stereo.image.segmentation.segment.cell_seg#

stereo.image.segmentation.segment.cell_seg(model_path, img_path, out_path, deep_crop_size=20000, overlap=100, post_processing_workers=10, is_water=False, method='v3', need_tissue_cut=True, tissue_seg_model_path=None, tissue_seg_staining_type=None, gpu='-1', num_threads=-1)[source]#

Cell segmentation on regist.tif/mask.tif by deeplearning model.

Parameters:

  • model_path (str) – the path of model used to cell segmentation.

  • img_path (str) – the path of regist.tif/mask.tif.

  • out_path (str) – the path of directory to save the result cell mask.tif.

  • deep_crop_size (int) – deep crop size, defaults to 20000

  • overlap (int) – over lap size, defaults to 100

  • post_processing_workers (int) – the number of processes on post processing, defaults to 10.

  • is_water (bool) – defaults to False.

  • method (str) – v1, v1_pro or v3, recommend to use v3.

  • need_tissue_cut – whether to run tissue segmentation, defaults to True. the method v1 and v1_pro have to run tissue segmentation, so these two methods must be based on regist.tif, method v3 can use mask.tif from tissue segmentation or regist.tif without tissue segmentation.

  • tissue_seg_model_path (Optional[str]) – the path of model used to tissue segmentation, defaults to None

  • tissue_seg_staining_type (Optional[str]) – the staining type of regist.mask, defaults to None

  • gpu (str) – the gpu on which the model works, available for both cell segmtation and tissue segmtation, ‘-1’ means working on cpu.

  • num_threads (int) – the number of threads when model work on cpu, available for both v3 cell segmtation and tissue segmtation, -1 means using all the cores.