Chris Oswald
commited on
Commit
•
5a56741
1
Parent(s):
48bacc5
added mask array
Browse files
SPIDER.py
CHANGED
@@ -168,9 +168,9 @@ class SPIDER(datasets.GeneratorBasedBuilder):
|
|
168 |
"patient_id": datasets.Value("string"),
|
169 |
"scan_type": datasets.Value("string"),
|
170 |
# "raw_image": datasets.Image(),
|
171 |
-
"image_array": datasets.Array3D(shape=image_size, dtype='
|
172 |
# "raw_mask": datasets.Image(),
|
173 |
-
"mask_array": datasets.Array3D(shape=image_size, dtype='
|
174 |
"metadata": {
|
175 |
"num_vertebrae": datasets.Value(dtype="string"), #TODO: more specific types
|
176 |
"num_discs": datasets.Value(dtype="string"),
|
@@ -289,10 +289,6 @@ class SPIDER(datasets.GeneratorBasedBuilder):
|
|
289 |
resize_shape: Tuple[int, int, int],
|
290 |
validate_share: float = 0.3,
|
291 |
test_share: float = 0.2,
|
292 |
-
raw_image: bool = True,
|
293 |
-
numeric_array: bool = True,
|
294 |
-
metadata: bool = True,
|
295 |
-
rad_gradings: bool = True,
|
296 |
random_seed: int = 9999,
|
297 |
) -> Tuple[str, Dict]:
|
298 |
"""
|
@@ -312,13 +308,6 @@ class SPIDER(datasets.GeneratorBasedBuilder):
|
|
312 |
test_share: float indicating share of data to use for testing;
|
313 |
must be in range (0.0, 1.0); note that training share is
|
314 |
calculated as (1 - validate_share - test_share)
|
315 |
-
raw_image: indicates whether to include .mha image file in example
|
316 |
-
numeric_array: indicates whether to include numpy numeric array of
|
317 |
-
image in example
|
318 |
-
metadata: indicates whether to include patient and scanner metadata
|
319 |
-
with image example
|
320 |
-
rad_gradings: indicates whether to include patient's radiological
|
321 |
-
gradings with image example
|
322 |
|
323 |
Yields
|
324 |
Tuple (unique patient-scan ID, dict of
|
@@ -483,7 +472,7 @@ class SPIDER(datasets.GeneratorBasedBuilder):
|
|
483 |
patient_id = scan_id.split('_')[0]
|
484 |
scan_type = '_'.join(scan_id.split('_')[1:])
|
485 |
|
486 |
-
# Load .mha file
|
487 |
image_path = os.path.join(paths_dict['images'], 'images', example)
|
488 |
image = sitk.ReadImage(image_path)
|
489 |
|
@@ -492,8 +481,15 @@ class SPIDER(datasets.GeneratorBasedBuilder):
|
|
492 |
sitk.GetArrayFromImage(image), resize_shape
|
493 |
)
|
494 |
|
495 |
-
#
|
|
|
|
|
496 |
|
|
|
|
|
|
|
|
|
|
|
497 |
# Extract overview data corresponding to image
|
498 |
image_overview = overview_dict[scan_id]
|
499 |
|
@@ -507,7 +503,7 @@ class SPIDER(datasets.GeneratorBasedBuilder):
|
|
507 |
'raw_image':None, #TODO
|
508 |
'raw_mask':None, #TODO
|
509 |
'image_array':image_array,
|
510 |
-
'mask_array':
|
511 |
'metadata':image_overview,
|
512 |
'rad_gradings':patient_grades_dict,
|
513 |
}
|
|
|
168 |
"patient_id": datasets.Value("string"),
|
169 |
"scan_type": datasets.Value("string"),
|
170 |
# "raw_image": datasets.Image(),
|
171 |
+
"image_array": datasets.Array3D(shape=image_size, dtype='float64'),
|
172 |
# "raw_mask": datasets.Image(),
|
173 |
+
"mask_array": datasets.Array3D(shape=image_size, dtype='float64'),
|
174 |
"metadata": {
|
175 |
"num_vertebrae": datasets.Value(dtype="string"), #TODO: more specific types
|
176 |
"num_discs": datasets.Value(dtype="string"),
|
|
|
289 |
resize_shape: Tuple[int, int, int],
|
290 |
validate_share: float = 0.3,
|
291 |
test_share: float = 0.2,
|
|
|
|
|
|
|
|
|
292 |
random_seed: int = 9999,
|
293 |
) -> Tuple[str, Dict]:
|
294 |
"""
|
|
|
308 |
test_share: float indicating share of data to use for testing;
|
309 |
must be in range (0.0, 1.0); note that training share is
|
310 |
calculated as (1 - validate_share - test_share)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
311 |
|
312 |
Yields
|
313 |
Tuple (unique patient-scan ID, dict of
|
|
|
472 |
patient_id = scan_id.split('_')[0]
|
473 |
scan_type = '_'.join(scan_id.split('_')[1:])
|
474 |
|
475 |
+
# Load .mha image file
|
476 |
image_path = os.path.join(paths_dict['images'], 'images', example)
|
477 |
image = sitk.ReadImage(image_path)
|
478 |
|
|
|
481 |
sitk.GetArrayFromImage(image), resize_shape
|
482 |
)
|
483 |
|
484 |
+
# Load .mha mask file
|
485 |
+
mask_path = os.path.join(paths_dict['masks'], 'masks', example)
|
486 |
+
mask = sitk.ReadImage(mask_path)
|
487 |
|
488 |
+
# Convert .mha mask to standardized numeric array
|
489 |
+
mask_array = standardize_3D_image(
|
490 |
+
sitk.GetArrayFromImage(mask), resize_shape
|
491 |
+
)
|
492 |
+
|
493 |
# Extract overview data corresponding to image
|
494 |
image_overview = overview_dict[scan_id]
|
495 |
|
|
|
503 |
'raw_image':None, #TODO
|
504 |
'raw_mask':None, #TODO
|
505 |
'image_array':image_array,
|
506 |
+
'mask_array':mask_array,
|
507 |
'metadata':image_overview,
|
508 |
'rad_gradings':patient_grades_dict,
|
509 |
}
|