Datasets:
File size: 15,279 Bytes
5919b4a | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 | """
Image normalization utilities for MedVision visualization.
Copied and adapted from:
- medvision_bm/utils/configs.py (label_map_regroup, CT_HU_windows_WL,
TASK_LIST_FORCE_STANDARD_IMAGE_NORMALIZATION)
- medvision_bm/sft/sft_utils.py (normalize_ct_img, normalize_general_img,
normalize_img)
"""
import numpy as np
# ---------------------------------------------------------------------------
# Tasks that must use standard (percentile) normalization even for CT images.
# Mainly designed to skip HU-based normalization for contrast CT scans (e.g. KiPA22).
# Source: medvision_bm/utils/configs.py — TASK_LIST_FORCE_STANDARD_IMAGE_NORMALIZATION
# ---------------------------------------------------------------------------
TASK_LIST_FORCE_STANDARD_IMAGE_NORMALIZATION = [
{"dataset_name": "KiPA22", "taskID": "01", "taskType": "Tumor-Lesion-Size"},
{"dataset_name": "KiPA22", "taskID": "01", "taskType": "Box-Size"},
{"dataset_name": "KiPA22", "taskID": "01", "taskType": "Mask-Size"},
]
# ---------------------------------------------------------------------------
# HU window presets (width, level).
# Source: medvision_bm/utils/configs.py — HU_window_WL_map
# ---------------------------------------------------------------------------
_HU_WINDOW_WL_MAP = {
"soft_tissue": (400, 40),
"lung": (1500, -600),
"brain": (80, 40),
"bone": (1800, 400),
}
# Per-anatomy-group HU window assignments.
# Source: medvision_bm/utils/configs.py — CT_HU_windows_WL
CT_HU_WINDOWS_WL = {
"Artery": _HU_WINDOW_WL_MAP["soft_tissue"],
"Vein": _HU_WINDOW_WL_MAP["soft_tissue"],
"Brain": _HU_WINDOW_WL_MAP["brain"],
"Brain Tumor/Lesion": _HU_WINDOW_WL_MAP["brain"],
"Heart": _HU_WINDOW_WL_MAP["soft_tissue"],
"Lung": _HU_WINDOW_WL_MAP["lung"],
"Lung Tumor/Lesion": _HU_WINDOW_WL_MAP["lung"],
"Liver": _HU_WINDOW_WL_MAP["soft_tissue"],
"Liver Tumor/Lesion": _HU_WINDOW_WL_MAP["soft_tissue"],
"Kidney": _HU_WINDOW_WL_MAP["soft_tissue"],
"Kidney Tumor/Lesion": _HU_WINDOW_WL_MAP["soft_tissue"],
"Pancreas": _HU_WINDOW_WL_MAP["soft_tissue"],
"Pancreas Tumor/Lesion": _HU_WINDOW_WL_MAP["soft_tissue"],
"Gallbladder": _HU_WINDOW_WL_MAP["soft_tissue"],
"Spleen": _HU_WINDOW_WL_MAP["soft_tissue"],
"Adrenal Gland": _HU_WINDOW_WL_MAP["soft_tissue"],
"Colon": _HU_WINDOW_WL_MAP["soft_tissue"],
"Colon Tumor/Lesion": _HU_WINDOW_WL_MAP["soft_tissue"],
"Intestine": _HU_WINDOW_WL_MAP["soft_tissue"],
"Esophagus": _HU_WINDOW_WL_MAP["soft_tissue"],
"Stomach": _HU_WINDOW_WL_MAP["soft_tissue"],
"Urinary System": _HU_WINDOW_WL_MAP["soft_tissue"],
"Uterus": _HU_WINDOW_WL_MAP["soft_tissue"],
"Prostate": _HU_WINDOW_WL_MAP["soft_tissue"],
"Head-Neck": _HU_WINDOW_WL_MAP["soft_tissue"],
"Head-Neck Tumor/Lesion": _HU_WINDOW_WL_MAP["soft_tissue"],
"Hip": _HU_WINDOW_WL_MAP["bone"],
"Rib": _HU_WINDOW_WL_MAP["bone"],
"Spine": _HU_WINDOW_WL_MAP["bone"],
"Knee Bone": _HU_WINDOW_WL_MAP["bone"],
"Knee Soft Tissue": _HU_WINDOW_WL_MAP["soft_tissue"],
"Metastatic Lymph Node": _HU_WINDOW_WL_MAP["soft_tissue"],
"Miscellaneous Tumor/Lesion":_HU_WINDOW_WL_MAP["soft_tissue"],
"Jawbone": _HU_WINDOW_WL_MAP["bone"],
"Tooth": _HU_WINDOW_WL_MAP["bone"],
}
# ---------------------------------------------------------------------------
# Maps fine-grained label names to anatomy groups used for HU-window lookup.
# Source: medvision_bm/utils/configs.py — label_map_regroup
# ---------------------------------------------------------------------------
LABEL_MAP_REGROUP = {
# vasculature
"aorta": "Artery",
"anterior communicating artery": "Artery",
"basilar artery": "Artery",
"left iliac artery": "Artery",
"left anterior cerebral artery": "Artery",
"left common carotid artery": "Artery",
"left internal carotid artery": "Artery",
"left middle cerebral artery": "Artery",
"left posterior cerebral artery": "Artery",
"left posterior communicating artery": "Artery",
"left subclavian artery": "Artery",
"renal artery": "Artery",
"right iliac artery": "Artery",
"right anterior cerebral artery": "Artery",
"right common carotid artery": "Artery",
"right internal carotid artery": "Artery",
"right middle cerebral artery": "Artery",
"right posterior cerebral artery": "Artery",
"right posterior communicating artery": "Artery",
"right subclavian artery": "Artery",
"third a2 segment": "Artery",
"brachiocephalic trunk": "Artery",
"superior vena cava": "Vein",
"inferior vena cava": "Vein",
"inferior vena cava (ivc)": "Vein",
"postcava": "Vein",
"postcava (inferior vena cava)": "Vein",
"portal and splenic veins": "Vein",
"portal vein and splenic vein": "Vein",
"left brachiocephalic vein": "Vein",
"right brachiocephalic vein": "Vein",
"left iliac vein": "Vein",
"right iliac vein": "Vein",
"renal vein": "Vein",
# brain
"brain": "Brain",
"skull": "Brain",
"anterior hippocampus": "Brain",
"posterior hippocampus": "Brain",
"deep grey matter": "Brain",
"grey matter": "Brain",
"white matter": "Brain",
"brainstem": "Brain",
"cerebellum": "Brain",
"ventricles": "Brain",
"external cerebrospinal fluid": "Brain",
"stroke infarct": "Brain Tumor/Lesion",
"peritumoral edema of brain": "Brain Tumor/Lesion",
"edema of brain": "Brain Tumor/Lesion",
"surrounding non-enhancing flair hyperintensity of brain": "Brain Tumor/Lesion",
"resection cavity of brain": "Brain Tumor/Lesion",
"enhancing brain tumor": "Brain Tumor/Lesion",
"enhancing brain tumor tissue": "Brain Tumor/Lesion",
"non-enhancing brain tumor": "Brain Tumor/Lesion",
"non-enhancing brain tumor core": "Brain Tumor/Lesion",
"gross tumor volume of brain": "Brain Tumor/Lesion",
"cystic component of brain": "Brain Tumor/Lesion",
# heart
"heart": "Heart",
"left atrium": "Heart",
"left atrium of heart": "Heart",
"left atrial appendage": "Heart",
"left ventricular cavity": "Heart",
"left ventricular myocardium": "Heart",
"left ventricle": "Heart",
"right ventricular cavity": "Heart",
"myocardium": "Heart",
# lungs
"left lung": "Lung",
"left lung lower lobe": "Lung",
"left lung upper lobe": "Lung",
"right lung": "Lung",
"right lung lower lobe": "Lung",
"right lung middle lobe": "Lung",
"right lung upper lobe": "Lung",
"lung cancer": "Lung Tumor/Lesion",
# liver
"liver": "Liver",
"liver vessel": "Liver",
"liver cancer": "Liver Tumor/Lesion",
"liver tumour": "Liver Tumor/Lesion",
"liver tumor": "Liver Tumor/Lesion",
# kidneys
"kidney": "Kidney",
"right kidney": "Kidney",
"left kidney": "Kidney",
"kidney cyst": "Kidney Tumor/Lesion",
"left kidney cyst": "Kidney Tumor/Lesion",
"right kidney cyst": "Kidney Tumor/Lesion",
"kidney tumor": "Kidney Tumor/Lesion",
# pancreas
"pancreas": "Pancreas",
"pancreas cancer": "Pancreas Tumor/Lesion",
# gallbladder
"gall bladder": "Gallbladder",
"gallbladder": "Gallbladder",
# spleen
"spleen": "Spleen",
# adrenal glands
"adrenal gland": "Adrenal Gland",
"left adrenal gland": "Adrenal Gland",
"left adrenal gland (lag)": "Adrenal Gland",
"right adrenal gland": "Adrenal Gland",
"right adrenal gland (rag)": "Adrenal Gland",
# colon
"colon": "Colon",
"colon cancer primaries": "Colon Tumor/Lesion",
# intestines
"rectum": "Intestine",
"duodenum": "Intestine",
"small bowel": "Intestine",
"esophagus": "Esophagus",
# stomach
"stomach": "Stomach",
# uro-gynae
"urinary bladder": "Urinary System",
"bladder": "Urinary System",
"uterus": "Uterus",
"prostate": "Prostate",
"peripheral zone of prostate": "Prostate",
"transition zone of prostate": "Prostate",
# head & neck
"cochlea": "Head-Neck",
"trachea": "Head-Neck",
"pharynx": "Head-Neck",
"thyroid gland": "Head-Neck",
"primary gross tumor volume (head & neck)": "Head-Neck Tumor/Lesion",
"vestibular schwannoma": "Head-Neck Tumor/Lesion",
# musculoskeletal — hip
"left hip": "Hip",
"right hip": "Hip",
"sacrum": "Hip",
"left gluteus maximus": "Hip",
"left gluteus medius": "Hip",
"left gluteus minimus": "Hip",
"right gluteus maximus": "Hip",
"right gluteus medius": "Hip",
"right gluteus minimus": "Hip",
"right iliopsoas": "Hip",
"left iliopsoas": "Hip",
# ribs
"left 1st rib": "Rib",
"left 2nd rib": "Rib",
"left 3rd rib": "Rib",
"right 1st rib": "Rib",
"right 2nd rib": "Rib",
"right 3rd rib": "Rib",
**{f"{side} {n}th rib": "Rib" for side in ("left", "right") for n in range(4, 13)},
"costal cartilages": "Rib",
# spine
**{
f"vertebra {lvl}": "Spine"
for lvl in (
"C1", "C2", "C3", "C4", "C5", "C6", "C7",
"T1", "T2", "T3", "T4", "T5", "T6", "T7", "T8", "T9", "T10", "T11", "T12",
"L1", "L2", "L3", "L4", "L5", "S1",
)
},
"vertebrae": "Spine",
"intervertebral discs": "Spine",
"spinal cord": "Spine",
# knee
"femur": "Knee Bone",
"tibia": "Knee Bone",
"left femur": "Knee Bone",
"right femur": "Knee Bone",
"femoral cartilage": "Knee Soft Tissue",
"lateral tibial cartilage": "Knee Soft Tissue",
"medial tibial cartilage": "Knee Soft Tissue",
"patellar cartilage": "Knee Soft Tissue",
"lateral meniscus": "Knee Soft Tissue",
"medial meniscus": "Knee Soft Tissue",
# lymphatics
"metastatic lymph node": "Metastatic Lymph Node",
# miscellaneous
"edema": "Miscellaneous Tumor/Lesion",
"tumor": "Miscellaneous Tumor/Lesion",
"cystic component": "Miscellaneous Tumor/Lesion",
# dentistry
"upper jawbone": "Jawbone",
"lower jawbone": "Jawbone",
"left inferior alveolar canal": "Jawbone",
"right inferior alveolar canal": "Jawbone",
**{t: "Tooth" for t in [
"upper left canine", "upper left central incisor", "upper left lateral incisor",
"upper left first premolar", "upper left second premolar",
"upper left first molar", "upper left second molar",
"upper left third molar (wisdom tooth)",
"upper right canine", "upper right central incisor", "upper right lateral incisor",
"upper right first premolar", "upper right second premolar",
"upper right first molar", "upper right second molar",
"upper right third molar (wisdom tooth)",
"lower left canine", "lower left central incisor", "lower left lateral incisor",
"lower left first premolar", "lower left second premolar",
"lower left first molar", "lower left second molar",
"lower left third molar (wisdom tooth)",
"lower right canine", "lower right central incisor", "lower right lateral incisor",
"lower right first premolar", "lower right second premolar",
"lower right first molar", "lower right second molar",
"lower right third molar (wisdom tooth)",
]},
# catch-alls
"na": "Others",
"implant": "Others",
"crown": "Others",
"bridge": "Others",
"left autochthon": "Others",
"right autochthon": "Others",
"sternum": "Others",
"humerus": "Others",
"left humerus": "Others",
"right humerus": "Others",
"left clavicle": "Others",
"right clavicle": "Others",
"left scapula": "Others",
"right scapula": "Others",
"prostate/uterus": "Others",
}
# ---------------------------------------------------------------------------
# Core normalization functions.
# Source: medvision_bm/sft/sft_utils.py
# ---------------------------------------------------------------------------
def normalize_ct_img(img, window_width, window_level):
"""Normalize CT Hounsfield Units to [0, 255] using a W/L window."""
v_min = window_level - (window_width / 2)
v_max = window_level + (window_width / 2)
img_normalized = np.clip(img, v_min, v_max)
img_normalized = ((img_normalized - v_min) / (v_max - v_min)) * 255.0
return img_normalized.astype(np.uint8)
def normalize_general_img(img):
"""Percentile (0.5–99.5) min-max normalization to [0, 255] for MRI, PET, etc."""
v_min = np.percentile(img, 0.5)
v_max = np.percentile(img, 99.5)
if v_max - v_min == 0:
return np.zeros_like(img, dtype=np.uint8)
img_normalized = np.clip(img, v_min, v_max)
img_normalized = ((img_normalized - v_min) / (v_max - v_min)) * 255.0
return img_normalized.astype(np.uint8)
def normalize_img(image_2d, image_modality, label_name, dataset_name, taskID, taskType):
"""Normalize a 2D image slice to [0, 255] for visualization.
Replicates the full normalization logic from medvision_bm/sft/sft_utils.py:
- CT with a known, non-"Others" label: HU window-based normalization.
- Everything else (MRI, PET, contrast CT, unknown label): percentile min-max.
Args:
image_2d: 2D numpy array of raw pixel/HU values.
image_modality: Modality string from task_info, e.g. "CT" or "MRI".
label_name: Human-readable label name from labels_map, or None.
dataset_name: Dataset name string (e.g. "BraTS24").
taskID: Task ID string (e.g. "01").
taskType: Task type string (e.g. "Tumor-Lesion-Size").
Returns:
uint8 numpy array normalized to [0, 255].
"""
# Check if this task forces standard normalization (skip HU windows)
is_standard_normalization_required = any(
t["dataset_name"] == dataset_name
and t["taskID"] == taskID
and t["taskType"] == taskType
for t in TASK_LIST_FORCE_STANDARD_IMAGE_NORMALIZATION
)
if image_modality.lower() == "ct":
group = LABEL_MAP_REGROUP.get(label_name) if label_name is not None else None
use_hu_window = (
group is not None
and not is_standard_normalization_required
and group.lower() != "others"
)
if use_hu_window:
hu_window_wl = CT_HU_WINDOWS_WL.get(group)
assert hu_window_wl is not None, (
f"No HU window defined for group '{group}' (label '{label_name}'). "
f"Check CT_HU_WINDOWS_WL in image_normalization.py."
)
return normalize_ct_img(image_2d, hu_window_wl[0], hu_window_wl[1])
else:
return normalize_general_img(image_2d)
else:
return normalize_general_img(image_2d)
|