create_masks.py 文件源码

def create_roi_masks(centers_of_mass, putative_nuclei_image, putative_somata_image=None, radius=3):
        """
        create roi masks for the outer segment of the cell (i.e. soma)

        :param radius: limits the size of the mask
        :return roi_mask_list: a list of pixels for each cell, for further analysis
        """
        roi_mask_list = []

        if putative_somata_image is None:
            putative_somata_image = np.zeros_like(putative_nuclei_image)
        putative_nuclei_image = remove_small_blobs(centers_of_mass, putative_nuclei_image)[0]
        watershed_image = np.logical_or(putative_nuclei_image, putative_somata_image)
        labelled_watershed = calculate_distance(centers_of_mass, watershed_image)
        labelled_putative_somata = putative_somata_image*labelled_watershed
        labelled_putative_nuclei = calculate_distance(centers_of_mass, putative_nuclei_image)*putative_nuclei_image  # nuclei need their own watershed

        for i in range(np.max(labelled_putative_somata)):  # for each nucleus

            # calculate the distance away from the nucleus boundary

            distance_from_blob_centre = ndimage.distance_transform_edt(labelled_putative_nuclei != i+1)
            bool_mask = np.ones_like(labelled_putative_nuclei)
            bool_mask[distance_from_blob_centre > radius] = 0
            bool_mask[distance_from_blob_centre == 0] = 0

            # take all indices within the radius number of pixels of the nucleus boundary

            idx = np.where(labelled_putative_somata*bool_mask == i+1)
            x = np.vstack((idx[0], idx[1]))
            m = np.reshape(x.T, (len(idx[0]), 2))
            roi_mask_list.append(m)

        return roi_mask_list