def main2():
dstroot = 'clusters'
listcells = os.listdir(dstroot)
cellsinfo = []
for n, i in zip(listcells, tqdm(range(len(listcells)))):
cellinfo = {}
cellpath = os.path.join(dstroot, n)
img = cv2.imread(cellpath)
img_gray = cv2.imread(cellpath, 0)
img_gray = bf.get_fit_img(img_gray)
# cv2.imwrite(cellpath+'abc0.png',img_gray)
value1_, value_2 = bf.get_2value(img_gray)
value_1, _ = bf.get_2value(img_gray, grien=3)
sign_, cell_nuclei_mask1, area1 = get_cell_nuclei_mask(
img_gray, img, value_1) #获取细胞核掩码、个数、面积、凸面积、周长、核形规则度、获取细胞>核深染程度
sign_, cell_nuclei_mask2, area2 = get_cell_nuclei_mask(
img_gray, img, value1_) #获取细胞核掩码、个数、面积、凸面积、周长、核形规则度、获取细胞核深染
if sign_ == 0:
continue
# cv2.imwrite(cellpath+'abc1.png',cell_nuclei_mask2)
temp = area1 / area2
temp1 = str(temp)[0:5]
if area1 / area2 > 0.85:
newpath = os.path.join('ab_clusters', n + '_' + temp1 + '.png')
shutil.copy(cellpath, newpath)
def main2():
dstroot = 'fovs'
list_dst = os.listdir(dstroot)
for n, i in zip(list_dst, tqdm(range(len(list_dst)))):
imgpath = os.path.join(dstroot, n)
# print(imgpath)
img_org = cv2.imread(imgpath)
img_gray = cv2.imread(imgpath, 0)
img_gray_fit = bf.get_fit_img(img_gray)
value_1, value_2 = bf.get_2value(img_gray_fit) # 获取细胞核、背景阈值
kernel_1 = np.ones((50, 50), np.uint8)
kernel_2 = np.ones((5, 5), np.uint8)
cells_xoy = get_cells(img_gray_fit,
imgpath,
value_2 + 10,
kernel_1,
50000,
1600,
side=5)
value_1, value_2 = bf.get_2value(img_gray_fit, grien=3)
clusters_xoy = get_clusters(img_gray_fit,
value_1,
kernel_2,
20000,
1000000,
side=5)
crop_from_fov(img_org, n, cells_xoy)
crop_from_fov(img_org, n, clusters_xoy, root_='clusters')
def main2():
dstroot = 'crop'
listcells = os.listdir(dstroot)
cellsinfo = []
for n, i in zip(listcells, tqdm(range(len(listcells)))):
cellinfo = {}
cellpath = os.path.join(dstroot, n)
img = cv2.imread(cellpath)
img_gray = cv2.imread(cellpath, 0)
img_gray = bf.get_fit_img(img_gray)
# cv2.imwrite(cellpath+'abc0.png',img_gray)
value_1, value_2 = bf.get_2value(img_gray)
sign_nuclei, cell_nuclei_mask, nuclei_cnt, nuclei_area, nuclei_hull_area, nuclei_circ, nuclei_rule, cell_nuclei_value = get_cell_nuclei_mask(
img_gray, img, value_1) #获取细胞核掩码、个数、面积、凸面积、周长、核形规则度、获取细胞核深染程度
# cv2.imwrite(cellpath+'abc1.png',cell_nuclei_mask)
if sign_nuclei == 0:
continue
sign_cytoplasm, cell_cytoplasm_mask, cytoplasm_area, cytoplasm_hull_area, cytoplasm_circ, cytoplasm_rule, cell_cytoplasm_value, cytoplasm_var = get_cell_cytoplasm_mask(
img_gray, cell_nuclei_mask, nuclei_area, img,
value_2) #获取细胞质掩码、面积、凸面积、周长、细胞规则度、获取细胞质情况
# cv2.imwrite(cellpath+'abc2.png',cell_cytoplasm_mask)
if sign_cytoplasm == 0 or (cytoplasm_area - nuclei_area) == 0:
continue
cell_N_C = nuclei_area / (cytoplasm_area - nuclei_area) #计算核质比
cellinfo_keys = [
'cellpath', 'nuclei_cnt', 'nuclei_area', 'nuclei_hull_area',
'nuclei_circ', 'nuclei_rule', 'cell_nuclei_value',
'cytoplasm_area', 'cytoplasm_hull_area', 'cytoplasm_circ',
'cytoplasm_rule', 'cell_cytoplasm_value', 'cytoplasm_var',
'cell_N_C'
]
cellinfo_values = [
cellpath, nuclei_cnt, nuclei_area, nuclei_hull_area, nuclei_circ,
nuclei_rule, cell_nuclei_value, cytoplasm_area,
cytoplasm_hull_area, cytoplasm_circ, cytoplasm_rule,
cell_cytoplasm_value, cytoplasm_var, cell_N_C
]
cellinfo = dict(zip(cellinfo_keys, cellinfo_values))
cell_save_sign = get_cell_save_sign(cellinfo)
if cell_save_sign == 0:
continue
cellsinfo.append(cellinfo)
newpath = os.path.join('valid', n)
shutil.copy(cellpath, newpath)
np.save("./cells_info/cells_info.npy", cellsinfo)
1000000,
side=5)
crop_from_fov(img_org, n, cells_xoy)
crop_from_fov(img_org, n, clusters_xoy, root_='clusters')
if __name__ == "__main__":
dstroot = 'fovs'
list_dst = os.listdir(dstroot)
for n, i in zip(list_dst, tqdm(range(len(list_dst)))):
imgpath = os.path.join(dstroot, n)
# print(imgpath)
img_org = cv2.imread(imgpath)
img_gray = cv2.imread(imgpath, 0)
img_gray_fit = bf.get_fit_img(img_gray)
value_1, value_2 = bf.get_2value(img_gray_fit) # 获取细胞核、背景阈值
kernel_1 = np.ones((50, 50), np.uint8)
kernel_2 = np.ones((5, 5), np.uint8)
cells_xoy = get_cells(img_gray_fit,
imgpath,
value_2 + 10,
kernel_1,
50000,
1600,
side=5)
value_1, value_2 = bf.get_2value(img_gray_fit, grien=3)
clusters_xoy = get_clusters(img_gray_fit,
value_1,
kernel_2,
20000,
side=5)
if area1 / area2 > 0.85:
newpath = os.path.join('ab_clusters', n + '_' + temp1 + '.png')
shutil.copy(cellpath, newpath)
if __name__ == "__main__":
dstroot = 'clusters'
listcells = os.listdir(dstroot)
cellsinfo = []
for n, i in zip(listcells, tqdm(range(len(listcells)))):
cellinfo = {}
cellpath = os.path.join(dstroot, n)
img = cv2.imread(cellpath)
img_gray = cv2.imread(cellpath, 0)
img_gray = bf.get_fit_img(img_gray)
# cv2.imwrite(cellpath+'abc0.png',img_gray)
value1_, value_2 = bf.get_2value(img_gray)
value_1, _ = bf.get_2value(img_gray, grien=3)
sign_, cell_nuclei_mask1, area1 = get_cell_nuclei_mask(
img_gray, img, value_1) #获取细胞核掩码、个数、面积、凸面积、周长、核形规则度、获取细胞>核深染程度
sign_, cell_nuclei_mask2, area2 = get_cell_nuclei_mask(
img_gray, img, value1_) #获取细胞核掩码、个数、面积、凸面积、周长、核形规则度、获取细胞核深染
if sign_ == 0:
continue
# cv2.imwrite(cellpath+'abc1.png',cell_nuclei_mask2)
temp = area1 / area2
temp1 = str(temp)[0:5]
if area1 / area2 > 0.85:
newpath = os.path.join('ab_clusters', n + '_' + temp1 + '.png')
shutil.copy(cellpath, newpath)
np.save("./cells_info/cells_info.npy", cellsinfo)
if __name__ == "__main__":
dstroot = 'crop'
listcells = os.listdir(dstroot)
cellsinfo = []
for n, i in zip(listcells, tqdm(range(len(listcells)))):
cellinfo = {}
cellpath = os.path.join(dstroot, n)
# print(cellpath)
img = cv2.imread(cellpath)
img_gray = cv2.imread(cellpath, 0)
img_gray = bf.get_fit_img(img_gray)
#cv2.imwrite(cellpath+'abc0.png',img_gray)
value_1, value_2 = bf.get_2value(img_gray)
sign_nuclei, cell_nuclei_mask, nuclei_cnt, nuclei_area, nuclei_hull_area, nuclei_circ, nuclei_rule, cell_nuclei_value = get_cell_nuclei_mask(
img_gray, img, value_1) #获取细胞核掩码、个数、面积、凸面积、周长、核形规则度、获取细胞核深染程度
#cv2.imwrite(cellpath+str(cell_nuclei_value)[0:5]+'abc1.png',cell_nuclei_mask)
if sign_nuclei == 0:
continue
sign_cytoplasm, cell_cytoplasm_mask, cytoplasm_area, cytoplasm_hull_area, cytoplasm_circ, cytoplasm_rule, cell_cytoplasm_value, cytoplasm_var = get_cell_cytoplasm_mask(
img_gray, cell_nuclei_mask, nuclei_area, img,
value_2) #获取细胞质掩码、面积、凸面积、周长、细胞规则度、获取细胞质情况
#cv2.imwrite(cellpath+'abc2.png',cell_cytoplasm_mask)
if sign_cytoplasm == 0 or (cytoplasm_area - nuclei_area) == 0:
continue
cell_N_C = nuclei_area / (cytoplasm_area - nuclei_area) #计算核质比
cellinfo_keys = [
'cellpath', 'nuclei_cnt', 'nuclei_area', 'nuclei_hull_area',
'nuclei_circ', 'nuclei_rule', 'cell_nuclei_value',
else:
return max(0, (A_max_std - A_std) / (A_max_std - A_min_std))
if __name__ == "__main__":
dstroot = 'crop'
# find_invalid_cells(dstroot,50)
list__ = os.listdir(dstroot)
for n in list__:
imgpath = os.path.join(dstroot, n)
img = cv2.imread(imgpath, 0)
value_dhash = get_mindHash(img)
img_fit = get_fit_img(img)
cnt = get_2value(img_fit)
res = get_entropy(img_fit)
_, value2 = bf.get_2value(img_fit)
ret_, thresh = cv2.threshold(img, value2, 255, cv2.THRESH_BINARY_INV)
matrix_01 = thresh / 255
matrix_01 = cv2.resize(matrix_01, (64, 64))
rand_value = get_stand_std(matrix_01)
kernel_1 = np.ones((5, 5), np.uint8)
thresh = bf.get_img_open(thresh, kernel_1)
# thresh = bf.get_img_close(thresh,kernel_1)
image_, contours, hierarchy_ = cv2.findContours(
thresh, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
rule = 100
if len(contours) == 0:
rule = 100
else:
area_ = 0
contours_max = []