HPACellSegmentatorPortable/generate_cell_crops.py at main · CellProfiling/HPACellSegmentatorPortable · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
import warnings
import cv2
import numpy as np
from skimage.measure import regionprops
from scipy.ndimage import grey_dilation
import pandas as pd
import image_utils


warnings.simplefilter(action="ignore", category=FutureWarning)
colors = ["red", "yellow", "blue", "green"]


def safe_crop(image, bbox):
    y1, x1, y2, x2 = bbox
    img_h, img_w = image.shape[:2]
    is_single_channel = len(image.shape) == 2
    if x1 < 0:
        pad_x1 = 0 - x1
        new_x1 = 0
    else:
        pad_x1 = 0
        new_x1 = x1
    if y1 < 0:
        pad_y1 = 0 - y1
        new_y1 = 0
    else:
        pad_y1 = 0
        new_y1 = y1
    if x2 > img_w - 1:
        pad_x2 = x2 - (img_w - 1)
        new_x2 = img_w - 1
    else:
        pad_x2 = 0
        new_x2 = x2
    if y2 > img_h - 1:
        pad_y2 = y2 - (img_h - 1)
        new_y2 = img_h - 1
    else:
        pad_y2 = 0
        new_y2 = y2

    patch = image[new_y1:new_y2, new_x1:new_x2]
    patch = (
        np.pad(
            patch,
            ((pad_y1, pad_y2), (pad_x1, pad_x2)),
            mode="constant",
            constant_values=0,
        )
        if is_single_channel
        else np.pad(
            patch,
            ((pad_y1, pad_y2), (pad_x1, pad_x2), (0, 0)),
            mode="constant",
            constant_values=0,
        )
    )
    return patch, (new_y1, new_x1, new_y2, new_x2)


# Optional code to generate the segmented cell crops
def generate_crops(image_stack, cell_mask, nuclei_mask, crop_size, crop_bitdepth, crop_mask, mask_cell, output_folder, output_prefix):
    regions = regionprops(cell_mask)

    cell_bboxes = []
    for region in regions:
        image_cp = image_stack[0][0].copy()

        bbox = region.bbox
        bbox_center = [(bbox[0] + bbox[2]) // 2, (bbox[1] + bbox[3]) // 2]
        h, w = bbox[2] - bbox[0], bbox[3] - bbox[1]
        fixed_bbox = (
            bbox_center[0] - crop_size // 2,
            bbox_center[1] - crop_size // 2,
            bbox_center[0] + crop_size // 2,
            bbox_center[1] + crop_size // 2,
        )

        if crop_mask:
            this_cell_mask = cell_mask.copy()
            this_cell_mask[this_cell_mask != region.label] = 0
            this_cell_mask[this_cell_mask == region.label] = 1
            this_cell_mask = grey_dilation(this_cell_mask, size=7)
            curr_cell_mask, _ = safe_crop(this_cell_mask, fixed_bbox)
            cv2.imwrite(f"{output_folder}/{output_prefix}cell{region.label}_mask.png", np.uint8(curr_cell_mask * 255))
            if nuclei_mask is not None:
                this_nuclei_mask = nuclei_mask.copy()
                this_nuclei_mask[this_nuclei_mask != region.label] = 0
                this_nuclei_mask[this_nuclei_mask == region.label] = 1
                this_nuclei_mask = grey_dilation(this_nuclei_mask, size=7)
                curr_nuclei_mask, _ = safe_crop(this_nuclei_mask, fixed_bbox)
                cv2.imwrite(f"{output_folder}/{output_prefix}cell{region.label}_nucleimask.png", np.uint8(curr_nuclei_mask * 255))

        for curr_img_index in range(len(image_stack)):
            if curr_img_index != 0:
                image_cp = image_stack[curr_img_index][0].copy()

            cell_crop, _ = safe_crop(image_cp, fixed_bbox)
            cv2.imwrite(f"{output_folder}/{output_prefix}cell{region.label}_crop_" + colors[curr_img_index] + ".png", image_utils.convert_bitdepth(cell_crop, crop_bitdepth))

            if mask_cell:
                this_cell_mask = cell_mask == region.label
                this_cell_mask = grey_dilation(this_cell_mask, size=7)
                image_cp[this_cell_mask == 0] = 0
                cell_mask_crop, _ = safe_crop(image_cp, fixed_bbox)
                cv2.imwrite(f"{output_folder}/{output_prefix}cell{region.label}_crop_masked_" + colors[curr_img_index] + ".png", image_utils.convert_bitdepth(cell_mask_crop, crop_bitdepth))

        new_center = (crop_size // 2, crop_size // 2)
        new_bbox = (
            new_center[0] - h // 2,
            new_center[1] - w // 2,
            new_center[0] + h // 2,
            new_center[1] + w // 2,
        )
        cell_bboxes.append(
            {
                "id" : output_prefix,
                "cell": region.label,
                "y1": new_bbox[0],
                "x1": new_bbox[1],
                "y2": new_bbox[2],
                "x2": new_bbox[3],
            }
        )

    return pd.DataFrame(cell_bboxes)