.ndarray x. if None, channels dimension is attempted to be automatically determined. Defaults to None.

  • z_axis (int, optional) – z axis in element of list x, or of np.ndarray x. if None, z dimension is attempted to be automatically determined. Defaults to None.

  • normalize (bool, optional) –

    if True, normalize data so 0.0=1st percentile and 1.0=99th percentile of image intensities in each channel; can also pass dictionary of parameters (all keys are optional, default values shown):

    • ”lowhigh”=None : pass in normalization values for 0.0 and 1.0 as list [low, high] (if not None, all following parameters ignored)

    • ”sharpen”=0 ; sharpen image with high pass filter, recommended to be 1/4-1/8 diameter of cells in pixels

    • ”normalize”=True ; run normalization (if False, all following parameters ignored)

    • ”percentile”=None : pass in percentiles to use as list [perc_low, perc_high]

    • ”tile_norm”=0 ; compute normalization in tiles across image to brighten dark areas, to turn on set to window size in pixels (e.g. 100)

    • ”norm3D”=False ; compute normalization across entire z-stack rather than plane-by-plane in stitching mode.

    Defaults to True.

  • invert (bool, optional) – invert image pixel intensity before running network. Defaults to False.

  • rescale (float, optional) – resize factor for each image, if None, set to 1.0; (only used if diameter is None). Defaults to None.

  • diameter (float, optional) – diameter for each image, if diameter is None, set to diam_mean or diam_train if available. Defaults to None.

  • flow_threshold (float, optional) – flow error threshold (all cells with errors below threshold are kept) (not used for 3D). Defaults to 0.4.

  • cellprob_threshold (float, optional) – all pixels with value above threshold kept for masks, decrease to find more and larger masks. Defaults to 0.0.

  • do_3D (bool, optional) – set to True to run 3D segmentation on 3D/4D image input. Defaults to False.

  • anisotropy (float, optional) – for 3D segmentation, optional rescaling factor (e.g. set to 2.0 if Z is sampled half as dense as X or Y). Defaults to None.

  • stitch_threshold (float, optional) – if stitch_threshold>0.0 and not do_3D, masks are stitched in 3D to return volume segmentation. Defaults to 0.0.

  • min_size (int, optional) – all ROIs below this size, in pixels, will be discarded. Defaults to 15.

  • niter (int, optional) – number of iterations for dynamics computation. if None, it is set proportional to the diameter. Defaults to None.

  • augment (bool, optional) – tiles image with overlapping tiles and flips overlapped regions to augment. Defaults to False.

  • tile (bool, optional) – tiles image to ensure GPU/CPU memory usage limited (recommended). Defaults to True.

  • tile_overlap (float, optional) – fraction of overlap of tiles when computing flows. Defaults to 0.1.

  • bsize (int, optional) – block size for tiles, recommended to keep at 224, like in training. Defaults to 224.

  • interp (bool, optional) – interpolate during 2D dynamics (not available in 3D) . Defaults to True.

  • compute_masks (bool, optional) – Whether or not to compute dynamics and return masks. This is set to False when retrieving the styles for the size model. Defaults to True.

  • progress (QProgressBar, optional) – pyqt progress bar. Defaults to None.

Returns

  • masks (list, np.ndarray): labelled image(s), where 0=no masks; 1,2,…=mask labels

  • flows (list): list of lists: flows[k][0] = XY flow in HSV 0-255; flows[k][1] = XY(Z) flows at each pixel; flows[k][2] = cell probability (if > cellprob_threshold, pixel used for dynamics); flows[k][3] = final pixel locations after Euler integration

  • styles (list, np.ndarray): style vector summarizing each image of size 256.

Return type

A tuple containing

CellposeDenoiseModel

class cellpose.denoise.CellposeDenoiseModel(gpu=False, pretrained_model=False, model_type=None, restore_type='denoise_cyto3', chan2_restore=False, device=None)[source]

model to run Cellpose and Image restoration

eval(x, batch_size=8, channels=None, channel_axis=None, z_axis=None, normalize=True, rescale=None, diameter=None, tile=True, tile_overlap=0.1, augment=False, resample=True, invert=False, flow_threshold=0.4, cellprob_threshold=0.0, do_3D=False, anisotropy=None, stitch_threshold=0.0, min_size=15, niter=None, interp=True)[source]

Restore array or list of images using the image restoration model, and then segment.

Parameters
  • x (list, np.ndarry) – can be list of 2D/3D/4D images, or array of 2D/3D/4D images

  • batch_size (int, optional) – number of 224x224 patches to run simultaneously on the GPU (can make smaller or bigger depending on GPU memory usage). Defaults to 8.

  • channels (list, optional) – list of channels, either of length 2 or of length number of images by 2. First element of list is the channel to segment (0=grayscale, 1=red, 2=green, 3=blue). Second element of list is the optional nuclear channel (0=none, 1=red, 2=green, 3=blue). For instance, to segment grayscale images, input [0,0]. To segment images with cells in green and nuclei in blue, input [2,3]. To segment one grayscale image and one image with cells in green and nuclei in blue, input [[0,0], [2,3]]. Defaults to None.

  • channel_axis (int, optional) – channel axis in element of list x, or of np.ndarray x. if None, channels dimension is attempted to be automatically determined. Defaults to None.

  • z_axis (int, optional) – z axis in element of list x, or of np.ndarray x. if None, z dimension is attempted to be automatically determined. Defaults to None.

  • normalize (bool, optional) –

    if True, normalize data so 0.0=1st percentile and 1.0=99th percentile of image intensities in each channel; can also pass dictionary of parameters (all keys are optional, default values shown):

    • ”lowhigh”=None : pass in normalization values for 0.0 and 1.0 as list [low, high] (if not None, all following parameters ignored)

    • ”sharpen”=0 ; sharpen image with high pass filter, recommended to be 1/4-1/8 diameter of cells in pixels

    • ”normalize”=True ; run normalization (if False, all following parameters ignored)

    • ”percentile”=None : pass in percentiles to use as list [perc_low, perc_high]

    • ”tile_norm”=0 ; compute normalization in tiles across image to brighten dark areas, to turn on set to window size in pixels (e.g. 100)

    • ”norm3D”=False ; compute normalization across entire z-stack rather than plane-by-plane in stitching mode.

    Defaults to True.

  • rescale (float, optional) – resize factor for each image, if None, set to 1.0; (only used if diameter is None). Defaults to None.

  • diameter (float, optional) – diameter for each image, if diameter is None, set to diam_mean or diam_train if available. Defaults to None.

  • tile (bool, optional) – tiles image to ensure GPU/CPU memory usage limited (recommended). Defaults to True.

  • tile_overlap (float, optional) – fraction of overlap of tiles when computing flows. Defaults to 0.1.

  • augment (bool, optional) – augment tiles by flipping and averaging for segmentation. Defaults to False.

  • resample (bool, optional) – run dynamics at original image size (will be slower but create more accurate boundaries). Defaults to True.

  • invert (bool, optional) – invert image pixel intensity before running network. Defaults to False.

  • flow_threshold (float, optional) – flow error threshold (all cells with errors below threshold are kept) (not used for 3D). Defaults to 0.4.

  • cellprob_threshold (float, optional) – all pixels with value above threshold kept for masks, decrease to find more and larger masks. Defaults to 0.0.

  • do_3D (bool, optional) – set to True to run 3D segmentation on 3D/4D image input. Defaults to False.

  • anisotropy (float, optional) – for 3D segmentation, optional rescaling factor (e.g. set to 2.0 if Z is sampled half as dense as X or Y). Defaults to None.

  • stitch_threshold (float, optional) – if stitch_threshold>0.0 and not do_3D, masks are stitched in 3D to return volume segmentation. Defaults to 0.0.

  • min_size (int, optional) – all ROIs below this size, in pixels, will be discarded. Defaults to 15.

  • niter (int, optional) – number of iterations for dynamics computation. if None, it is set proportional to the diameter. Defaults to None.

  • interp (bool, optional) – interpolate during 2D dynamics (not available in 3D) . Defaults to True.

Returns

labelled image(s), where 0=no masks; 1,2,…=mask labels flows (list): list of lists: flows[k][0] = XY flow in HSV 0-255; flows[k][1] = XY(Z) flows at each pixel; flows[k][2] = cell probability (if > cellprob_threshold, pixel used for dynamics); flows[k][3] = final pixel locations after Euler integration styles (list, np.ndarray): style vector summarizing each image of size 256. imgs (list of 2D/3D arrays): Restored images

Return type

masks (list, np.ndarray)

DenoiseModel

class cellpose.denoise.DenoiseModel(gpu=False, pretrained_model=False, nchan=1, model_type=None, chan2=False, diam_mean=30.0, device=None)[source]

DenoiseModel class for denoising images using Cellpose denoising model.

Parameters
  • gpu (bool, optional) – Whether to use GPU for computation. Defaults to False.

  • pretrained_model (bool or str or Path, optional) – Pretrained model to use for denoising. Can be a string or path. Defaults to False.

  • nchan (int, optional) – Number of channels in the input images, all Cellpose 3 models were trained with nchan=1. Defaults to 1.

  • model_type (str, optional) – Type of pretrained model to use (“denoise_cyto3”, “deblur_cyto3”, “upsample_cyto3”, …). Defaults to None.

  • chan2 (bool, optional) – Whether to use a separate model for the second channel. Defaults to False.

  • diam_mean (float, optional) – Mean diameter of the objects in the images. Defaults to 30.0.

  • device (torch.device, optional) – Device to use for computation. Defaults to None.

nchan

Number of channels in the input images.

Type

int

diam_mean

Mean diameter of the objects in the images.

Type

float

net

Cellpose network for denoising.

Type

CPnet

pretrained_model

Pretrained model path to use for denoising.

Type

bool or str or Path

net_chan2

Cellpose network for the second channel, if applicable.

Type

CPnet or None

net_type

Type of the denoising network.

Type

str

eval(x, batch_size=8, channels=None, channel_axis=None, z_axis=None,

normalize=True, rescale=None, diameter=None, tile=True, tile_overlap=0.1)

Denoise array or list of images using the denoising model.

_eval(net, x, normalize=True, rescale=None, diameter=None, tile=True,

tile_overlap=0.1)

Run denoising model on a single channel.

eval(x, batch_size=8, channels=None, channel_axis=None, z_axis=None, normalize=True, rescale=None, diameter=None, tile=True, tile_overlap=0.1)[source]

Restore array or list of images using the image restoration model.

Parameters
  • x (list, np.ndarry) – can be list of 2D/3D/4D images, or array of 2D/3D/4D images

  • batch_size (int, optional) – number of 224x224 patches to run simultaneously on the GPU (can make smaller or bigger depending on GPU memory usage). Defaults to 8.

  • channels (list, optional) – list of channels, either of length 2 or of length number of images by 2. First element of list is the channel to segment (0=grayscale, 1=red, 2=green, 3=blue). Second element of list is the optional nuclear channel (0=none, 1=red, 2=green, 3=blue). For instance, to segment grayscale images, input [0,0]. To segment images with cells in green and nuclei in blue, input [2,3]. To segment one grayscale image and one image with cells in green and nuclei in blue, input [[0,0], [2,3]]. Defaults to None.

  • channel_axis (int, optional) – channel axis in element of list x, or of np.ndarray x. if None, channels dimension is attempted to be automatically determined. Defaults to None.

  • z_axis (int, optional) – z axis in element of list x, or of np.ndarray x. if None, z dimension is attempted to be automatically determined. Defaults to None.

  • normalize (bool, optional) –

    if True, normalize data so 0.0=1st percentile and 1.0=99th percentile of image intensities in each channel; can also pass dictionary of parameters (all keys are optional, default values shown):

    • ”lowhigh”=None : pass in normalization values for 0.0 and 1.0 as list [low, high] (if not None, all following parameters ignored)

    • ”sharpen”=0 ; sharpen image with high pass filter, recommended to be 1/4-1/8 diameter of cells in pixels

    • ”normalize”=True ; run normalization (if False, all following parameters ignored)

    • ”percentile”=None : pass in percentiles to use as list [perc_low, perc_high]

    • ”tile_norm”=0 ; compute normalization in tiles across image to brighten dark areas, to turn on set to window size in pixels (e.g. 100)

    • ”norm3D”=False ; compute normalization across entire z-stack rather than plane-by-plane in stitching mode.

    Defaults to True.

  • rescale (float, optional) – resize factor for each image, if None, set to 1.0; (only used if diameter is None). Defaults to None.

  • diameter (float, optional) – diameter for each image, if diameter is None, set to diam_mean or diam_train if available. Defaults to None.

  • tile (bool, optional) – tiles image to ensure GPU/CPU memory usage limited (recommended). Defaults to True.

  • tile_overlap (float, optional) – fraction of overlap of tiles when computing flows. Defaults to 0.1.

Returns

Restored images

Return type

imgs (list of 2D/3D arrays)

SizeModel

class cellpose.models.SizeModel(cp_model, device=None, pretrained_size=None, **kwargs)[source]

Linear regression model for determining the size of objects in image used to rescale before input to cp_model. Uses styles from cp_model.

pretrained_size

Path to pretrained size model.

Type

str

cp

Model from which to get styles.

Type

UnetModel or CellposeModel

device

Device used for model running / training (torch.device(“cuda”) or torch.device(“cpu”)), overrides gpu input, recommended if you want to use a specific GPU (e.g. torch.device(“cuda:1”)).

Type

torch device

diam_mean

Mean diameter of objects.

Type

float

eval(self, x, channels=None, channel_axis=None, normalize=True, invert=False,

augment=False, tile=True, batch_size=8, progress=None, interp=True):

Use images x to produce style or use style input to predict size of objects in image.

Raises

ValueError – If no pretrained cellpose model is specified, cannot compute size.

eval(x, channels=None, channel_axis=None, normalize=True, invert=False, augment=False, tile=True, batch_size=8, progress=None)[source]

Use images x to produce style or use style input to predict size of objects in image.

Object size estimation is done in two steps: 1. Use a linear regression model to predict size from style in image. 2. Resize image to predicted size and run CellposeModel to get output masks.

Take the median object size of the predicted masks as the final predicted size.

Parameters
  • x (list, np.ndarry) – can be list of 2D/3D/4D images, or array of 2D/3D/4D images

  • channels (list, optional) – list of channels, either of length 2 or of length number of images by 2. First element of list is the channel to segment (0=grayscale, 1=red, 2=green, 3=blue). Second element of list is the optional nuclear channel (0=none, 1=red, 2=green, 3=blue). For instance, to segment grayscale images, input [0,0]. To segment images with cells in green and nuclei in blue, input [2,3]. To segment one grayscale image and one image with cells in green and nuclei in blue, input [[0,0], [2,3]]. Defaults to None.

  • channel_axis (int, optional) – channel axis in element of list x, or of np.ndarray x. if None, channels dimension is attempted to be automatically determined. Defaults to None.

  • normalize (bool, optional) –

    if True, normalize data so 0.0=1st percentile and 1.0=99th percentile of image intensities in each channel; can also pass dictionary of parameters (all keys are optional, default values shown):

    • ”lowhigh”=None : pass in normalization values for 0.0 and 1.0 as list [low, high] (if not None, all following parameters ignored)

    • ”sharpen”=0 ; sharpen image with high pass filter, recommended to be 1/4-1/8 diameter of cells in pixels

    • ”normalize”=True ; run normalization (if False, all following parameters ignored)

    • ”percentile”=None : pass in percentiles to use as list [perc_low, perc_high]

    • ”tile_norm”=0 ; compute normalization in tiles across image to brighten dark areas, to turn on set to window size in pixels (e.g. 100)

    • ”norm3D”=False ; compute normalization across entire z-stack rather than plane-by-plane in stitching mode.

    Defaults to True.

  • invert (bool, optional) – Invert image pixel intensity before running network (if True, image is also normalized). Defaults to False.

  • augment (bool, optional) – tiles image with overlapping tiles and flips overlapped regions to augment. Defaults to False.

  • tile (bool, optional) – tiles image to ensure GPU/CPU memory usage limited (recommended). Defaults to True.

  • batch_size (int, optional) – number of 224x224 patches to run simultaneously on the GPU (can make smaller or bigger depending on GPU memory usage). Defaults to 8.

  • progress (QProgressBar, optional) – pyqt progress bar. Defaults to None.

Returns

  • diam (np.ndarray): Final estimated diameters from images x or styles style after running both steps.

  • diam_style (np.ndarray): Estimated diameters from style alone.

Return type

A tuple containing

Training

cellpose.train.train_seg(net, train_data=None, train_labels=None, train_files=None, train_labels_files=None, train_probs=None, test_data=None, test_labels=None, test_files=None, test_labels_files=None, test_probs=None, load_files=True, batch_size=8, learning_rate=0.005, n_epochs=2000, weight_decay=1e-05, momentum=0.9, SGD=False, channels=None, channel_axis=None, normalize=True, compute_flows=False, save_path=None, save_every=100, nimg_per_epoch=None, nimg_test_per_epoch=None, rescale=True, min_train_masks=5, model_name=None)[source]

Train the network with images for segmentation.

Parameters
  • net (object) – The network model to train.

  • train_data (List[np.ndarray], optional) – List of arrays (2D or 3D) - images for training. Defaults to None.

  • train_labels (List[np.ndarray], optional) – List of arrays (2D or 3D) - labels for train_data, where 0=no masks; 1,2,…=mask labels. Defaults to None.

  • train_files (List[str], optional) – List of strings - file names for images in train_data (to save flows for future runs). Defaults to None.

  • train_labels_files (List[str], optional) – List of strings - file names for labels in train_labels. Defaults to None.

  • train_probs (List[float], optional) – List of floats - probabilities for each image to be selected during training. Defaults to None.

  • test_data (List[np.ndarray], optional) – List of arrays (2D or 3D) - images for testing. Defaults to None.

  • test_labels (List[np.ndarray], optional) – List of arrays (2D or 3D) - labels for test_data, where 0=no masks; 1,2,…=mask labels. Defaults to None.

  • test_files (List[str], optional) – List of strings - file names for images in test_data (to save flows for future runs). Defaults to None.

  • test_labels_files (List[str], optional) – List of strings - file names for labels in test_labels. Defaults to None.

  • test_probs (List[float], optional) – List of floats - probabilities for each image to be selected during testing. Defaults to None.

  • load_files (bool, optional) – Boolean - whether to load images and labels from files. Defaults to True.

  • batch_size (int, optional) – Integer - number of patches to run simultaneously on the GPU. Defaults to 8.

  • learning_rate (float or List[float], optional) – Float or list/np.ndarray - learning rate for training. Defaults to 0.005.

  • n_epochs (int, optional) – Integer - number of times to go through the whole training set during training. Defaults to 2000.

  • weight_decay (float, optional) – Float - weight decay for the optimizer. Defaults to 1e-5.

  • momentum (float, optional) – Float - momentum for the optimizer. Defaults to 0.9.

  • SGD (bool, optional) – Boolean - whether to use SGD as optimization instead of RAdam. Defaults to False.

  • channels (List[int], optional) – List of ints - channels to use for training. Defaults to None.

  • channel_axis (int, optional) – Integer - axis of the channel dimension in the input data. Defaults to None.

  • normalize (bool or dict, optional) – Boolean or dictionary - whether to normalize the data. Defaults to True.

  • compute_flows (bool, optional) – Boolean - whether to compute flows during training. Defaults to False.

  • save_path (str, optional) – String - where to save the trained model. Defaults to None.

  • save_every (int, optional) – Integer - save the network every [save_every] epochs. Defaults to 100.

  • nimg_per_epoch (int, optional) – Integer - minimum number of images to train on per epoch. Defaults to None.

  • nimg_test_per_epoch (int, optional) – Integer - minimum number of images to test on per epoch. Defaults to None.

  • rescale (bool, optional) – Boolean - whether or not to rescale images during training. Defaults to True.

  • min_train_masks (int, optional) – Integer - minimum number of masks an image must have to use in the training set. Defaults to 5.

  • model_name (str, optional) – String - name of the network. Defaults to None.

Returns

path to saved model weights

Return type

Path

cellpose.train.train_size(net, pretrained_model, train_data=None, train_labels=None, train_files=None, train_labels_files=None, train_probs=None, test_data=None, test_labels=None, test_files=None, test_labels_files=None, test_probs=None, load_files=True, min_train_masks=5, channels=None, channel_axis=None, normalize=True, nimg_per_epoch=None, nimg_test_per_epoch=None, batch_size=128, l2_regularization=1.0, n_epochs=10)[source]

Train the size model.

Parameters
  • net (object) – The neural network model.

  • pretrained_model (str) – The path to the pretrained model.

  • train_data (numpy.ndarray, optional) – The training data. Defaults to None.

  • train_labels (numpy.ndarray, optional) – The training labels. Defaults to None.

  • train_files (list, optional) – The training file paths. Defaults to None.

  • train_labels_files (list, optional) – The training label file paths. Defaults to None.

  • train_probs (numpy.ndarray, optional) – The training probabilities. Defaults to None.

  • test_data (numpy.ndarray, optional) – The test data. Defaults to None.

  • test_labels (numpy.ndarray, optional) – The test labels. Defaults to None.

  • test_files (list, optional) – The test file paths. Defaults to None.

  • test_labels_files (list, optional) – The test label file paths. Defaults to None.

  • test_probs (numpy.ndarray, optional) – The test probabilities. Defaults to None.

  • load_files (bool, optional) – Whether to load files. Defaults to True.

  • min_train_masks (int, optional) – The minimum number of training masks. Defaults to 5.

  • channels (list, optional) – The channels. Defaults to None.

  • channel_axis (int, optional) – The channel axis. Defaults to None.

  • normalize (bool or dict, optional) – Whether to normalize the data. Defaults to True.

  • nimg_per_epoch (int, optional) – The number of images per epoch. Defaults to None.

  • nimg_test_per_epoch (int, optional) – The number of test images per epoch. Defaults to None.

  • batch_size (int, optional) – The batch size. Defaults to 128.

  • l2_regularization (float, optional) – The L2 regularization factor. Defaults to 1.0.

  • n_epochs (int, optional) – The number of epochs. Defaults to 10.

Returns

The trained size model parameters.

Return type

dict

Metrics

Copyright © 2023 Howard Hughes Medical Institute, Authored by Carsen Stringer and Marius Pachitariu.

cellpose.metrics.aggregated_jaccard_index(masks_true, masks_pred)[source]

AJI = intersection of all matched masks / union of all masks

Parameters
  • masks_true (list of np.ndarrays (int) or np.ndarray (int)) – where 0=NO masks; 1,2… are mask labels

  • masks_pred (list of np.ndarrays (int) or np.ndarray (int)) – np.ndarray (int) where 0=NO masks; 1,2… are mask labels

Returns

aggregated jaccard index for each set of masks

Return type

aji (float)

cellpose.metrics.average_precision(masks_true, masks_pred, threshold=[0.5, 0.75, 0.9])[source]

Average precision estimation: AP = TP / (TP + FP + FN)

This function is based heavily on the fast stardist matching functions (https://github.com/mpicbg-csbd/stardist/blob/master/stardist/matching.py)

Parameters
  • masks_true (list of np.ndarrays (int) or np.ndarray (int)) – where 0=NO masks; 1,2… are mask labels

  • masks_pred (list of np.ndarrays (int) or np.ndarray (int)) – np.ndarray (int) where 0=NO masks; 1,2… are mask labels

Returns

average precision at thresholds tp (array [len(masks_true) x len(threshold)]):

number of true positives at thresholds

fp (array [len(masks_true) x len(threshold)]):

number of false positives at thresholds

fn (array [len(masks_true) x len(threshold)]):

number of false negatives at thresholds

Return type

ap (array [len(masks_true) x len(threshold)])

cellpose.metrics.boundary_scores(masks_true, masks_pred, scales)[source]

Calculate boundary precision, recall, and F-score.

Parameters
  • masks_true (list) – List of true masks.

  • masks_pred (list) – List of predicted masks.

  • scales (list) – List of scales.

Returns

A tuple containing precision, recall, and F-score arrays.

Return type

tuple

cellpose.metrics.flow_error(maski, dP_net, device=None)[source]

Error in flows from predicted masks vs flows predicted by network run on image.

This function serves to benchmark the quality of masks. It works as follows: 1. The predicted masks are used to create a flow diagram. 2. The mask-flows are compared to the flows that the network predicted.

If there is a discrepancy between the flows, it suggests that the mask is incorrect. Masks with flow_errors greater than 0.4 are discarded by default. This setting can be changed in Cellpose.eval or CellposeModel.eval.

Parameters
  • maski (np.ndarray, int) – Masks produced from running dynamics on dP_net, where 0=NO masks; 1,2… are mask labels.

  • dP_net (np.ndarray, float) – ND flows where dP_net.shape[1:] = maski.shape.

Returns

Mean squared error between predicted flows and flows from masks. dP_masks (np.ndarray, float): ND flows produced from the predicted masks.

Return type

flow_errors (np.ndarray, float)

cellpose.metrics.mask_ious(masks_true, masks_pred)[source]

Return best-matched masks.

Flows to masks

Copyright © 2023 Howard Hughes Medical Institute, Authored by Carsen Stringer and Marius Pachitariu.

cellpose.dynamics.compute_masks(dP, cellprob, p=None, niter=200, cellprob_threshold=0.0, flow_threshold=0.4, interp=True, do_3D=False, min_size=15, device=None)[source]

Compute masks using dynamics from dP and cellprob.

Parameters
  • dP (numpy.ndarray) – The dynamics flow field array.

  • cellprob (numpy.ndarray) – The cell probability array.

  • p (numpy.ndarray, optional) – The pixels on which to run dynamics. Defaults to None

  • niter (int, optional) – The number of iterations for mask computation. Defaults to 200.

  • cellprob_threshold (float, optional) – The threshold for cell probability. Defaults to 0.0.

  • flow_threshold (float, optional) – The threshold for quality control metrics. Defaults to 0.4.

  • interp (bool, optional) – Whether to interpolate during dynamics computation. Defaults to True.

  • do_3D (bool, optional) – Whether to perform mask computation in 3D. Defaults to False.

  • min_size (int, optional) – The minimum size of the masks. Defaults to 15.

  • device (str, optional) – The torch device to use for computation. Defaults to None.

Returns

A tuple containing the computed masks and the final pixel locations.

Return type

tuple

cellpose.dynamics.follow_flows(dP, mask=None, niter=200, interp=True, device=None)[source]

Run dynamics to recover masks in 2D or 3D.

Pixels are represented as a meshgrid. Only pixels with non-zero cell-probability are used (as defined by inds).

Parameters
  • dP (np.ndarray) – Flows [axis x Ly x Lx] or [axis x Lz x Ly x Lx].

  • mask (np.ndarray, optional) – Pixel mask to seed masks. Useful when flows have low magnitudes.

  • niter (int, optional) – Number of iterations of dynamics to run. Default is 200.

  • interp (bool, optional) – Interpolate during 2D dynamics (not available in 3D). Default is True.

  • use_gpu (bool, optional) – Use GPU to run interpolated dynamics (faster than CPU). Default is False.

Returns

  • p (np.ndarray): Final locations of each pixel after dynamics; [axis x Ly x Lx] or [axis x Lz x Ly x Lx].

  • inds (np.ndarray): Indices of pixels used for dynamics; [axis x Ly x Lx] or [axis x Lz x Ly x Lx].

Return type

tuple containing

cellpose.dynamics.get_centers(masks, slices)

Get the centers of the masks and their extents.

Parameters
  • masks (ndarray) – The labeled masks.

  • slices (ndarray) – The slices of the masks.

Returns

tuple containing
  • centers (ndarray): The centers of the masks.

  • ext (ndarray): The extents of the masks.

cellpose.dynamics.get_masks(p, iscell=None, rpad=20)[source]

Create masks using pixel convergence after running dynamics.

Makes a histogram of final pixel locations p, initializes masks at peaks of histogram and extends the masks from the peaks so that they include all pixels with more than 2 final pixels p. Discards masks with flow errors greater than the threshold.

Parameters
  • p (float32, 3D or 4D array) – Final locations of each pixel after dynamics, size [axis x Ly x Lx] or [axis x Lz x Ly x Lx].

  • iscell (bool, 2D or 3D array) – If iscell is not None, set pixels that are iscell False to stay in their original location.

  • rpad (int, optional) – Histogram edge padding. Default is 20.

Returns

Masks with inconsistent flow masks removed,

0=NO masks; 1,2,…=mask labels, size [Ly x Lx] or [Lz x Ly x Lx].

Return type

M0 (int, 2D or 3D array)

cellpose.dynamics.labels_to_flows(labels, files=None, device=None, redo_flows=False, niter=None)[source]

Converts labels (list of masks or flows) to flows for training model.

Parameters
  • labels (list of ND-arrays) – The labels to convert. labels[k] can be 2D or 3D. If [3 x Ly x Lx], it is assumed that flows were precomputed. Otherwise, labels[k][0] or labels[k] (if 2D) is used to create flows and cell probabilities.

  • files (list of str, optional) – The files to save the flows to. If provided, flows are saved to files to be reused. Defaults to None.

  • device (str, optional) – The device to use for computation. Defaults to None.

  • redo_flows (bool, optional) – Whether to recompute the flows. Defaults to False.

  • niter (int, optional) – The number of iterations for computing flows. Defaults to None.

Returns

The flows for training the model. flows[k][0] is labels[k], flows[k][1] is cell distance transform, flows[k][2] is Y flow, flows[k][3] is X flow, and flows[k][4] is heat distribution.

Return type

list of [4 x Ly x Lx] arrays

cellpose.dynamics.map_coordinates(I, yc, xc, Y)

Bilinear interpolation of image “I” in-place with y-coordinates yc and x-coordinates xc to Y.

Parameters
  • I (numpy.ndarray) – Input image of shape (C, Ly, Lx).

  • yc (numpy.ndarray) – New y-coordinates.

  • xc (numpy.ndarray) – New x-coordinates.

  • Y (numpy.ndarray) – Output array of shape (C, ni).

Returns

None

cellpose.dynamics.masks_to_flows(masks, device=None, niter=None)[source]

Convert masks to flows using diffusion from center pixel.

Center of masks where diffusion starts is defined to be the closest pixel to the mean of all pixels that is inside the mask. Result of diffusion is converted into flows by computing the gradients of the diffusion density map.

Parameters

masks (int, 2D or 3D array) – Labelled masks 0=NO masks; 1,2,…=mask labels

Returns

Flows in Y = mu[-2], flows in X = mu[-1].

If masks are 3D, flows in Z = mu[0].

Return type

mu (float, 3D or 4D array)

cellpose.dynamics.masks_to_flows_cpu(masks, device=None, niter=None)[source]

Convert masks to flows using diffusion from center pixel.

Center of masks where diffusion starts is defined to be the closest pixel to the mean of all pixels that is inside the mask. Result of diffusion is converted into flows by computing the gradients of the diffusion density map.

Parameters

masks (int, 2D or 3D array) – Labelled masks 0=NO masks; 1,2,…=mask labels

Returns

tuple containing
  • mu (float, 3D or 4D array): Flows in Y = mu[-2], flows in X = mu[-1].

    If masks are 3D, flows in Z = mu[0].

  • meds (float, 2D or 3D array): cell centers

cellpose.dynamics.masks_to_flows_gpu(masks, device=None, niter=None)[source]

Convert masks to flows using diffusion from center pixel.

Center of masks where diffusion starts is defined using COM.

Parameters

masks (int, 2D or 3D array) – Labelled masks. 0=NO masks; 1,2,…=mask labels.

Returns

tuple containing
  • mu (float, 3D or 4D array): Flows in Y = mu[-2], flows in X = mu[-1].

    If masks are 3D, flows in Z = mu[0].

  • meds_p (float, 2D or 3D array): cell centers

cellpose.dynamics.masks_to_flows_gpu_3d(masks, device=None)[source]

Convert masks to flows using diffusion from center pixel.

Parameters

masks (int, 2D or 3D array) – Labelled masks. 0=NO masks; 1,2,…=mask labels.

Returns

tuple containing
  • mu (float, 3D or 4D array): Flows in Y = mu[-2], flows in X = mu[-1]. If masks are 3D, flows in Z = mu[0].

  • mu_c (float, 2D or 3D array): zeros

cellpose.dynamics.remove_bad_flow_masks(masks, flows, threshold=0.4, device=None)[source]

Remove masks which have inconsistent flows.

Uses metrics.flow_error to compute flows from predicted masks and compare flows to predicted flows from the network. Discards masks with flow errors greater than the threshold.

Parameters
  • masks (int, 2D or 3D array) – Labelled masks, 0=NO masks; 1,2,…=mask labels, size [Ly x Lx] or [Lz x Ly x Lx].

  • flows (float, 3D or 4D array) – Flows [axis x Ly x Lx] or [axis x Lz x Ly x Lx].

  • threshold (float, optional) – Masks with flow error greater than threshold are discarded. Default is 0.4.

Returns

Masks with inconsistent flow masks removed,

0=NO masks; 1,2,…=mask labels, size [Ly x Lx] or [Lz x Ly x Lx].

Return type

masks (int, 2D or 3D array)

cellpose.dynamics.resize_and_compute_masks(dP, cellprob, p=None, niter=200, cellprob_threshold=0.0, flow_threshold=0.4, interp=True, do_3D=False, min_size=15, resize=None, device=None)[source]

Compute masks using dynamics from dP and cellprob, and resizes masks if resize is not None.

Parameters
  • dP (numpy.ndarray) – The dynamics flow field array.

  • cellprob (numpy.ndarray) – The cell probability array.

  • p (numpy.ndarray, optional) – The pixels on which to run dynamics. Defaults to None

  • niter (int, optional) – The number of iterations for mask computation. Defaults to 200.

  • cellprob_threshold (float, optional) – The threshold for cell probability. Defaults to 0.0.

  • flow_threshold (float, optional) – The threshold for quality control metrics. Defaults to 0.4.

  • interp (bool, optional) – Whether to interpolate during dynamics computation. Defaults to True.

  • do_3D (bool, optional) – Whether to perform mask computation in 3D. Defaults to False.

  • min_size (int, optional) – The minimum size of the masks. Defaults to 15.

  • resize (tuple, optional) – The desired size for resizing the masks. Defaults to None.

  • device (str, optional) – The torch device to use for computation. Defaults to None.

Returns

A tuple containing the computed masks and the final pixel locations.

Return type

tuple

cellpose.dynamics.steps2D(p, dP, inds, niter)

Run dynamics of pixels to recover masks in 2D.

Euler integration of dynamics dP for niter steps.

Parameters
  • p (np.ndarray) – Pixel locations [axis x Ly x Lx] (start at initial meshgrid).

  • dP (np.ndarray) – Flows [axis x Ly x Lx].

  • inds (np.ndarray) – Non-zero pixels to run dynamics on [npixels x 2].

  • niter (int) – Number of iterations of dynamics to run.

Returns

Final locations of each pixel after dynamics.

Return type

np.ndarray

cellpose.dynamics.steps2D_interp(p, dP, niter, device=None)[source]

Run dynamics of pixels to recover masks in 2D, with interpolation between pixel values.

Euler integration of dynamics dP for niter steps.

Parameters
  • p (numpy.ndarray) – Array of shape (n_points, 2) representing the initial pixel locations.

  • dP (numpy.ndarray) – Array of shape (2, Ly, Lx) representing the flow field.

  • niter (int) – Number of iterations to perform.

  • device (torch.device, optional) – Device to use for computation. Defaults to None.

Returns

Array of shape (n_points, 2) representing the final pixel locations.

Return type

numpy.ndarray

Raises

None

cellpose.dynamics.steps3D(p, dP, inds, niter)

Run dynamics of pixels to recover masks in 3D.

Euler integration of dynamics dP for niter steps.

Parameters
  • p (np.ndarray) – Pixel locations [axis x Lz x Ly x Lx] (start at initial meshgrid).

  • dP (np.ndarray) – Flows [axis x Lz x Ly x Lx].

  • inds (np.ndarray) – Non-zero pixels to run dynamics on [npixels x 3].

  • niter (int) – Number of iterations of dynamics to run.

Returns

Final locations of each pixel after dynamics.

Return type

np.ndarray

Image transforms

Copyright © 2023 Howard Hughes Medical Institute, Authored by Carsen Stringer and Marius Pachitariu.

cellpose.transforms.average_tiles(y, ysub, xsub, Ly, Lx)[source]

Average the results of the network over tiles.

Parameters
  • y (float) – Output of cellpose network for each tile. Shape: [ntiles x nclasses x bsize x bsize]

  • ysub (list) – List of arrays with start and end of tiles in Y of length ntiles

  • xsub (list) – List of arrays with start and end of tiles in X of length ntiles

  • Ly (int) – Size of pre-tiled image in Y (may be larger than original image if image size is less than bsize)

  • Lx (int) – Size of pre-tiled image in X (may be larger than original image if image size is less than bsize)

Returns

Network output averaged over tiles. Shape: [nclasses x Ly x Lx]

Return type

yf (float32)

cellpose.transforms.convert_image(x, channels, channel_axis=None, z_axis=None, do_3D=False, nchan=2)[source]

Converts the image to have the z-axis first, channels last, and normalized intensities.

Parameters
  • x (numpy.ndarray or torch.Tensor) – The input image.

  • channels (list or None) – The list of channels to use (ones-based, 0=gray). If None, all channels are kept.

  • channel_axis (int or None) – The axis of the channels in the input image. If None, the axis is determined automatically.

  • z_axis (int or None) – The axis of the z-dimension in the input image. If None, the axis is determined automatically.

  • do_3D (bool) – Whether to process the image in 3D mode. Defaults to False.

  • nchan (int) – The number of channels to keep if the input image has more than nchan channels.

Returns

The converted image.

Return type

numpy.ndarray

Raises
  • ValueError – If the input image has less than two channels and channels are not specified.

  • ValueError – If the input image is 2D and do_3D is True.

  • ValueError – If the input image is 4D and do_3D is False.

cellpose.transforms.gaussian_kernel(sigma, Ly, Lx, device=torch.device)[source]

Generates a 2D Gaussian kernel.

Parameters
  • sigma (float) – Standard deviation of the Gaussian distribution.

  • Ly (int) – Number of pixels in the y-axis.

  • Lx (int) – Number of pixels in the x-axis.

  • device (torch.device, optional) – Device to store the kernel tensor. Defaults to torch.device(“cpu”).

Returns

2D Gaussian kernel tensor.

Return type

torch.Tensor

cellpose.transforms.make_tiles(imgi, bsize=224, augment=False, tile_overlap=0.1)[source]

Make tiles of image to run at test-time.

Parameters
  • imgi (np.ndarray) – Array of shape (nchan, Ly, Lx) representing the input image.

  • bsize (int, optional) – Size of tiles. Defaults to 224.

  • augment (bool, optional) – Whether to flip tiles and set tile_overlap=2. Defaults to False.

  • tile_overlap (float, optional) – Fraction of overlap of tiles. Defaults to 0.1.

Returns

tuple containing
  • IMG (np.ndarray): Array of shape (ntiles, nchan, bsize, bsize) representing the tiles.

  • ysub (list): List of arrays with start and end of tiles in Y of length ntiles.

  • xsub (list): List of arrays with start and end of tiles in X of length ntiles.

  • Ly (int): Height of the input image.

  • Lx (int): Width of the input image.

cellpose.transforms.move_axis(img, m_axis=-1, first=True)[source]

move axis m_axis to first or last position

cellpose.transforms.move_min_dim(img, force=False)[source]

Move the minimum dimension last as channels if it is less than 10 or force is True.

Parameters
  • img (ndarray) – The input image.

  • force (bool, optional) – If True, the minimum dimension will always be moved. Defaults to False.

Returns

The image with the minimum dimension moved to the last axis as channels.

Return type

ndarray

cellpose.transforms.normalize99(Y, lower=1, upper=99, copy=True)[source]

Normalize the image so that 0.0 corresponds to the 1st percentile and 1.0 corresponds to the 99th percentile.

Parameters
  • Y (ndarray) – The input image.

  • lower (int, optional) – The lower percentile. Defaults to 1.

  • upper (int, optional) – The upper percentile. Defaults to 99.

  • copy (bool, optional) – Whether to create a copy of the input image. Defaults to True.

Returns

The normalized image.

Return type

ndarray

cellpose.transforms.normalize99_tile(img, blocksize=100, lower=1.0, upper=99.0, tile_overlap=0.1, norm3D=False, smooth3D=1, is3D=False)[source]

Compute normalization like normalize99 function but in tiles.

Parameters
  • img (numpy.ndarray) – Array of shape (Lz x) Ly x Lx (x nchan) containing the image.

  • blocksize (float, optional) – Size of tiles. Defaults to 100.

  • lower (float, optional) – Lower percentile for normalization. Defaults to 1.0.

  • upper (float, optional) – Upper percentile for normalization. Defaults to 99.0.

  • tile_overlap (float, optional) – Fraction of overlap of tiles. Defaults to 0.1.

  • norm3D (bool, optional) – Use same tiled normalization for each z-plane. Defaults to False.

  • smooth3D (int, optional) – Smoothing factor for 3D normalization. Defaults to 1.

  • is3D (bool, optional) – Set to True if image is a 3D stack. Defaults to False.

Returns

Normalized image array of shape (Lz x) Ly x Lx (x nchan).

Return type

numpy.ndarray

cellpose.transforms.normalize_img(img, normalize=True, norm3D=False, invert=False, lowhigh=None, percentile=None, sharpen_radius=0, smooth_radius=0, tile_norm_blocksize=0, tile_norm_smooth3D=1, axis=-1)[source]

Normalize each channel of the image.

Parameters
  • img (ndarray) – The input image. It should have at least 3 dimensions. If it is 4-dimensional, it assumes the first non-channel axis is the Z dimension.

  • normalize (bool, optional) – Whether to perform normalization. Defaults to True.

  • norm3D (bool, optional) – Whether to normalize in 3D. Defaults to False.

  • invert (bool, optional) – Whether to invert the image. Useful if cells are dark instead of bright. Defaults to False.

  • lowhigh (tuple, optional) – The lower and upper bounds for normalization. If provided, it should be a tuple of two values. Defaults to None.

  • percentile (tuple, optional) – The lower and upper percentiles for normalization. If provided, it should be a tuple of two values. Each value should be between 0 and 100. Defaults to None.

  • sharpen_radius (int, optional) – The radius for sharpening the image. Defaults to 0.

  • smooth_radius (int, optional) – The radius for smoothing the image. Defaults to 0.

  • tile_norm_blocksize (int, optional) – The block size for tile-based normalization. Defaults to 0.

  • tile_norm_smooth3D (int, optional) – The smoothness factor for tile-based normalization in 3D. Defaults to 1.

  • axis (int, optional) – The channel axis to loop over for normalization. Defaults to -1.

Returns

The normalized image of the same size.

Return type

ndarray

Raises
  • ValueError – If the image has less than 3 dimensions.

  • ValueError – If the provided lowhigh or percentile values are invalid.

  • ValueError – If the image is inverted without normalization.

cellpose.transforms.pad_image_ND(img0, div=16, extra=1, min_size=None)[source]

Pad image for test-time so that its dimensions are a multiple of 16 (2D or 3D).

Parameters
  • img0 (ndarray) – Image of size [nchan (x Lz) x Ly x Lx].

  • div (int, optional) – Divisor for padding. Defaults to 16.

  • extra (int, optional) – Extra padding. Defaults to 1.

  • min_size (tuple, optional) – Minimum size of the image. Defaults to None.

Returns

tuple containing
  • I (ndarray): Padded image.

  • ysub (ndarray): Y range of pixels in the padded image corresponding to img0.

  • xsub (ndarray): X range of pixels in the padded image corresponding to img0.

cellpose.transforms.random_rotate_and_resize(X, Y=None, scale_range=1.0, xy=(224, 224), do_3D=False, do_flip=True, rotate=True, rescale=None, unet=False, random_per_image=True)[source]

Augmentation by random rotation and resizing.

Parameters
  • X (list of ND-arrays, float) – List of image arrays of size [nchan x Ly x Lx] or [Ly x Lx].

  • Y (list of ND-arrays, float, optional) – List of image labels of size [nlabels x Ly x Lx] or [Ly x Lx]. The 1st channel of Y is always nearest-neighbor interpolated (assumed to be masks or 0-1 representation). If Y.shape[0]==3 and not unet, then the labels are assumed to be [cell probability, Y flow, X flow]. If unet, second channel is dist_to_bound. Defaults to None.

  • scale_range (float, optional) – Range of resizing of images for augmentation. Images are resized by (1-scale_range/2) + scale_range * np.random.rand(). Defaults to 1.0.

  • xy (tuple, int, optional) – Size of transformed images to return. Defaults to (224,224).

  • do_flip (bool, optional) – Whether or not to flip images horizontally. Defaults to True.

  • rotate (bool, optional) – Whether or not to rotate images. Defaults to True.

  • rescale (array, float, optional) – How much to resize images by before performing augmentations. Defaults to None.

  • unet (bool, optional) – Whether or not to use unet. Defaults to False.

  • random_per_image (bool, optional) – Different random rotate and resize per image. Defaults to True.

Returns

tuple containing
  • imgi (ND-array, float): Transformed images in array [nimg x nchan x xy[0] x xy[1]].

  • lbl (ND-array, float): Transformed labels in array [nimg x nchan x xy[0] x xy[1]].

  • scale (array, float): Amount each image was resized by.

cellpose.transforms.reshape(data, channels=[0, 0], chan_first=False)[source]

Reshape data using channels.

Parameters
  • data (numpy.ndarray) – The input data. It should have shape (Z x ) Ly x Lx x nchan if data.ndim==8 and data.shape[0]<8, it is assumed to be nchan x Ly x Lx.

  • channels (list of int, optional) – The channels to use for reshaping. The first element of the list is the channel to segment (0=grayscale, 1=red, 2=green, 3=blue). The second element of the list is the optional nuclear channel (0=none, 1=red, 2=green, 3=blue). For instance, to train on grayscale images, input [0,0]. To train on images with cells in green and nuclei in blue, input [2,3]. Defaults to [0, 0].

  • chan_first (bool, optional) – Whether to return the reshaped data with channel as the first dimension. Defaults to False.

Returns

The reshaped data with shape (Z x ) Ly x Lx x nchan (if chan_first==False).

Return type

numpy.ndarray

cellpose.transforms.resize_image(img0, Ly=None, Lx=None, rsz=None, interpolation=cv2.INTER_LINEAR, no_channels=False)[source]

Resize image for computing flows / unresize for computing dynamics.

Parameters
  • img0 (ndarray) – Image of size [Y x X x nchan] or [Lz x Y x X x nchan] or [Lz x Y x X].

  • Ly (int, optional) – Desired height of the resized image. Defaults to None.

  • Lx (int, optional) – Desired width of the resized image. Defaults to None.

  • rsz (float, optional) – Resize coefficient(s) for the image. If Ly is None, rsz is used. Defaults to None.

  • interpolation (int, optional) – OpenCV interpolation method. Defaults to cv2.INTER_LINEAR.

  • no_channels (bool, optional) – Flag indicating whether to treat the third dimension as a channel. Defaults to False.

Returns

Resized image of size [Ly x Lx x nchan] or [Lz x Ly x Lx x nchan].

Return type

ndarray

Raises

ValueError – If Ly is None and rsz is None.

cellpose.transforms.smooth_sharpen_img(img, smooth_radius=6, sharpen_radius=12, device=torch.device, is3D=False)[source]

Sharpen blurry images with surround subtraction and/or smooth noisy images.

Parameters
  • img (float32) – Array that’s (Lz x) Ly x Lx (x nchan).

  • smooth_radius (float, optional) – Size of gaussian smoothing filter, recommended to be 1/10-1/4 of cell diameter (if also sharpening, should be 2-3x smaller than sharpen_radius). Defaults to 6.

  • sharpen_radius (float, optional) – Size of gaussian surround filter, recommended to be 1/8-1/2 of cell diameter (if also smoothing, should be 2-3x larger than smooth_radius). Defaults to 12.

  • device (torch.device, optional) – Device on which to perform sharpening. Will be faster on GPU but need to ensure GPU has RAM for image. Defaults to torch.device(“cpu”).

  • is3D (bool, optional) – If image is 3D stack (only necessary to set if img.ndim==3). Defaults to False.

Returns

Array that’s (Lz x) Ly x Lx (x nchan).

Return type

img_sharpen (float32)

cellpose.transforms.unaugment_tiles(y)[source]

Reverse test-time augmentations for averaging (includes flipping of flowsY and flowsX).

Parameters

y (float32) – Array of shape (ntiles_y, ntiles_x, chan, Ly, Lx) where chan = (flowsY, flowsX, cell prob).

Returns

Array of shape (ntiles_y, ntiles_x, chan, Ly, Lx).

Return type

float32

cellpose.transforms.update_axis(m_axis, to_squeeze, ndim)[source]

Squeeze the axis value based on the given parameters.

Parameters
  • m_axis (int) – The current axis value.

  • to_squeeze (numpy.ndarray) – An array of indices to squeeze.

  • ndim (int) – The number of dimensions.

Returns

The updated axis value.

Return type

int or None

Plot functions

Copyright © 2023 Howard Hughes Medical Institute, Authored by Carsen Stringer and Marius Pachitariu.

cellpose.plot.disk(med, r, Ly, Lx)[source]

Returns the pixels of a disk with a given radius and center.

Parameters
  • med (tuple) – The center coordinates of the disk.

  • r (float) – The radius of the disk.

  • Ly (int) – The height of the image.

  • Lx (int) – The width of the image.

Returns

A tuple containing the y and x coordinates of the pixels within the disk.

Return type

tuple

cellpose.plot.dx_to_circ(dP, transparency=False, mask=None)[source]

Converts the optic flow representation to a circular color representation.

Parameters
  • dP (ndarray) – Flow field components [dy, dx].

  • transparency (bool, optional) – Controls the opacity based on the magnitude of flow. Defaults to False.

  • mask (ndarray, optional) – Multiplies each RGB component to suppress noise.

Returns

The circular color representation of the optic flow.

Return type

ndarray

cellpose.plot.image_to_rgb(img0, channels=[0, 0])[source]

Converts image from 2 x Ly x Lx or Ly x Lx x 2 to RGB Ly x Lx x 3.

Parameters

img0 (ndarray) – Input image of shape 2 x Ly x Lx or Ly x Lx x 2.

Returns

RGB image of shape Ly x Lx x 3.

Return type

ndarray

cellpose.plot.interesting_patch(mask, bsize=130)[source]

Get patch of size bsize x bsize with most masks.

Parameters
  • mask (ndarray) – Input mask.

  • bsize (int) – Size of the patch.

Returns

Patch coordinates (y, x).

Return type

tuple

cellpose.plot.mask_overlay(img, masks, colors=None)[source]

Overlay masks on image (set image to grayscale).

Parameters
  • img (int or float, 2D or 3D array) – Image of size [Ly x Lx (x nchan)].

  • masks (int, 2D array) – Masks where 0=NO masks; 1,2,…=mask labels.

  • colors (int, 2D array, optional) – Size [nmasks x 3], each entry is a color in 0-255 range.

Returns

Array of masks overlaid on grayscale image.

Return type

RGB (uint8, 3D array)

cellpose.plot.mask_rgb(masks, colors=None)[source]

Masks in random RGB colors.

Parameters
  • masks (int, 2D array) – Masks where 0=NO masks; 1,2,…=mask labels.

  • colors (int, 2D array, optional) – Size [nmasks x 3], each entry is a color in 0-255 range.

Returns

Array of masks overlaid on grayscale image.

Return type

RGB (uint8, 3D array)

cellpose.plot.outline_view(img0, maski, color=[1, 0, 0], mode='inner')[source]

Generates a red outline overlay onto the image.

Parameters
  • img0 (numpy.ndarray) – The input image.

  • maski (numpy.ndarray) – The mask representing the region of interest.

  • color (list, optional) – The color of the outline overlay. Defaults to [1, 0, 0] (red).

  • mode (str, optional) – The mode for generating the outline. Defaults to “inner”.

Returns

The image with the red outline overlay.

Return type

numpy.ndarray

cellpose.plot.show_segmentation(fig, img, maski, flowi, channels=[0, 0], file_name=None)[source]

Plot segmentation results (like on website).

Can save each panel of figure with file_name option. Use channels option if img input is not an RGB image with 3 channels.

Parameters
  • fig (matplotlib.pyplot.figure) – Figure in which to make plot.

  • img (ndarray) – 2D or 3D array. Image input into cellpose.

  • maski (int, ndarray) – For image k, masks[k] output from Cellpose.eval, where 0=NO masks; 1,2,…=mask labels.

  • flowi (int, ndarray) – For image k, flows[k][0] output from Cellpose.eval (RGB of flows).

  • channels (list of int, optional) – Channels used to run Cellpose, no need to use if image is RGB. Defaults to [0, 0].

  • file_name (str, optional) – File name of image. If file_name is not None, figure panels are saved. Defaults to None.

  • seg_norm (bool, optional) – Improve cell visibility under labels. Defaults to False.

I/O functions

Copyright © 2023 Howard Hughes Medical Institute, Authored by Carsen Stringer and Marius Pachitariu.

cellpose.io.add_model(filename)[source]

add model to .cellpose models folder to use with GUI or CLI

cellpose.io.get_image_files(folder, mask_filter, imf=None, look_one_level_down=False)[source]

Finds all images in a folder and its subfolders (if specified) with the given file extensions.

Parameters
  • folder (str) – The path to the folder to search for images.

  • mask_filter (str) – The filter for mask files.

  • imf (str, optional) – The additional filter for image files. Defaults to None.

  • look_one_level_down (bool, optional) – Whether to search for images in subfolders. Defaults to False.

Returns

A list of image file paths.

Return type

list

Raises
  • ValueError – If no files are found in the specified folder.

  • ValueError – If no images are found in the specified folder with the supported file extensions.

  • ValueError – If no images are found in the specified folder without the mask or flow file endings.

cellpose.io.get_label_files(image_names, mask_filter, imf=None)[source]

Get the label files corresponding to the given image names and mask filter.

Parameters
  • image_names (list) – List of image names.

  • mask_filter (str) – Mask filter to be applied.

  • imf (str, optional) – Image file extension. Defaults to None.

Returns

A tuple containing the label file names and flow file names (if present).

Return type

tuple

cellpose.io.imread(filename)[source]

Read in an image file with tif or image file type supported by cv2.

Parameters

filename (str) – The path to the image file.

Returns

The image data as a NumPy array.

Return type

numpy.ndarray

Raises

None

Raises an error if the image file format is not supported.

Examples

>>> img = imread("image.tif")
cellpose.io.imsave(filename, arr)[source]

Saves an image array to a file.

Parameters
  • filename (str) – The name of the file to save the image to.

  • arr (numpy.ndarray) – The image array to be saved.

Returns

None

cellpose.io.load_images_labels(tdir, mask_filter='_masks', image_filter=None, look_one_level_down=False)[source]

Loads images and corresponding labels from a directory.

Parameters
  • tdir (str) – The directory path.

  • mask_filter (str, optional) – The filter for mask files. Defaults to “_masks”.

  • image_filter (str, optional) – The filter for image files. Defaults to None.

  • look_one_level_down (bool, optional) – Whether to look for files one level down. Defaults to False.

Returns

A tuple containing a list of images, a list of labels, and a list of image names.

Return type

tuple

cellpose.io.load_train_test_data(train_dir, test_dir=None, image_filter=None, mask_filter='_masks', look_one_level_down=False)[source]

Loads training and testing data for a Cellpose model.

Parameters
  • train_dir (str) – The directory path containing the training data.

  • test_dir (str, optional) – The directory path containing the testing data. Defaults to None.

  • image_filter (str, optional) – The filter for selecting image files. Defaults to None.

  • mask_filter (str, optional) – The filter for selecting mask files. Defaults to “_masks”.

  • look_one_level_down (bool, optional) – Whether to look for data in subdirectories of train_dir and test_dir. Defaults to False.

Returns

A list of training images. labels (list): A list of labels corresponding to the training images. image_names (list): A list of names of the training images. test_images (list, optional): A list of testing images. None if test_dir is not provided. test_labels (list, optional): A list of labels corresponding to the testing images. None if test_dir is not provided. test_image_names (list, optional): A list of names of the testing images. None if test_dir is not provided.

Return type

images (list)

cellpose.io.masks_flows_to_seg(images, masks, flows, file_names, diams=30.0, channels=None, imgs_restore=None, restore_type=None, ratio=1.0)[source]

Save output of model eval to be loaded in GUI.

Can be list output (run on multiple images) or single output (run on single image).

Saved to file_names[k]+”_seg.npy”.

Parameters
  • images (list) – Images input into cellpose.

  • masks (list) – Masks output from Cellpose.eval, where 0=NO masks; 1,2,…=mask labels.

  • flows (list) – Flows output from Cellpose.eval.

  • file_names (list, str) – Names of files of images.

  • diams (float array) – Diameters used to run Cellpose. Defaults to 30.

  • channels (list, int, optional) – Channels used to run Cellpose. Defaults to None.

Returns

None

cellpose.io.remove_model(filename, delete=False)[source]

remove model from .cellpose custom model list

cellpose.io.save_masks(images, masks, flows, file_names, png=True, tif=False, channels=[0, 0], suffix='', save_flows=False, save_outlines=False, dir_above=False, in_folders=False, savedir=None, save_txt=False, save_mpl=False)[source]

Save masks + nicely plotted segmentation image to png and/or tiff.

Can save masks, flows to different directories, if in_folders is True.

If png, masks[k] for images[k] are saved to file_names[k]+”_cp_masks.png”.

If tif, masks[k] for images[k] are saved to file_names[k]+”_cp_masks.tif”.

If png and matplotlib installed, full segmentation figure is saved to file_names[k]+”_cp.png”.

Only tif option works for 3D data, and only tif option works for empty masks.

Parameters
  • images (list) – Images input into cellpose.

  • masks (list) – Masks output from Cellpose.eval, where 0=NO masks; 1,2,…=mask labels.

  • flows (list) – Flows output from Cellpose.eval.

  • file_names (list, str) – Names of files of images.

  • png (bool, optional) – Save masks to PNG. Defaults to True.

  • tif (bool, optional) – Save masks to TIF. Defaults to False.

  • channels (list, int, optional) – Channels used to run Cellpose. Defaults to [0,0].

  • suffix (str, optional) – Add name to saved masks. Defaults to “”.

  • save_flows (bool, optional) – Save flows output from Cellpose.eval. Defaults to False.

  • save_outlines (bool, optional) – Save outlines of masks. Defaults to False.

  • dir_above (bool, optional) – Save masks/flows in directory above. Defaults to False.

  • in_folders (bool, optional) – Save masks/flows in separate folders. Defaults to False.

  • savedir (str, optional) – Absolute path where images will be saved. If None, saves to image directory. Defaults to None.

  • save_txt (bool, optional) – Save masks as list of outlines for ImageJ. Defaults to False.

  • save_mpl (bool, optional) – If True, saves a matplotlib figure of the original image/segmentation/flows. Does not work for 3D. This takes a long time for large images. Defaults to False.

Returns

None

cellpose.io.save_rois(masks, file_name)[source]

save masks to .roi files in .zip archive for ImageJ/Fiji

Parameters
  • masks (np.ndarray) – masks output from Cellpose.eval, where 0=NO masks; 1,2,…=mask labels

  • file_name (str) – name to save the .zip file to

Returns

None

cellpose.io.save_server(parent=None, filename=None)[source]

Uploads a *_seg.npy file to the bucket.

Parameters
  • parent (PyQt.MainWindow, optional) – GUI window to grab file info from. Defaults to None.

  • filename (str, optional) – if no GUI, send this file to server. Defaults to None.

cellpose.io.save_to_png(images, masks, flows, file_names)[source]

deprecated (runs io.save_masks with png=True)

does not work for 3D images

Utils functions

Copyright © 2023 Howard Hughes Medical Institute, Authored by Carsen Stringer and Marius Pachitariu.

class cellpose.utils.TqdmToLogger(logger, level=None)[source]

Output stream for TQDM which will output to logger module instead of the StdOut.

flush()[source]

Flush write buffers, if applicable.

This is not implemented for read-only and non-blocking streams.

write(buf)[source]

Write string to file.

Returns the number of characters written, which is always equal to the length of the string.

cellpose.utils.circleMask(d0)[source]

Creates an array with indices which are the radius of that x,y point.

Parameters

d0 (tuple) – Patch of (-d0, d0+1) over which radius is computed.

Returns

A tuple containing:
  • rs (ndarray): Array of radii with shape (2*d0[0]+1, 2*d0[1]+1).

  • dx (ndarray): Indices of the patch along the x-axis.

  • dy (ndarray): Indices of the patch along the y-axis.

Return type

tuple

cellpose.utils.diameters(masks)[source]

Calculate the diameters of the objects in the given masks.

Parameters: masks (ndarray): masks (0=no cells, 1=first cell, 2=second cell,…)

Returns: tuple: A tuple containing the median diameter and an array of diameters for each object.

Examples: >>> masks = np.array([[0, 1, 1], [1, 0, 0], [1, 1, 0]]) >>> diameters(masks) (1.0, array([1.41421356, 1.0, 1.0]))

cellpose.utils.dilate_masks(masks, n_iter=5)[source]

Dilate masks by n_iter pixels.

Parameters
  • masks (ndarray) – Array of masks.

  • n_iter (int, optional) – Number of pixels to dilate the masks. Defaults to 5.

Returns

Dilated masks.

Return type

ndarray

cellpose.utils.distance_to_boundary(masks)[source]

Get the distance to the boundary of mask pixels.

Parameters

masks (int, 2D or 3D array) – The masks array. Size [Ly x Lx] or [Lz x Ly x Lx], where 0 represents no mask and 1, 2, … represent mask labels.

Returns

The distance to the boundary. Size [Ly x Lx] or [Lz x Ly x Lx].

Return type

dist_to_bound (2D or 3D array)

Raises

ValueError – If the masks array is not 2D or 3D.

cellpose.utils.download_url_to_file(url, dst, progress=True)[source]
Download object at the given URL to a local path.

Thanks to torch, slightly modified

Parameters
  • url (string) – URL of the object to download

  • dst (string) – Full path where object will be saved, e.g. /tmp/temporary_file

  • progress (bool, optional) – whether or not to display a progress bar to stderr Default: True

cellpose.utils.fill_holes_and_remove_small_masks(masks, min_size=15)[source]

Fills holes in masks (2D/3D) and discards masks smaller than min_size.

This function fills holes in each mask using scipy.ndimage.morphology.binary_fill_holes. It also removes masks that are smaller than the specified min_size.

Parameters: masks (ndarray): Int, 2D or 3D array of labelled masks.

0 represents no mask, while positive integers represent mask labels. The size can be [Ly x Lx] or [Lz x Ly x Lx].

min_size (int, optional): Minimum number of pixels per mask.

Masks smaller than min_size will be removed. Set to -1 to turn off this functionality. Default is 15.

Returns: ndarray: Int, 2D or 3D array of masks with holes filled and small masks removed.

0 represents no mask, while positive integers represent mask labels. The size is [Ly x Lx] or [Lz x Ly x Lx].

cellpose.utils.get_mask_compactness(masks)[source]

Calculate the compactness of masks.

Parameters

masks (ndarray) – Binary masks representing objects.

Returns

Array of compactness values for each mask.

Return type

ndarray

cellpose.utils.get_mask_perimeters(masks)[source]

Calculate the perimeters of the given masks.

Parameters

masks (numpy.ndarray) – Binary masks representing objects.

Returns

Array containing the perimeters of each mask.

Return type

numpy.ndarray

cellpose.utils.get_mask_stats(masks_true)[source]

Calculate various statistics for the given binary masks.

Parameters

masks_true (ndarray) – masks (0=no cells, 1=first cell, 2=second cell,…)

Returns

Convexity values for each mask. solidity (ndarray): Solidity values for each mask. compactness (ndarray): Compactness values for each mask.

Return type

convexity (ndarray)

cellpose.utils.get_masks_unet(output, cell_threshold=0, boundary_threshold=0)[source]

Create masks using cell probability and cell boundary.

Parameters
  • output (ndarray) – The output array containing cell probability and cell boundary.

  • cell_threshold (float, optional) – The threshold value for cell probability. Defaults to 0.

  • boundary_threshold (float, optional) – The threshold value for cell boundary. Defaults to 0.

Returns

The masks representing the segmented cells.

Return type

ndarray

cellpose.utils.get_outline_multi(args)[source]

Get the outline of a specific mask in a multi-mask image.

Parameters

args (tuple) – A tuple containing the masks and the mask number.

Returns

The outline of the specified mask as an array of coordinates.

Return type

numpy.ndarray

cellpose.utils.get_perimeter(points)[source]

Calculate the perimeter of a set of points.

Parameters

points (ndarray) – An array of points with shape (npoints, ndim).

Returns

The perimeter of the points.

Return type

float

cellpose.utils.masks_to_edges(masks, threshold=1.0)[source]

Get edges of masks as a 0-1 array.

Parameters
  • masks (int, 2D or 3D array) – Size [Ly x Lx] or [Lz x Ly x Lx], where 0=NO masks and 1,2,…=mask labels.

  • threshold (float, optional) – Threshold value for distance to boundary. Defaults to 1.0.

Returns

Size [Ly x Lx] or [Lz x Ly x Lx], where True pixels are edge pixels.

Return type

edges (2D or 3D array)

cellpose.utils.masks_to_outlines(masks)[source]

Get outlines of masks as a 0-1 array.

Parameters

masks (int, 2D or 3D array) – Size [Ly x Lx] or [Lz x Ly x Lx], where 0=NO masks and 1,2,…=mask labels.

Returns

Size [Ly x Lx] or [Lz x Ly x Lx], where True pixels are outlines.

Return type

outlines (2D or 3D array)

cellpose.utils.outlines_list(masks, multiprocessing_threshold=1000, multiprocessing=None)[source]

Get outlines of masks as a list to loop over for plotting.

Parameters
  • masks (ndarray) – Array of masks.

  • multiprocessing_threshold (int, optional) – Threshold for enabling multiprocessing. Defaults to 1000.

  • multiprocessing (bool, optional) – Flag to enable multiprocessing. Defaults to None.

Returns

List of outlines.

Return type

list

Raises

None

Notes

  • This function is a wrapper for outlines_list_single and outlines_list_multi.

  • Multiprocessing is disabled for Windows.

cellpose.utils.outlines_list_multi(masks, num_processes=None)[source]

Get outlines of masks as a list to loop over for plotting.

Parameters

masks (ndarray) – masks (0=no cells, 1=first cell, 2=second cell,…)

Returns

List of outlines as pixel coordinates.

Return type

list

cellpose.utils.outlines_list_single(masks)[source]

Get outlines of masks as a list to loop over for plotting.

Parameters

masks (ndarray) – masks (0=no cells, 1=first cell, 2=second cell,…)

Returns

List of outlines as pixel coordinates.

Return type

list

cellpose.utils.radius_distribution(masks, bins)[source]

Calculate the radius distribution of masks.

Parameters
  • masks (ndarray) – masks (0=no cells, 1=first cell, 2=second cell,…)

  • bins (int) – Number of bins for the histogram.

Returns

A tuple containing:
  • nb (ndarray): Normalized histogram of radii.

  • md (float): Median radius.

  • radii (ndarray): Array of radii.

Return type

tuple

cellpose.utils.remove_edge_masks(masks, change_index=True)[source]

Removes masks with pixels on the edge of the image.

Parameters
  • masks (int, 2D or 3D array) – The masks to be processed. Size [Ly x Lx] or [Lz x Ly x Lx], where 0 represents no mask and 1, 2, … represent mask labels.

  • change_index (bool, optional) – If True, after removing masks, changes the indexing so that there are no missing label numbers. Defaults to True.

Returns

The processed masks. Size [Ly x Lx] or [Lz x Ly x Lx], where 0 represents no mask and 1, 2, … represent mask labels.

Return type

outlines (2D or 3D array)

cellpose.utils.size_distribution(masks)[source]

Calculates the size distribution of masks.

Parameters

masks (ndarray) – masks (0=no cells, 1=first cell, 2=second cell,…)

Returns

The ratio of the 25th percentile of mask sizes to the 75th percentile of mask sizes.

Return type

float

cellpose.utils.stitch3D(masks, stitch_threshold=0.25)[source]

Stitch 2D masks into a 3D volume using a stitch_threshold on IOU.

Parameters
  • masks (list or ndarray) – List of 2D masks.

  • stitch_threshold (float, optional) – Threshold value for stitching. Defaults to 0.25.

Returns

List of stitched 3D masks.

Return type

list

Network classes

Copyright © 2023 Howard Hughes Medical Institute, Authored by Carsen Stringer and Marius Pachitariu.

class cellpose.resnet_torch.CPnet(*args: Any, **kwargs: Any)[source]

CPnet is the Cellpose neural network model used for cell segmentation and image restoration.

Parameters
  • nbase (list) – List of integers representing the number of channels in each layer of the downsample path.

  • nout (int) – Number of output channels.

  • sz (int) – Size of the input image.

  • mkldnn (bool, optional) – Whether to use MKL-DNN acceleration. Defaults to False.

  • conv_3D (bool, optional) – Whether to use 3D convolution. Defaults to False.

  • max_pool (bool, optional) – Whether to use max pooling. Defaults to True.

  • diam_mean (float, optional) – Mean diameter of the cells. Defaults to 30.0.

nbase

List of integers representing the number of channels in each layer of the downsample path.

Type

list

nout

Number of output channels.

Type

int

sz

Size of the input image.

Type

int

residual_on

Whether to use residual connections.

Type

bool

style_on

Whether to use style transfer.

Type

bool

concatenation

Whether to use concatenation.

Type

bool

conv_3D

Whether to use 3D convolution.

Type

bool

mkldnn

Whether to use MKL-DNN acceleration.

Type

bool

downsample

Downsample blocks of the network.

Type

nn.Module

upsample

Upsample blocks of the network.

Type

nn.Module

make_style

Style module, avgpool’s over all spatial positions.

Type

nn.Module

output

Output module - batchconv layer.

Type

nn.Module

diam_mean

Parameter representing the mean diameter to which the cells are rescaled to during training.

Type

nn.Parameter

diam_labels

Parameter representing the mean diameter of the cells in the training set (before rescaling).

Type

nn.Parameter

property device

Get the device of the model.

Returns

The device of the model.

Return type

torch.device

forward(data)[source]

Forward pass of the CPnet model.

Parameters

data (torch.Tensor) – Input data.

Returns

A tuple containing the output tensor, style tensor, and downsampled tensors.

Return type

tuple

load_model(filename, device=None)[source]

Load the model from a file.

Parameters
  • filename (str) – The path to the file where the model is saved.

  • device (torch.device, optional) – The device to load the model on. Defaults to None.

save_model(filename)[source]

Save the model to a file.

Parameters

filename (str) – The path to the file where the model will be saved.

class cellpose.resnet_torch.batchconvstyle(*args: Any, **kwargs: Any)[source]
class cellpose.resnet_torch.convup(*args: Any, **kwargs: Any)[source]
class cellpose.resnet_torch.downsample(*args: Any, **kwargs: Any)[source]
class cellpose.resnet_torch.make_style(*args: Any, **kwargs: Any)[source]
class cellpose.resnet_torch.resdown(*args: Any, **kwargs: Any)[source]
class cellpose.resnet_torch.resup(*args: Any, **kwargs: Any)[source]
class cellpose.resnet_torch.upsample(*args: Any, **kwargs: Any)[source]

Core functions

Copyright © 2023 Howard Hughes Medical Institute, Authored by Carsen Stringer and Marius Pachitariu.

cellpose.core.assign_device(use_torch=True, gpu=False, device=0)[source]

Assigns the device (CPU or GPU or mps) to be used for computation.

Parameters
  • use_torch (bool, optional) – Whether to use torch for GPU detection. Defaults to True.

  • gpu (bool, optional) – Whether to use GPU for computation. Defaults to False.

  • device (int or str, optional) – The device index or name to be used. Defaults to 0.

Returns

The assigned device. bool: True if GPU is used, False otherwise.

Return type

torch.device

cellpose.core.check_mkl(use_torch=True)[source]

Checks if MKL-DNN is enabled and working.

Parameters

use_torch (bool, optional) – Whether to use torch. Defaults to True.

Returns

True if MKL-DNN is enabled, False otherwise.

Return type

bool

cellpose.core.run_3D(net, imgs, batch_size=8, rsz=1.0, anisotropy=None, augment=False, tile=True, tile_overlap=0.1, bsize=224, progress=None)[source]

Run network on image z-stack.

(faster if augment is False)

Parameters
  • imgs (np.ndarray) – The input image stack of size [Lz x Ly x Lx x nchan].

  • batch_size (int, optional) – Number of tiles to run in a batch. Defaults to 8.

  • rsz (float, optional) – Resize coefficient(s) for image. Defaults to 1.0.

  • anisotropy (float, optional) – for 3D segmentation, optional rescaling factor (e.g. set to 2.0 if Z is sampled half as dense as X or Y). Defaults to None.

  • augment (bool, optional) – Tiles image with overlapping tiles and flips overlapped regions to augment. Defaults to False.

  • tile (bool, optional) – Tiles image to ensure GPU/CPU memory usage limited (recommended); cannot be turned off for 3D segmentation. Defaults to True.

  • tile_overlap (float, optional) – Fraction of overlap of tiles when computing flows. Defaults to 0.1.

  • bsize (int, optional) – Size of tiles to use in pixels [bsize x bsize]. Defaults to 224.

  • progress (QProgressBar, optional) – pyqt progress bar. Defaults to None.

Returns

output of network, if tiled it is averaged in tile overlaps. Size of [Ly x Lx x 3] or [Lz x Ly x Lx x 3].

y[…,0] is Y flow; y[…,1] is X flow; y[…,2] is cell probability.

style (np.ndarray): 1D array of size 256 summarizing the style of the image, if tiled it is averaged over tiles.

Return type

y (np.ndarray)

cellpose.core.run_net(net, imgs, batch_size=8, augment=False, tile=True, tile_overlap=0.1, bsize=224)[source]

Run network on image or stack of images.

(faster if augment is False)

Parameters
  • imgs (np.ndarray) – The input image or stack of images of size [Ly x Lx x nchan] or [Lz x Ly x Lx x nchan].

  • batch_size (int, optional) – Number of tiles to run in a batch. Defaults to 8.

  • rsz (float, optional) – Resize coefficient(s) for image. Defaults to 1.0.

  • augment (bool, optional) – Tiles image with overlapping tiles and flips overlapped regions to augment. Defaults to False.

  • tile (bool, optional) – Tiles image to ensure GPU/CPU memory usage limited (recommended); cannot be turned off for 3D segmentation. Defaults to True.

  • tile_overlap (float, optional) – Fraction of overlap of tiles when computing flows. Defaults to 0.1.

  • bsize (int, optional) – Size of tiles to use in pixels [bsize x bsize]. Defaults to 224.

Returns

output of network, if tiled it is averaged in tile overlaps. Size of [Ly x Lx x 3] or [Lz x Ly x Lx x 3].

y[…,0] is Y flow; y[…,1] is X flow; y[…,2] is cell probability.

style (np.ndarray): 1D array of size 256 summarizing the style of the image, if tiled it is averaged over tiles.

Return type

y (np.ndarray)

cellpose.core.use_gpu(gpu_number=0, use_torch=True)[source]

Check if GPU is available for use.

Parameters
  • gpu_number (int) – The index of the GPU to be used. Default is 0.

  • use_torch (bool) – Whether to use PyTorch for GPU check. Default is True.

Returns

True if GPU is available, False otherwise.

Return type

bool

Raises

ValueError – If use_torch is False, as cellpose only runs with PyTorch now.

All models functions

Copyright © 2023 Howard Hughes Medical Institute, Authored by Carsen Stringer and Marius Pachitariu.

class cellpose.models.Cellpose(gpu=False, model_type='cyto3', device=None)[source]

Main model which combines SizeModel and CellposeModel.

Parameters
  • gpu (bool, optional) – Whether or not to use GPU, will check if GPU available. Defaults to False.

  • model_type (str, optional) – Model type. “cyto”=cytoplasm model; “nuclei”=nucleus model; “cyto2”=cytoplasm model with additional user images; “cyto3”=super-generalist model; Defaults to “cyto3”.

  • device (torch device, optional) – Device used for model running / training. Overrides gpu input. Recommended if you want to use a specific GPU (e.g. torch.device(“cuda:1”)). Defaults to None.

device

Device used for model running / training.

Type

torch device

gpu

Flag indicating if GPU is used.

Type

bool

diam_mean

Mean diameter for cytoplasm model.

Type

float

cp

CellposeModel instance.

Type

CellposeModel

pretrained_size

Pretrained size model path.

Type

str

sz

SizeModel instance.

Type

SizeModel

eval(x, batch_size=8, channels=[0, 0], channel_axis=None, invert=False, normalize=True, diameter=30.0, do_3D=False, **kwargs)[source]

Run cellpose size model and mask model and get masks.

Parameters
  • x (list or array) – List or array of images. Can be list of 2D/3D images, or array of 2D/3D images, or 4D image array.

  • batch_size (int, optional) – Number of 224x224 patches to run simultaneously on the GPU. Can make smaller or bigger depending on GPU memory usage. Defaults to 8.

  • channels (list, optional) – List of channels, either of length 2 or of length number of images by 2. First element of list is the channel to segment (0=grayscale, 1=red, 2=green, 3=blue). Second element of list is the optional nuclear channel (0=none, 1=red, 2=green, 3=blue). For instance, to segment grayscale images, input [0,0]. To segment images with cells in green and nuclei in blue, input [2,3]. To segment one grayscale image and one image with cells in green and nuclei in blue, input [[0,0], [2,3]]. Defaults to [0,0].

  • channel_axis (int, optional) – If None, channels dimension is attempted to be automatically determined. Defaults to None.

  • invert (bool, optional) – Invert image pixel intensity before running network (if True, image is also normalized). Defaults to False.

  • normalize (bool, optional) – If True, normalize data so 0.0=1st percentile and 1.0=99th percentile of image intensities in each channel; can also pass dictionary of parameters (see CellposeModel for details). Defaults to True.

  • diameter (float, optional) – If set to None, then diameter is automatically estimated if size model is loaded. Defaults to 30..

  • do_3D (bool, optional) – Set to True to run 3D segmentation on 4D image input. Defaults to False.

Returns

tuple containing
  • masks (list of 2D arrays or single 3D array): Labelled image, where 0=no masks; 1,2,…=mask labels.

  • flows (list of lists 2D arrays or list of 3D arrays):
    • flows[k][0] = XY flow in HSV 0-255

    • flows[k][1] = XY flows at each pixel

    • flows[k][2] = cell probability (if > cellprob_threshold, pixel used for dynamics)

    • flows[k][3] = final pixel locations after Euler integration

  • styles (list of 1D arrays of length 256 or single 1D array): Style vector summarizing each image, also used to estimate size of objects in image.

  • diams (list of diameters or float): List of diameters or float (if do_3D=True).

class cellpose.models.CellposeModel(gpu=False, pretrained_model=False, model_type=None, diam_mean=30.0, device=None, nchan=2)[source]

Class representing a Cellpose model.

diam_mean

Mean “diameter” value for the model.

Type

float

builtin

Whether the model is a built-in model or not.

Type

bool

device

Device used for model running / training.

Type

torch device

mkldnn

MKLDNN flag for the model.

Type

None or bool

nchan

Number of channels used as input to the network.

Type

int

nclasses

Number of classes in the model.

Type

int

nbase

List of base values for the model.

Type

list

net

Cellpose network.

Type

CPnet

pretrained_model

Full path to pretrained cellpose model(s).

Type

str or list of strings

diam_labels

Diameter labels of the model.

Type

numpy array

net_type

Type of the network.

Type

str

__init__(self, gpu=False, pretrained_model=False, model_type=None, diam_mean=30., device=None, nchan=2)[source]

Initialize the CellposeModel.

eval(self, x, batch_size=8, resample=True, channels=None, channel_axis=None, z_axis=None, normalize=True, invert=False, rescale=None, diameter=None, flow_threshold=0.4, cellprob_threshold=0.0, do_3D=False, anisotropy=None, stitch_threshold=0.0, min_size=15, niter=None, augment=False, tile=True, tile_overlap=0.1, bsize=224, interp=True, compute_masks=True, progress=None)[source]

Segment list of images x, or 4D array - Z x nchan x Y x X.

eval(x, batch_size=8, resample=True, channels=None, channel_axis=None, z_axis=None, normalize=True, invert=False, rescale=None, diameter=None, flow_threshold=0.4, cellprob_threshold=0.0, do_3D=False, anisotropy=None, stitch_threshold=0.0, min_size=15, niter=None, augment=False, tile=True, tile_overlap=0.1, bsize=224, interp=True, compute_masks=True, progress=None)[source]

segment list of images x, or 4D array - Z x nchan x Y x X

Parameters
  • x (list, np.ndarry) – can be list of 2D/3D/4D images, or array of 2D/3D/4D images

  • batch_size (int, optional) – number of 224x224 patches to run simultaneously on the GPU (can make smaller or bigger depending on GPU memory usage). Defaults to 8.

  • resample (bool, optional) – run dynamics at original image size (will be slower but create more accurate boundaries). Defaults to True.

  • channels (list, optional) – list of channels, either of length 2 or of length number of images by 2. First element of list is the channel to segment (0=grayscale, 1=red, 2=green, 3=blue). Second element of list is the optional nuclear channel (0=none, 1=red, 2=green, 3=blue). For instance, to segment grayscale images, input [0,0]. To segment images with cells in green and nuclei in blue, input [2,3]. To segment one grayscale image and one image with cells in green and nuclei in blue, input [[0,0], [2,3]]. Defaults to None.

  • channel_axis (int, optional) – channel axis in element of list x, or of np.ndarray x. if None, channels dimension is attempted to be automatically determined. Defaults to None.

  • z_axis (int, optional) – z axis in element of list x, or of np.ndarray x. if None, z dimension is attempted to be automatically determined. Defaults to None.

  • normalize (bool, optional) –

    if True, normalize data so 0.0=1st percentile and 1.0=99th percentile of image intensities in each channel; can also pass dictionary of parameters (all keys are optional, default values shown):

    • ”lowhigh”=None : pass in normalization values for 0.0 and 1.0 as list [low, high] (if not None, all following parameters ignored)

    • ”sharpen”=0 ; sharpen image with high pass filter, recommended to be 1/4-1/8 diameter of cells in pixels

    • ”normalize”=True ; run normalization (if False, all following parameters ignored)

    • ”percentile”=None : pass in percentiles to use as list [perc_low, perc_high]

    • ”tile_norm”=0 ; compute normalization in tiles across image to brighten dark areas, to turn on set to window size in pixels (e.g. 100)

    • ”norm3D”=False ; compute normalization across entire z-stack rather than plane-by-plane in stitching mode.

    Defaults to True.

  • invert (bool, optional) – invert image pixel intensity before running network. Defaults to False.

  • rescale (float, optional) – resize factor for each image, if None, set to 1.0; (only used if diameter is None). Defaults to None.

  • diameter (float, optional) – diameter for each image, if diameter is None, set to diam_mean or diam_train if available. Defaults to None.

  • flow_threshold (float, optional) – flow error threshold (all cells with errors below threshold are kept) (not used for 3D). Defaults to 0.4.

  • cellprob_threshold (float, optional) – all pixels with value above threshold kept for masks, decrease to find more and larger masks. Defaults to 0.0.

  • do_3D (bool, optional) – set to True to run 3D segmentation on 3D/4D image input. Defaults to False.

  • anisotropy (float, optional) – for 3D segmentation, optional rescaling factor (e.g. set to 2.0 if Z is sampled half as dense as X or Y). Defaults to None.

  • stitch_threshold (float, optional) – if stitch_threshold>0.0 and not do_3D, masks are stitched in 3D to return volume segmentation. Defaults to 0.0.

  • min_size (int, optional) – all ROIs below this size, in pixels, will be discarded. Defaults to 15.

  • niter (int, optional) – number of iterations for dynamics computation. if None, it is set proportional to the diameter. Defaults to None.

  • augment (bool, optional) – tiles image with overlapping tiles and flips overlapped regions to augment. Defaults to False.

  • tile (bool, optional) – tiles image to ensure GPU/CPU memory usage limited (recommended). Defaults to True.

  • tile_overlap (float, optional) – fraction of overlap of tiles when computing flows. Defaults to 0.1.

  • bsize (int, optional) – block size for tiles, recommended to keep at 224, like in training. Defaults to 224.

  • interp (bool, optional) – interpolate during 2D dynamics (not available in 3D) . Defaults to True.

  • compute_masks (bool, optional) – Whether or not to compute dynamics and return masks. This is set to False when retrieving the styles for the size model. Defaults to True.

  • progress (QProgressBar, optional) – pyqt progress bar. Defaults to None.

Returns

  • masks (list, np.ndarray): labelled image(s), where 0=no masks; 1,2,…=mask labels

  • flows (list): list of lists: flows[k][0] = XY flow in HSV 0-255; flows[k][1] = XY(Z) flows at each pixel; flows[k][2] = cell probability (if > cellprob_threshold, pixel used for dynamics); flows[k][3] = final pixel locations after Euler integration

  • styles (list, np.ndarray): style vector summarizing each image of size 256.

Return type

A tuple containing

class cellpose.models.SizeModel(cp_model, device=None, pretrained_size=None, **kwargs)[source]

Linear regression model for determining the size of objects in image used to rescale before input to cp_model. Uses styles from cp_model.

pretrained_size

Path to pretrained size model.

Type

str

cp

Model from which to get styles.

Type

UnetModel or CellposeModel

device

Device used for model running / training (torch.device(“cuda”) or torch.device(“cpu”)), overrides gpu input, recommended if you want to use a specific GPU (e.g. torch.device(“cuda:1”)).

Type

torch device

diam_mean

Mean diameter of objects.

Type

float

eval(self, x, channels=None, channel_axis=None, normalize=True, invert=False,

augment=False, tile=True, batch_size=8, progress=None, interp=True):

Use images x to produce style or use style input to predict size of objects in image.

Raises

ValueError – If no pretrained cellpose model is specified, cannot compute size.

eval(x, channels=None, channel_axis=None, normalize=True, invert=False, augment=False, tile=True, batch_size=8, progress=None)[source]

Use images x to produce style or use style input to predict size of objects in image.

Object size estimation is done in two steps: 1. Use a linear regression model to predict size from style in image. 2. Resize image to predicted size and run CellposeModel to get output masks.

Take the median object size of the predicted masks as the final predicted size.

Parameters
  • x (list, np.ndarry) – can be list of 2D/3D/4D images, or array of 2D/3D/4D images

  • channels (list, optional) – list of channels, either of length 2 or of length number of images by 2. First element of list is the channel to segment (0=grayscale, 1=red, 2=green, 3=blue). Second element of list is the optional nuclear channel (0=none, 1=red, 2=green, 3=blue). For instance, to segment grayscale images, input [0,0]. To segment images with cells in green and nuclei in blue, input [2,3]. To segment one grayscale image and one image with cells in green and nuclei in blue, input [[0,0], [2,3]]. Defaults to None.

  • channel_axis (int, optional) – channel axis in element of list x, or of np.ndarray x. if None, channels dimension is attempted to be automatically determined. Defaults to None.

  • normalize (bool, optional) –

    if True, normalize data so 0.0=1st percentile and 1.0=99th percentile of image intensities in each channel; can also pass dictionary of parameters (all keys are optional, default values shown):

    • ”lowhigh”=None : pass in normalization values for 0.0 and 1.0 as list [low, high] (if not None, all following parameters ignored)

    • ”sharpen”=0 ; sharpen image with high pass filter, recommended to be 1/4-1/8 diameter of cells in pixels

    • ”normalize”=True ; run normalization (if False, all following parameters ignored)

    • ”percentile”=None : pass in percentiles to use as list [perc_low, perc_high]

    • ”tile_norm”=0 ; compute normalization in tiles across image to brighten dark areas, to turn on set to window size in pixels (e.g. 100)

    • ”norm3D”=False ; compute normalization across entire z-stack rather than plane-by-plane in stitching mode.

    Defaults to True.

  • invert (bool, optional) – Invert image pixel intensity before running network (if True, image is also normalized). Defaults to False.

  • augment (bool, optional) – tiles image with overlapping tiles and flips overlapped regions to augment. Defaults to False.

  • tile (bool, optional) – tiles image to ensure GPU/CPU memory usage limited (recommended). Defaults to True.

  • batch_size (int, optional) – number of 224x224 patches to run simultaneously on the GPU (can make smaller or bigger depending on GPU memory usage). Defaults to 8.

  • progress (QProgressBar, optional) – pyqt progress bar. Defaults to None.

Returns

  • diam (np.ndarray): Final estimated diameters from images x or styles style after running both steps.

  • diam_style (np.ndarray): Estimated diameters from style alone.

Return type

A tuple containing