def it_locates_the_biggest_bbox(self, tmpdir, slide_fixture, tissue_mask,
save_scaled_image, expectation):
slide = Slide(slide_fixture, os.path.join(tmpdir, "processed"))
if save_scaled_image:
slide.save_scaled_image(3)
expected_img = load_expectation(
os.path.join("bbox-location-images", expectation),
type_="png",
)
bbox_location_img = slide.locate_biggest_tissue_box(
tissue_mask=tissue_mask, scale_factor=3)
np.testing.assert_array_almost_equal(np.asarray(bbox_location_img),
expected_img)
def it_can_save_scaled_image(self, tmpdir, resampled_dims_):
tmp_path_ = tmpdir.mkdir("myslide")
image = PILIMG.RGBA_COLOR_500X500_155_249_240
image.save(os.path.join(tmp_path_, "mywsi.png"), "PNG")
slide_path = os.path.join(tmp_path_, "mywsi.png")
slide = Slide(slide_path, os.path.join(tmp_path_, "processed"))
resampled_dims_.return_value = (100, 200, 300, 400)
slide.save_scaled_image(32)
assert slide.scaled_image_path(32) == os.path.join(
tmp_path_, "processed", "mywsi-32x-100x200-300x400.png")
assert os.path.exists(
os.path.join(tmp_path_, slide.scaled_image_path(32)))
def it_locates_tiles_on_the_slide(self, request, fixture_slide,
expectation, tmpdir):
slide = Slide(fixture_slide, os.path.join(tmpdir, "processed"))
slide.save_scaled_image(10)
random_tiles_extractor = RandomTiler(tile_size=(512, 512),
n_tiles=2,
level=0,
seed=42,
check_tissue=False)
expected_img = load_expectation(
expectation,
type_="png",
)
tiles_location_img = random_tiles_extractor.locate_tiles(
slide, scale_factor=10)
# --- Expanding test report with actual and expected images ---
expand_tests_report(request,
actual=tiles_location_img,
expected=expected_img)
np.testing.assert_array_almost_equal(np.asarray(tiles_location_img),
expected_img)
# plt.imshow(pil_img); plt.axis('off');
pil_img.save(tmp_outpath / 'pil_img.png', format='png')
# Apply mask to image
pil_img_boxed = histolab.util.apply_mask_image(
pil_img, pdx_slide.biggest_tissue_box_mask)
pil_img_boxed.save(tmp_outpath / 'pil_img_boxed.png', format='png')
# Scale and save image
scale_factor = 16
np_img_scaled = pdx_slide.resampled_array(
scale_factor=scale_factor) # scale and return ndarray
print(type(np_img_scaled))
print(np_img_scaled.shape)
# out image file path: {name}-{scale_factor}x-{large_w}x{large_h}-{new_w}x{new_h}.{IMG_EXT}
pdx_slide.save_scaled_image(
scale_factor=scale_factor) # scale and save into file
# Show img from array
# plt.imshow(img_scaled); plt.axis('off');
fig, ax = plt.subplots(figsize=(5, 5))
ax.imshow(np_img_scaled)
ax.axis('off')
# PIL.Image.fromarray(np_img_scaled)
# -----------------------------------
# Metadata
# -----------------------------------
# Load meta
if (metapath / 'meta_from_wsi_images.csv').exists():
meta_df = pd.read_csv(metapath / 'meta_from_wsi_images.csv')
