zhengyushan / kat Goto Github PK

def extract_tile(image_dir, tile_size, x, y, width, height):
x_start_tile = x // tile_size
y_start_tile = y // tile_size
x_end_tile = (x + width) // tile_size
y_end_tile = (y + height) // tile_size

tmp_image = np.ones(
    ((y_end_tile - y_start_tile + 1) * tile_size, (x_end_tile - x_start_tile + 1) * tile_size, 3),
    np.uint8) * 240

for y_id, col in enumerate(range(x_start_tile, x_end_tile + 1)):
    for x_id, row in enumerate(range(y_start_tile, y_end_tile + 1)):
        img_path = os.path.join(image_dir, '{:04d}_{:04d}.jpg'.format(row, col))
        if not os.path.exists(img_path):
            continue
        img = cv2.imread(img_path)
        h, w, _ = img.shape
        tmp_image[(x_id * tile_size):(x_id * tile_size + h), (y_id * tile_size):(y_id * tile_size + w), :] = img

x_off = x % tile_size
y_off = y % tile_size
output = tmp_image[y_off:y_off + height, x_off:x_off + width]

return output

what's the meaning of 'pos[1] * step' in the below function? why you regard it as x and pass it to function extract_file?
def extract_and_save_tiles(image_dir, slide_save_dir, position_list, tile_size,
imsize, step, invert_rgb=False):
for pos in position_list:
img = extract_tile(image_dir, tile_size, pos[1] * step, pos[0] * step,
imsize, imsize)

    if len(img) > 0:
        if invert_rgb:
            img = cv2.cvtColor(img, cv2.COLOR_RGB2BGR)
        cv2.imwrite(
            os.path.join(slide_save_dir, '{:04d}_{:04d}.jpg'.format(pos[1], pos[0])), img)

When I run 'cnn_sample', the majority of the 500 patches extracted from each Whole Slide Image (WSI) are blank. Is there an issue with this?

Hello,
very impressed with your work.

But I have a few small questions about Table1 and Table2, specifically about the model FLOP, GPU memory cost, and speed.

As far as I know, the size of each slide is different, which causes the efficiency of each image calculation to be different.

So my question is, what size slice are you computing on, and how many patches are there?

Looking forward to your reply.

Yours.
pzSuen.

How can I get the structure of the whole slide image dataset as you prove. Now i have the part of datasets of TCGA, and they are in the format of SVS. How i can get this photo of each patches. Thanks