Immune_pheno.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "cfdda519-ac61-461a-b33a-ec0c5debea1b",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "import shutil\n",
    "import gc\n",
    "\n",
    "# synthetic data\n",
    "from datetime import datetime\n",
    "from PIL import Image\n",
    "import os, os.path\n",
    "import torchvision.transforms.functional as TF\n",
    "import matplotlib.image as mpimg\n",
    "import matplotlib.pyplot as plt\n",
    "import torch\n",
    "import math\n",
    "import sys\n",
    "from skimage.transform import rescale, resize, downscale_local_mean\n",
    "import ssl\n",
    "ssl._create_default_https_context = ssl._create_unverified_context\n",
    "import numpy as np\n",
    "from sklearn import model_selection\n",
    "import torch.optim.lr_scheduler as Lrate\n",
    "import random\n",
    "from random import randrange\n",
    "import fnmatch\n",
    "import cv2\n",
    "import openslide"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9204a536-f02f-45e5-8ea7-62eed483e777",
   "metadata": {},
   "source": [
    "# Check the dimensions tile predictions for TIL, stroma and Tumor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "67b47a9f-e982-4389-88af-67ab621647ca",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1664369_MR16-1693 I4_LN_HE.svs 1664369_MR16-1693 I4_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/1664369_MR16-1693 I4_LN_HE_gray.png\n",
      "(191, 342) (191, 342)\n",
      "2305194_CR12-4869 K7_LN_HE.svs 2305194_CR12-4869 K7_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/2305194_CR12-4869 K7_LN_HE_gray.png\n",
      "(192, 237) (192, 237)\n",
      "2332894_CR10-3906 L6_LN_HE.svs 2332894_CR10-3906 L6_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/2332894_CR10-3906 L6_LN_HE_gray.png\n",
      "(178, 229) (178, 229)\n",
      "2339093_MR13-3004 G4_LN_HE.svs 2339093_MR13-3004 G4_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/2339093_MR13-3004 G4_LN_HE_gray.png\n",
      "(203, 308) (203, 308)\n",
      "2369433_CR09-3283 D11_LN_HE.svs 2369433_CR09-3283 D11_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/2369433_CR09-3283 D11_LN_HE_gray.png\n",
      "(186, 298) (186, 298)\n",
      "2747756_CR10-1140 E6_LN_HE.svs 2747756_CR10-1140 E6_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/2747756_CR10-1140 E6_LN_HE_gray.png\n",
      "(199, 327) (199, 327)\n",
      "2991376_CR09-1508 A18_LN_HE.svs 2991376_CR09-1508 A18_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/2991376_CR09-1508 A18_LN_HE_gray.png\n",
      "(183, 357) (183, 357)\n",
      "3024155_CR09-7755 H6_LN_HE.svs 3024155_CR09-7755 H6_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/3024155_CR09-7755 H6_LN_HE_gray.png\n",
      "(181, 304) (181, 304)\n",
      "3107974_MR13-6574 H8_LN_HE.svs 3107974_MR13-6574 H8_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/3107974_MR13-6574 H8_LN_HE_gray.png\n",
      "(152, 160) (152, 160)\n",
      "3432956_CR11-5321 A17_LN_HE.svs 3432956_CR11-5321 A17_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/3432956_CR11-5321 A17_LN_HE_gray.png\n",
      "(178, 294) (178, 294)\n",
      "3537256_MR11-1488 C3_LN_HE.svs 3537256_MR11-1488 C3_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/3537256_MR11-1488 C3_LN_HE_gray.png\n",
      "(179, 258) (179, 258)\n",
      "3824123_CR09-2241  A6_LN_HE.svs 3824123_CR09-2241  A6_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/3824123_CR09-2241  A6_LN_HE_gray.png\n",
      "(124, 300) (124, 300)\n",
      "3850843_CR10-9102 A3_LN_HE.svs 3850843_CR10-9102 A3_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/3850843_CR10-9102 A3_LN_HE_gray.png\n",
      "(205, 233) (205, 233)\n",
      "4109721_CR10-9713 A3_LN_HE.svs 4109721_CR10-9713 A3_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/4109721_CR10-9713 A3_LN_HE_gray.png\n",
      "(176, 275) (176, 275)\n",
      "4216047_CR08-1239 J5_LN_HE.svs 4216047_CR08-1239 J5_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/4216047_CR08-1239 J5_LN_HE_gray.png\n",
      "(176, 230) (176, 230)\n",
      "4216530_MR14-3865 C1_LN_HE.svs 4216530_MR14-3865 C1_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/4216530_MR14-3865 C1_LN_HE_gray.png\n",
      "(167, 197) (167, 197)\n",
      "4311337_CR12-7931 A13_LN_HE.svs 4311337_CR12-7931 A13_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/4311337_CR12-7931 A13_LN_HE_gray.png\n",
      "(199, 311) (199, 311)\n",
      "5072942_MR13-3219 G4_LN_HE.svs 5072942_MR13-3219 G4_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/5072942_MR13-3219 G4_LN_HE_gray.png\n",
      "(172, 262) (172, 262)\n",
      "5104186_MR12-2175 G8_LN_HE.svs 5104186_MR12-2175 G8_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/5104186_MR12-2175 G8_LN_HE_gray.png\n",
      "(195, 332) (195, 332)\n",
      "5279864_MR12-7571 M5_LN_HE.svs 5279864_MR12-7571 M5_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/5279864_MR12-7571 M5_LN_HE_gray.png\n",
      "(208, 349) (208, 349)\n",
      "5286554_CR12-5474 F18_LN_HE.svs 5286554_CR12-5474 F18_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/5286554_CR12-5474 F18_LN_HE_gray.png\n",
      "(174, 268) (174, 268)\n",
      "5333027_MR15-1375 A17_LN_HE.svs 5333027_MR15-1375 A17_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/5333027_MR15-1375 A17_LN_HE_gray.png\n",
      "(193, 309) (193, 309)\n",
      "5402709_MR14-2821 A8_LN_HE.svs 5402709_MR14-2821 A8_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/5402709_MR14-2821 A8_LN_HE_gray.png\n",
      "(167, 181) (167, 181)\n",
      "5428365_CR12-3292 A6_LN_HE.svs 5428365_CR12-3292 A6_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/5428365_CR12-3292 A6_LN_HE_gray.png\n",
      "(196, 264) (196, 264)\n",
      "5499473_MR13-6727 A17_LN_HE.svs 5499473_MR13-6727 A17_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/5499473_MR13-6727 A17_LN_HE_gray.png\n",
      "(192, 327) (192, 327)\n",
      "6037584_MR10-3389 M5_LN_HE.svs 6037584_MR10-3389 M5_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/6037584_MR10-3389 M5_LN_HE_gray.png\n",
      "(196, 289) (196, 289)\n",
      "6087786_MR14-5313 F16_LN_HE.svs 6087786_MR14-5313 F16_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/6087786_MR14-5313 F16_LN_HE_gray.png\n",
      "(192, 302) (192, 302)\n",
      "6413725_CR07-4496 I20_LN_HE.svs 6413725_CR07-4496 I20_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/6413725_CR07-4496 I20_LN_HE_gray.png\n",
      "(158, 258) (158, 258)\n",
      "6440421_CR07-8025 B6_LN_HE.svs 6440421_CR07-8025 B6_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/6440421_CR07-8025 B6_LN_HE_gray.png\n",
      "(166, 271) (166, 271)\n",
      "6448598_CR07-9333 A4_LN_HE.svs 6448598_CR07-9333 A4_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/6448598_CR07-9333 A4_LN_HE_gray.png\n",
      "(177, 276) (177, 276)\n",
      "6660500_CR10-2136 G8_LN_HE.svs 6660500_CR10-2136 G8_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/6660500_CR10-2136 G8_LN_HE_gray.png\n",
      "(174, 248) (174, 248)\n",
      "7003888_MR13-5946 A17_LN_HE.svs 7003888_MR13-5946 A17_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7003888_MR13-5946 A17_LN_HE_gray.png\n",
      "(173, 311) (173, 311)\n",
      "7007848_CR08-6646 A7_LN_HE.svs 7007848_CR08-6646 A7_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7007848_CR08-6646 A7_LN_HE_gray.png\n",
      "(183, 241) (183, 241)\n",
      "7013673_CR08-7577 K9_LN_HE.svs 7013673_CR08-7577 K9_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7013673_CR08-7577 K9_LN_HE_gray.png\n",
      "(177, 230) (177, 230)\n",
      "7041998_CR09-701 A25_LN_HE.svs 7041998_CR09-701 A25_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7041998_CR09-701 A25_LN_HE_gray.png\n",
      "(182, 205) (182, 205)\n",
      "7049339_CR09-1615 G15_LN_HE.svs 7049339_CR09-1615 G15_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7049339_CR09-1615 G15_LN_HE_gray.png\n",
      "(192, 332) (192, 332)\n",
      "7064074_CR09-3186 C15_LN_HE.svs 7064074_CR09-3186 C15_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7064074_CR09-3186 C15_LN_HE_gray.png\n",
      "(173, 271) (173, 271)\n",
      "7085615_CR09-6053 J12_LN_HE.svs 7085615_CR09-6053 J12_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7085615_CR09-6053 J12_LN_HE_gray.png\n",
      "(196, 338) (196, 338)\n",
      "7098750_CR09-7449 H3_LN_HE.svs 7098750_CR09-7449 H3_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7098750_CR09-7449 H3_LN_HE_gray.png\n",
      "(187, 327) (187, 327)\n",
      "7134164_MR10-1940 I3_LN_HE.svs 7134164_MR10-1940 I3_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7134164_MR10-1940 I3_LN_HE_gray.png\n",
      "(195, 287) (195, 287)\n",
      "7139421_CR10-2839 A2 _LN_HE.svs 7139421_CR10-2839 A2 _LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7139421_CR10-2839 A2 _LN_HE_gray.png\n",
      "(176, 306) (176, 306)\n",
      "7139472_CR10-2985 G1 _LN_HE.svs 7139472_CR10-2985 G1 _LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7139472_CR10-2985 G1 _LN_HE_gray.png\n",
      "(176, 280) (176, 280)\n",
      "7143632_CR10-3542 A15_LN_HE.svs 7143632_CR10-3542 A15_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7143632_CR10-3542 A15_LN_HE_gray.png\n",
      "(188, 265) (188, 265)\n",
      "7146579_CR10-3969 H25_LN_HE.svs 7146579_CR10-3969 H25_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7146579_CR10-3969 H25_LN_HE_gray.png\n",
      "(146, 201) (146, 201)\n",
      "7153027_CR10-4679 O14_LN_HE.svs 7153027_CR10-4679 O14_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7153027_CR10-4679 O14_LN_HE_gray.png\n",
      "(151, 241) (151, 241)\n",
      "7187362_CR10-9130 A7_LN_HE.svs 7187362_CR10-9130 A7_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7187362_CR10-9130 A7_LN_HE_gray.png\n",
      "(176, 193) (176, 193)\n",
      "7199977_CR11-1062 A7_LN_HE.svs 7199977_CR11-1062 A7_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7199977_CR11-1062 A7_LN_HE_gray.png\n",
      "(169, 273) (169, 273)\n",
      "7200749_CR11-349 F15_LN_HE.svs 7200749_CR11-349 F15_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7200749_CR11-349 F15_LN_HE_gray.png\n",
      "(186, 250) (186, 250)\n",
      "7206384_CR11-1511 A7_LN_HE.svs 7206384_CR11-1511 A7_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7206384_CR11-1511 A7_LN_HE_gray.png\n",
      "(156, 213) (156, 213)\n",
      "7232287_CR11-4766 A6_LN_HE.svs 7232287_CR11-4766 A6_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7232287_CR11-4766 A6_LN_HE_gray.png\n",
      "(208, 355) (208, 355)\n",
      "7244024_MR13-7015 A8_LN_HE.svs 7244024_MR13-7015 A8_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7244024_MR13-7015 A8_LN_HE_gray.png\n",
      "(195, 298) (195, 298)\n",
      "7253300_MR14-4083 H3_LN_HE.svs 7253300_MR14-4083 H3_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7253300_MR14-4083 H3_LN_HE_gray.png\n",
      "(181, 227) (181, 227)\n",
      "7311555_CR12-5077 G7_LN_HE.svs 7311555_CR12-5077 G7_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7311555_CR12-5077 G7_LN_HE_gray.png\n",
      "(206, 313) (206, 313)\n",
      "7330795_CR12-7369 C5_LN_HE.svs 7330795_CR12-7369 C5_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7330795_CR12-7369 C5_LN_HE_gray.png\n",
      "(168, 220) (168, 220)\n",
      "7335343_CR12-7911 I1_LN_HE.svs 7335343_CR12-7911 I1_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7335343_CR12-7911 I1_LN_HE_gray.png\n",
      "(208, 323) (208, 323)\n",
      "7358033_CR13-1109 F1_LN_HE.svs 7358033_CR13-1109 F1_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7358033_CR13-1109 F1_LN_HE_gray.png\n",
      "(202, 247) (202, 247)\n",
      "7362883_CR13-1985 K11_LN_HE.svs 7362883_CR13-1985 K11_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7362883_CR13-1985 K11_LN_HE_gray.png\n",
      "(190, 281) (190, 281)\n",
      "7415202_MR13-6944 I5_LN_HE.svs 7415202_MR13-6944 I5_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/7415202_MR13-6944 I5_LN_HE_gray.png\n",
      "(201, 306) (201, 306)\n",
      "8361486_MR16-773 A10_LN_HE.svs 8361486_MR16-773 A10_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/8361486_MR16-773 A10_LN_HE_gray.png\n",
      "(173, 293) (173, 293)\n",
      "8411283_CR12-7570 L10_LN_HE.svs 8411283_CR12-7570 L10_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/8411283_CR12-7570 L10_LN_HE_gray.png\n",
      "(195, 309) (195, 309)\n",
      "8731856_MR16-2354 H11_LN_HE.svs 8731856_MR16-2354 H11_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/8731856_MR16-2354 H11_LN_HE_gray.png\n",
      "(190, 309) (190, 309)\n",
      "8937583_MR15-4569 H6_LN_HE.svs 8937583_MR15-4569 H6_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/8937583_MR15-4569 H6_LN_HE_gray.png\n",
      "(172, 316) (172, 316)\n",
      "9022920_MR15-970 F16_LN_HE.svs 9022920_MR15-970 F16_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/9022920_MR15-970 F16_LN_HE_gray.png\n",
      "(195, 293) (195, 293)\n",
      "9073175_MR15-5728 V9_LN_HE.svs 9073175_MR15-5728 V9_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/9073175_MR15-5728 V9_LN_HE_gray.png\n",
      "(176, 218) (176, 218)\n",
      "9101243_MR16-53 H1 _LN_HE.svs 9101243_MR16-53 H1 _LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/9101243_MR16-53 H1 _LN_HE_gray.png\n",
      "(180, 271) (180, 271)\n",
      "9127506_MR16-2637 D15_LN_HE.svs 9127506_MR16-2637 D15_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/9127506_MR16-2637 D15_LN_HE_gray.png\n",
      "(180, 311) (180, 311)\n",
      "9135020_MR16-3554 I15_LN_HE.svs 9135020_MR16-3554 I15_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/9135020_MR16-3554 I15_LN_HE_gray.png\n",
      "(178, 299) (178, 299)\n",
      "9168841_MR16-6765 B7_LN_HE.svs 9168841_MR16-6765 B7_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/9168841_MR16-6765 B7_LN_HE_gray.png\n",
      "(193, 263) (193, 263)\n",
      "9244977_MR14-5843 G5_LN_HE.svs 9244977_MR14-5843 G5_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/9244977_MR14-5843 G5_LN_HE_gray.png\n",
      "(172, 281) (172, 281)\n",
      "9961502_MR16-5145 G2_LN_HE.svs 9961502_MR16-5145 G2_LN_HE_gray.png\n",
      "hxu_predictions-Jinhwan/pred_tils/9961502_MR16-5145 G2_LN_HE_gray.png\n",
      "(168, 235) (168, 235)\n"
     ]
    }
   ],
   "source": [
    "input_path=\"../shared_bucket/datasets/VanAbel_Oropharynx/\"\n",
    "h_path='hxu_predictions-Jinhwan/pred_tils/'\n",
    "tils_hm_files = fnmatch.filter(os.listdir(input_path+'Van Abel-K_req32469_OPX TILs HE/'), '*LN*.svs')\n",
    "tis_img_files=fnmatch.filter(os.listdir(h_path), '*LN*'+'*gray*.png')\n",
    "coord_path='CSV_coordinates_heatmap_LN/'\n",
    "tis_img_files.sort()\n",
    "#print(tis_img_files)\n",
    "tils_hm_files.sort()\n",
    "i=0\n",
    "Tile_size=[112,112]\n",
    "for (tils_file,img_file) in zip(tils_hm_files,tis_img_files) :\n",
    "    print(tils_file,img_file)\n",
    "    i+=1\n",
    "    HE_file=input_path+'Van Abel-K_req32469_OPX TILs HE/'+tils_file\n",
    "    h_file=h_path+img_file\n",
    "    print(h_file)\n",
    "    Slide = openslide.OpenSlide(HE_file)\n",
    "    xr = float(Slide.properties['openslide.mpp-x'])  # pixel resolution at x direction\n",
    "    yr = float(Slide.properties['openslide.mpp-y'])  # pixel resolution at y direction\n",
    "    Dims = Slide.level_dimensions\n",
    "    Stride = [round(Tile_size[0] / xr), round(Tile_size[1] / yr)]\n",
    "    #print(Stride)\n",
    "    X = np.arange(0, Dims[0][0] + 1, Stride[0])\n",
    "    Y = np.arange(0, Dims[0][1] + 1, Stride[1])\n",
    "    #print(X.shape,Y.shape)\n",
    "    X, Y = np.meshgrid(X, Y)\n",
    "    img=plt.imread(h_file)\n",
    "    print(X.shape,Y.shape)\n",
    "    with open(coord_path+img_file.split('_')[0]+'_coord_files.csv', 'w') as outfile :\n",
    "        outfile.write('index_x,index_y,WSI_x, WSI_y, til_score\\n')\n",
    "        for i in range(X.shape[0] - 1):\n",
    "            for j in range(X.shape[1] - 1):\n",
    "                score=img[i,j]\n",
    "                \n",
    "                outfile.write(str(i)\n",
    "                      + ',' + str(j)\n",
    "                      + ',' + str(X[i,j])+','+str(Y[i,j]) + ','+str(score[0])+'\\n')\n",
    "    \n",
    "\n",
    "        "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2865498d-18cf-44e5-9ebe-c617c3559b04",
   "metadata": {},
   "source": [
    "# Following function computed density of TILs in various regions of the tissue in grids on the predicted tiles"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 375,
   "id": "209fb592-9ad6-4712-a286-04d8cfeac287",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.7/site-packages/ipykernel_launcher.py:77: RuntimeWarning: invalid value encountered in long_scalars\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1664369 58 8 99 33\n",
      "2305194 30 21 77 26\n",
      "2332894 18 17 43 8\n",
      "2339093 19 15 42 8\n",
      "2369433 78 0 78 0\n",
      "2466516 64 29 104 11\n",
      "2747756 4 9 21 8\n",
      "2991376 38 3 55 14\n",
      "3024155 40 5 47 2\n",
      "3107974 51 23 122 48\n",
      "3268469 50 17 67 0\n",
      "3392106 14 18 43 11\n",
      "3432956 7 18 50 25\n",
      "3537256 5 0 5 0\n",
      "3824123 36 4 45 5\n",
      "3850843 29 14 102 59\n",
      "4109721 4 27 90 59\n",
      "4160780 18 5 44 21\n",
      "4216047 46 8 60 6\n",
      "4216530 15 6 37 16\n",
      "4311337 63 16 90 11\n",
      "5060209 18 14 41 9\n",
      "5072942 27 15 46 4\n",
      "5104186 54 21 92 17\n",
      "5279864 31 17 74 26\n",
      "5286554 21 19 51 11\n",
      "5333027 7 12 61 42\n",
      "5402709 7 15 55 33\n",
      "5428365 32 13 50 5\n",
      "5429415 38 21 77 18\n",
      "5499473 46 13 64 5\n",
      "6037584 23 24 87 40\n",
      "6087786 53 14 93 26\n",
      "6413725 5 9 15 1\n",
      "6440421 81 19 127 27\n",
      "6448598 24 6 41 11\n",
      "6660500 3 2 26 21\n",
      "7003888 11 24 56 21\n",
      "7007848 44 9 68 15\n",
      "7013673 2 3 46 41\n",
      "7041998 29 8 44 7\n",
      "7049339 5 0 17 12\n",
      "7064074 3 3 9 3\n",
      "7085615 20 32 65 13\n",
      "7098750 33 11 63 19\n",
      "7134164 3 27 59 29\n",
      "7139421 8 21 35 6\n",
      "7139472 78 7 91 6\n",
      "7143632 24 38 84 22\n",
      "7146579 20 25 70 25\n",
      "7153025 7 19 50 24\n",
      "7153027 36 24 100 40\n",
      "7187362 44 27 94 23\n",
      "7192748 67 3 111 41\n",
      "7199977 12 20 97 65\n",
      "7200749 42 13 67 12\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.7/site-packages/ipykernel_launcher.py:79: RuntimeWarning: invalid value encountered in long_scalars\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "7206384 25 33 75 17\n",
      "7232287 11 10 43 22\n",
      "7244024 22 20 51 9\n",
      "7253300 16 37 75 22\n",
      "7261355 0 5 36 31\n",
      "7279442 26 20 76 30\n",
      "7305656 15 2 30 13\n",
      "7311555 55 11 88 22\n",
      "7330795 24 13 49 12\n",
      "7335343 28 22 77 27\n",
      "7358033 64 29 96 3\n",
      "7362883 11 8 25 6\n",
      "7368286 19 35 77 23\n",
      "7383556 25 13 48 10\n",
      "7414302 35 4 42 3\n",
      "7415202 46 6 70 18\n",
      "7423885 9 25 110 76\n",
      "8361486 39 30 93 24\n",
      "8406491 9 16 61 36\n",
      "8411283 51 3 56 2\n",
      "8731856 10 22 62 30\n",
      "8937583 44 5 57 8\n",
      "9022920 24 12 72 36\n",
      "9038929 44 23 79 12\n",
      "9063099 0 6 70 64\n",
      "9073175 24 17 69 28\n",
      "9101243 14 24 80 42\n",
      "9127506 13 3 52 36\n",
      "9135020 27 5 63 31\n",
      "9168841 19 22 78 37\n",
      "9244977 10 10 26 6\n",
      "9961502 13 15 41 13\n"
     ]
    }
   ],
   "source": [
    "def tils_density(input_path):\n",
    "    tils_hm_files = fnmatch.filter(os.listdir(input_path+'pred_tils/'), '*Tumor*_gray.png')\n",
    "    tils_hm_files.sort()\n",
    "    \n",
    "    M=17; N=17\n",
    "    with open('result_Tumor_tils_density_test.csv', 'a') as outfile :\n",
    "        outfile.write('file_name,immune_infl,immune_excl,immune_des\\n')\n",
    "        for tils_file in tils_hm_files :          \n",
    "\n",
    "            tissue_mask_file = tils_file.split('_gray')[0] + '.png'\n",
    "            tumor_file = tils_file\n",
    "            til = input_path + '/pred_tils/' + tils_file\n",
    "            tissue_mask = input_path + '/tissue_mask/'+ tissue_mask_file\n",
    "            tumor = input_path + '/pred_tumor/' + tumor_file\n",
    "            stroma=input_path + '/pred_stromal/' + tils_file\n",
    "            til_img = plt.imread(til)\n",
    "            #plt.imshow(til_img)\n",
    "            #plt.show()\n",
    "            #print('til:',til_img.shape)\n",
    "            R, G, B = til_img[:,:,0], til_img[:,:,1], til_img[:,:,2]\n",
    "            til_img_gray=np.mean([R,G,B], axis=0)\n",
    "            til_img_gray[til_img_gray >= 0.5] = 1\n",
    "            til_img_gray[til_img_gray < 0.5] = 0\n",
    "            tumor_img = plt.imread(tumor)\n",
    "            #print('Tumor:',tumor_img.shape)\n",
    "            #plt.imshow(tumor_img)\n",
    "            #plt.show()\n",
    "            R, G, B = tumor_img[:,:,0], tumor_img[:,:,1], tumor_img[:,:,2]\n",
    "            tumor_img_gray = np.mean([R,G,B], axis=0)\n",
    "            tumor_img_gray[tumor_img_gray >= 0.5] = 1\n",
    "            tumor_img_gray[tumor_img_gray < 0.5] = 0\n",
    "            tumor_img_gray = resize(tumor_img_gray, (til_img_gray.shape[0],til_img_gray.shape[1]))\n",
    "            tissue_mask_img = plt.imread(tissue_mask)\n",
    "            #print('tissue:',tissue_mask_img.shape)\n",
    "            #plt.imshow( tissue_mask_img)\n",
    "            #plt.show()\n",
    "            tissue_mask_img = resize(tissue_mask_img, (til_img.shape[0],til_img.shape[1], 3))\n",
    "            R, G, B = tissue_mask_img[:,:,0], tissue_mask_img[:,:,1], tissue_mask_img[:,:,2]\n",
    "            tissue_mask_img_gray = np.mean([R,G,B], axis=0)\n",
    "            tissue_mask_img_gray[tissue_mask_img_gray >= 0.5] = 1\n",
    "            tissue_mask_img_gray[tissue_mask_img_gray < 0.5] = 0\n",
    "            tiles = [tissue_mask_img_gray[x:x+M,y:y+N] for x in range(0,tissue_mask_img_gray.shape[0],M) for y in range(0,tissue_mask_img_gray.shape[1],N)]\n",
    "            #print(tiles)\n",
    "            stroma_img = plt.imread(stroma)\n",
    "            stroma_img = resize(stroma_img, (til_img_gray.shape[0],til_img_gray.shape[1]))\n",
    "            R, G, B = stroma_img[:,:,0], stroma_img[:,:,1], stroma_img[:,:,2]\n",
    "            stroma_img = np.mean([R,G,B], axis=0)\n",
    "            #print('stroma:',stroma_img.shape)\n",
    "            #plt.imshow(stroma_img)\n",
    "            #plt.show()\n",
    "\n",
    "            overlap_tt = til_img_gray + tumor_img_gray #overlap TIL and Tumor\n",
    "            overlap_ts=til_img_gray + stroma_img#overlap of til and stroma\n",
    "            overlap_tst=tissue_mask_img_gray + til_img_gray\n",
    "            c_overlap_tt = overlap_tt > 1 #binarize\n",
    "            c_overlap_ts=overlap_ts > 1\n",
    "            c_overlap_tst=overlap_tst > 1\n",
    "            imm_inf=0\n",
    "            imm_excl=0\n",
    "            imm_des=0\n",
    "            c_tissue=0\n",
    "            i=0\n",
    "            for x in range(0,tissue_mask_img_gray.shape[0],M):\n",
    "                for y in range(0,tissue_mask_img_gray.shape[1],N):\n",
    "                    #print(i)\n",
    "                    #print(tiles[i].sum())\n",
    "                    if (tiles[i].sum())>100:\n",
    "                        tissue_sum=tiles[i].sum()\n",
    "                        c_tissue+=1 #count the number of tissue tiles\n",
    "                        overlap_count_tt = sum(map(lambda x : x == True, c_overlap_tt[x:x+M,y:y+N]))#count the overlap ratio\n",
    "                        overlap_count_ts = sum(map(lambda x : x == True, c_overlap_ts[x:x+M,y:y+N]))\n",
    "                        overlap_count_tst = sum(map(lambda x : x == True, c_overlap_tst[x:x+M,y:y+N]))\n",
    "                        sum_overlap_tt = np.sum(overlap_count_tt)\n",
    "                        sum_overlap_ts = np.sum(overlap_count_ts)\n",
    "                        sum_overlap_tst = np.sum(overlap_count_tst)\n",
    "                        #print(sum_overlap_tt/np.count_nonzero(tumor_img_gray[x:x+M,y:y+N]),sum_overlap_ts/np.count_nonzero(stroma_img[x:x+M,y:y+N]),sum_overlap_tst)\n",
    "                        if np.isnan(sum_overlap_tt/np.count_nonzero(tumor_img_gray[x:x+M,y:y+N]))==False and sum_overlap_tt/np.count_nonzero(tumor_img_gray[x:x+M,y:y+N]) > 0.5:\n",
    "                            imm_inf+=1\n",
    "                        elif np.isnan(sum_overlap_ts/np.count_nonzero(stroma_img[x:x+M,y:y+N]))==False and sum_overlap_ts/np.count_nonzero(stroma_img[x:x+M,y:y+N]) >0.5:\n",
    "                            imm_excl+=1\n",
    "                        else:\n",
    "                            imm_des+=1\n",
    "                    i+=1\n",
    "            print(tissue_mask_file.split('_')[0],imm_inf,imm_excl,c_tissue,imm_des)\n",
    "            outfile.write(tissue_mask_file.split('_')[0]\n",
    "                              + ',' + str(imm_inf/c_tissue)\n",
    "                              + ',' + str(imm_excl/c_tissue)\n",
    "                              + ',' + str(imm_des/c_tissue) + '\\n')\n",
    "    outfile.close()\n",
    "\n",
    "input_path = 'hxu_predictions-Jinhwan/'\n",
    "tils_density(input_path)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4c2f8f68",
   "metadata": {},
   "source": [
    "# Patient level label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "daf9ade8",
   "metadata": {},
   "outputs": [],
   "source": [
    "df=pd.read_csv(\"path_to_immune_score_file_all_patinets\")\n",
    "df_ext=df_m[[\"MRN\",\"immune_infl\",\"immune_excl\",\"immune_des\"]]\n",
    "df = pd.DataFrame(\n",
    "   {\n",
    "      \"MRN\": [],\n",
    "      \"Immune_ph\": []\n",
    "   }\n",
    ")\n",
    "for i,row in enumerate(df_ext.iterrows()):\n",
    "    if row[1][\"immune_infl\"]>0.33:\n",
    "        df.loc[i]=[row[1][\"MRN\"], \"immune_inflamed\"]\n",
    "    elif row[1][\"immune_excl\"]>0.33:\n",
    "        df.loc[i]=[row[1][\"MRN\"], \"immune_excluded\"]\n",
    "    else:\n",
    "        df.loc[i]=[row[1][\"MRN\"], \"immune_desert\"]"
   ]
  }
 ],
 "metadata": {
  "environment": {
   "kernel": "python3",
   "name": "pytorch-gpu.1-10.m84",
   "type": "gcloud",
   "uri": "gcr.io/deeplearning-platform-release/pytorch-gpu.1-10:m84"
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}