File size: 10,804 Bytes
74ecfd7 | 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 | {
"cells": [
{
"cell_type": "code",
"execution_count": 6,
"id": "83d855d7",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Saved: train_effusion_findings.json\n",
"Input1: 551, Kept(matched): 85, Dropped: 466\n"
]
}
],
"source": [
"import json, re\n",
"from pathlib import Path\n",
"from collections import OrderedDict\n",
"\n",
"# ====== paths (modify) ======\n",
"jsonl_1 = \"/home/shuhan/blobdata_sd/CT-RATE/train_4k_effusion.json\"\n",
"jsonl_2 = \"/home/shuhan/blobdata_sd/CT-RATE/disease_mask_json/disease_train_single_prompt_checked_label.json\"\n",
"out_jsonl = \"train_effusion_findings.json\"\n",
"\n",
"# jsonl_1 = \"/home/shuhan/blobdata_sd/CT-RATE/valid_effusion.json\"\n",
"# jsonl_2 = \"/home/shuhan/blobdata_sd/CT-RATE/disease_mask_json/disease_valid_single_prompt_checked_label.json\"\n",
"# out_jsonl = \"valid_effusion_findings.json\"\n",
"\n",
"\n",
"def iter_jsonl(path):\n",
" with open(path, \"r\", encoding=\"utf-8\") as f:\n",
" for line in f:\n",
" line = line.strip()\n",
" if line:\n",
" yield json.loads(line)\n",
"\n",
"def volume_key(volume_path: str) -> str:\n",
" \"\"\"match rule: ignore trailing _1/_2 before extension\"\"\"\n",
" name = Path(volume_path).name\n",
" if name.endswith(\".nii.gz\"):\n",
" core = name[:-7]\n",
" elif name.endswith(\".mha\"):\n",
" core = name[:-4]\n",
" else:\n",
" core = Path(name).stem\n",
" core = re.sub(r\"_(?:1|2)$\", \"\", core)\n",
" return core\n",
"\n",
"# Build key -> merged disease_findings (dedup)\n",
"key2findings = OrderedDict()\n",
"for obj in iter_jsonl(jsonl_2):\n",
" k = volume_key(obj.get(\"volume_path\", \"\"))\n",
" df = (obj.get(\"disease_findings\") or \"\").strip()\n",
" if not k:\n",
" continue\n",
" key2findings.setdefault(k, [])\n",
" if df and df not in key2findings[k]:\n",
" key2findings[k].append(df)\n",
"\n",
"kept = 0\n",
"seen = 0\n",
"with open(out_jsonl, \"w\", encoding=\"utf-8\") as w:\n",
" for obj in iter_jsonl(jsonl_1):\n",
" seen += 1\n",
" k = volume_key(obj.get(\"volume_path\", \"\"))\n",
" if k not in key2findings or not key2findings[k]:\n",
" continue # only keep matched\n",
" obj[\"disease_findings\"] = \" | \".join(key2findings[k])\n",
" w.write(json.dumps(obj, ensure_ascii=False) + \"\\n\")\n",
" kept += 1\n",
"\n",
"print(f\"Saved: {out_jsonl}\")\n",
"print(f\"Input1: {seen}, Kept(matched): {kept}, Dropped: {seen - kept}\")\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 7,
"id": "008a1daa",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"==== Effusion compare (ignore _1/_2) ====\n",
"Common volumes (exist in both): 2566\n",
"1) JSONL has effusion AND CSV has effusion: 153\n",
"2) JSONL has effusion BUT CSV no effusion: 23\n",
"3) JSONL no effusion BUT CSV has effusion:108\n",
"4) JSONL no effusion AND CSV no effusion: 2282\n",
"\n",
"[Coverage]\n",
"Only in JSONL: 0\n",
"Only in CSV: 22427\n"
]
}
],
"source": [
"import json, re\n",
"from pathlib import Path\n",
"import pandas as pd\n",
"\n",
"# ====== paths (modify) ======\n",
"jsonl_2_path = \"/home/shuhan/blobdata_sd/CT-RATE/disease_mask_json/disease_train_single_prompt_checked_label.json\"\n",
"csv_path = \"/home/shuhan/blobdata_sd/CT-RATE/multi_abnormality_labels/train_predicted_labels.csv\"\n",
"\n",
"def volume_key(p: str) -> str:\n",
" \"\"\"Ignore trailing _1/_2 before extension (nii.gz/mha).\"\"\"\n",
" name = Path(str(p).strip().strip('\"').strip(\"'\")).name\n",
" if name.endswith(\".nii.gz\"):\n",
" core = name[:-7]\n",
" elif name.endswith(\".mha\"):\n",
" core = name[:-4]\n",
" else:\n",
" core = Path(name).stem\n",
" core = re.sub(r\"_(?:1|2)$\", \"\", core) # drop trailing _1/_2\n",
" return core\n",
"\n",
"def json_has_effusion(obj: dict) -> bool:\n",
" s = \" \".join([\n",
" str(obj.get(\"disease_label\", \"\") or \"\"),\n",
" str(obj.get(\"disease_label_text\", \"\") or \"\"),\n",
" str(obj.get(\"disease_findings\", \"\") or \"\"),\n",
" ]).lower()\n",
" return \"effusion\" in s\n",
"\n",
"# -----------------------------\n",
"# 1) JSONL: key -> effusion?\n",
"# -----------------------------\n",
"eff_json = {}\n",
"with open(jsonl_2_path, \"r\", encoding=\"utf-8\") as f:\n",
" for line in f:\n",
" line = line.strip()\n",
" if not line:\n",
" continue\n",
" obj = json.loads(line)\n",
" k = volume_key(obj.get(\"volume_path\", \"\"))\n",
" if not k:\n",
" continue\n",
" eff_json[k] = eff_json.get(k, False) or json_has_effusion(obj)\n",
"\n",
"# -----------------------------\n",
"# 2) CSV: key -> effusion?\n",
"# Any column containing \"effusion\" == 1 -> effusion\n",
"# -----------------------------\n",
"df = pd.read_csv(csv_path)\n",
"if \"VolumeName\" not in df.columns:\n",
" raise ValueError(\"CSV must contain column: VolumeName\")\n",
"\n",
"eff_cols = [c for c in df.columns if \"effusion\" in c.lower()]\n",
"if not eff_cols:\n",
" raise ValueError(\"No effusion-related columns found in CSV (col name contains 'effusion').\")\n",
"\n",
"eff_csv = {}\n",
"for _, row in df.iterrows():\n",
" k = volume_key(row[\"VolumeName\"])\n",
" vals = []\n",
" for c in eff_cols:\n",
" v = row.get(c, 0)\n",
" try:\n",
" vals.append(int(v))\n",
" except:\n",
" vals.append(0)\n",
" is_eff = (max(vals) == 1)\n",
" eff_csv[k] = eff_csv.get(k, False) or is_eff # aggregate _1/_2\n",
"\n",
"# -----------------------------\n",
"# 3) Compare on common keys\n",
"# -----------------------------\n",
"common = set(eff_json) & set(eff_csv)\n",
"\n",
"both_yes = sum(1 for k in common if eff_json[k] and eff_csv[k])\n",
"json_yes_csv_no = sum(1 for k in common if eff_json[k] and (not eff_csv[k]))\n",
"json_no_csv_yes = sum(1 for k in common if (not eff_json[k]) and eff_csv[k])\n",
"both_no = sum(1 for k in common if (not eff_json[k]) and (not eff_csv[k]))\n",
"\n",
"print(\"==== Effusion compare (ignore _1/_2) ====\")\n",
"print(f\"Common volumes (exist in both): {len(common)}\")\n",
"print(f\"1) JSONL has effusion AND CSV has effusion: {both_yes}\")\n",
"print(f\"2) JSONL has effusion BUT CSV no effusion: {json_yes_csv_no}\")\n",
"print(f\"3) JSONL no effusion BUT CSV has effusion:{json_no_csv_yes}\")\n",
"print(f\"4) JSONL no effusion AND CSV no effusion: {both_no}\")\n",
"\n",
"# (optional) if you also want to know unmatched coverage:\n",
"print(\"\\n[Coverage]\")\n",
"print(f\"Only in JSONL: {len(set(eff_json) - set(eff_csv))}\")\n",
"print(f\"Only in CSV: {len(set(eff_csv) - set(eff_json))}\")\n"
]
},
{
"cell_type": "code",
"execution_count": 10,
"id": "35ba659d",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"total=370, effusion_records=72\n",
"effusion label types:\n",
"- Atelectasis\n"
]
}
],
"source": [
"import json\n",
"from pathlib import Path\n",
"\n",
"in_jsonl = \"/home/shuhan/blobdata_sd/CT-RATE/disease_mask_json/disease_valid_single_prompt_checked_label.json\"\n",
"out_jsonl = \"disease_valid_single_prompt_atelectasis, consolidation.json\" # 不想保存就改成 None\n",
"\n",
"def split_labels(s: str):\n",
" return [x.strip() for x in str(s or \"\").split(\",\") if x.strip()]\n",
"\n",
"def is_effusion_label(label: str) -> bool:\n",
" return \"atelectasis\" in label.lower()\n",
"\n",
"n_total = 0\n",
"n_hit = 0\n",
"effusion_types = set()\n",
"\n",
"out_f = open(out_jsonl, \"w\", encoding=\"utf-8\") if out_jsonl else None\n",
"\n",
"with open(in_jsonl, \"r\", encoding=\"utf-8\") as f:\n",
" for line in f:\n",
" line = line.strip()\n",
" if not line:\n",
" continue\n",
" n_total += 1\n",
" rec = json.loads(line)\n",
"\n",
" labels = split_labels(rec.get(\"disease_label\", \"\")) # 只从 disease_label 中提取\n",
" hit = [lb for lb in labels if is_effusion_label(lb)]\n",
"\n",
" if hit:\n",
" n_hit += 1\n",
" for h in hit:\n",
" effusion_types.add(h)\n",
" if out_f:\n",
" out_f.write(json.dumps(rec, ensure_ascii=False) + \"\\n\")\n",
"\n",
"if out_f:\n",
" out_f.close()\n",
"\n",
"print(f\"total={n_total}, effusion_records={n_hit}\")\n",
"print(\"effusion label types:\")\n",
"for x in sorted(effusion_types):\n",
" print(\"-\", x)\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "41db104f",
"metadata": {},
"outputs": [],
"source": [
"import json\n",
"\n",
"src_path = \"/home/shuhan/blobdata_sd/CT-RATE/disease_mask_json/disease_train_single_prompt_checked_label.json\"\n",
"target_label = \"Pleural effusion or thickening\"\n",
"\n",
"filtered = []\n",
"with open(src_path, \"r\", encoding=\"utf-8\") as f:\n",
" for line in f:\n",
" line = line.strip()\n",
" if not line:\n",
" continue\n",
" obj = json.loads(line)\n",
" if obj.get(\"disease_label\") == target_label:\n",
" filtered.append(obj)\n",
"\n",
"filtered[:3] # preview\n"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "genct",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.10.0"
}
},
"nbformat": 4,
"nbformat_minor": 5
}
|