analysis_visualize_results.py.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "import cv2\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.metrics import auc, roc_auc_score, roc_curve, precision_recall_curve\n",
    "from sklearn import metrics\n",
    "from mpl_toolkits.axes_grid1 import ImageGrid\n",
    "import numpy as np\n",
    "from skimage import io, transform\n",
    "import os\n",
    "import random\n",
    "import pandas as pd\n",
    "from tqdm.notebook import tqdm\n",
    "from datetime import datetime\n",
    "\n",
    "\n",
    "SMALL_SIZE = 22\n",
    "MEDIUM_SIZE = 24\n",
    "BIGGER_SIZE = 30\n",
    "\n",
    "plt.rc('font', size=SMALL_SIZE)          # controls default text sizes\n",
    "plt.rc('axes', titlesize=BIGGER_SIZE)     # fontsize of the axes title\n",
    "plt.rc('axes', labelsize=BIGGER_SIZE)    # fontsize of the x and y labels\n",
    "plt.rc('xtick', labelsize=MEDIUM_SIZE)    # fontsize of the tick labels\n",
    "plt.rc('ytick', labelsize=MEDIUM_SIZE)    # fontsize of the tick labels\n",
    "plt.rc('legend', fontsize=MEDIUM_SIZE)    # legend fontsize\n",
    "plt.rc('figure', titlesize=BIGGER_SIZE)  # fontsize of the figure title\n",
    "plt.rc('lines', linewidth=5)\n",
    "\n",
    "#WORKING_FILE = '/home/simon/Documents/20210531_799_ablation/fusion/fused.csv'\n",
    "WORKING_FILE = '/home/simon/Documents/20210719_299_ablation/fused.csv'\n",
    "OUTPUT_FOLDER = '/home/simon/Pictures/Attention Maps 799/'\n",
    "CUR_DATE = datetime.today().strftime(\"%Y%m%d\")\n",
    "SWITCH_CLASSES = True"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.0007951782899908721, 0.25260868899485406, 0.5044221996997172, 0.7562357104045805, 1.0080492211094436, 1.2598627318143067, 1.51167624251917, 1.7634897532240332, 2.0153032639288964, 2.2671167746337595, 2.5189302853386226, 2.7707437960434858, 3.0225573067483493, 3.2743708174532125, 3.5261843281580756, 3.56]\n"
     ]
    },
    {
     "data": {
      "text/plain": "<Figure size 1152x720 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Histogram of attention weights\n",
    "def reject_outliers(data, m=2):\n",
    "    return data[data - np.mean(data) < m * np.std(data)]\n",
    "\n",
    "def get_normalized_weights(df, col_idx = 5, mode='extend'):\n",
    "    acc_attention_weights = []\n",
    "    for row in df.itertuples():\n",
    "        weights = eval(row[col_idx])\n",
    "        weights = [w * len(weights) for w in weights]\n",
    "        if mode == 'extend':\n",
    "            acc_attention_weights.extend(weights)\n",
    "        else:\n",
    "            acc_attention_weights.append(weights)\n",
    "    \n",
    "    #print(f'Number of attention weights: {len(acc_attention_weights) if mode==\"extend\" else len(acc_attention_weights[0])}, 'f' Number of Batches: {len(df_att)}')\n",
    "    att_w = np.array(acc_attention_weights)\n",
    "    return att_w\n",
    "\n",
    "#att_w = reject_outliers(att_w, m=2)\n",
    "\n",
    "num_bins = 150\n",
    "upper_value = 3.56\n",
    "#df_att = pd.read_csv('/home/simon/Code/krepelschrofen/MasterThesis/docs/training_mil_avg_0.744_0.700.csv')\n",
    "#df_att = pd.read_csv('~/Code/krepelschrofen/pad_classification/20201210_1635_fold5_PAD/49_last_pad_model_0.615.csv')\n",
    "df_att = pd.read_csv(WORKING_FILE)\n",
    "att_w = get_normalized_weights(df_att)\n",
    "\n",
    "_, bin_edges = np.histogram(att_w, bins=num_bins)\n",
    "bin_edges = [b for b in bin_edges if b < upper_value] + [upper_value]\n",
    "att_w[att_w >= upper_value] = upper_value\n",
    "print(bin_edges)\n",
    "\n",
    "fig = plt.figure(figsize=(16, 10))\n",
    "plt.hist(att_w, density=False, alpha=0.7, rwidth=0.85, bins=bin_edges)\n",
    "#mn, mx = plt.xlim()\n",
    "#kde_xs = np.linspace(mn, mx, 301)\n",
    "#kde = st.gaussian_kde(att_w)\n",
    "#plt.plot(kde_xs, kde.pdf(kde_xs), label=\"PDF\")\n",
    "plt.ylabel('Count')\n",
    "plt.xlabel('$w_{att}$')\n",
    "plt.title(\"Histogram of attention weights\")\n",
    "plt.savefig(os.path.join('/home/simon/Pictures', f'{CUR_DATE}_Histogram of attention weights.png'))\n",
    "\n",
    "if SWITCH_CLASSES:\n",
    "    df_att['label'] = df_att['label'].replace({0:1, 1:0})\n",
    "    df_att['prediction'] = df_att['prediction'].replace({0:1, 1:0})\n",
    "    df_att['probability'] = 1.0 - df_att['probability']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC AUC:  0.628130460104834\n"
     ]
    },
    {
     "data": {
      "text/plain": "<Figure size 1152x720 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# calc ROC AUC\n",
    "def get_roc_values(path, no_skill = False, group_by_patient=False):\n",
    "    # df_att = pd.read_csv(path)\n",
    "    df_val = df_att.iloc[:, :4]\n",
    "\n",
    "    if group_by_patient:\n",
    "        df_val = df_val.groupby(['eye_id']).agg(\n",
    "            label=pd.NamedAgg(column='label', aggfunc='first'),\n",
    "            prediction=pd.NamedAgg(column='prediction', aggfunc=lambda x: 1 if np.mean(x) >= 0.5 else 0),\n",
    "            probability=pd.NamedAgg(column='probability', aggfunc='sum')\n",
    "        ).reset_index()\n",
    "\n",
    "    #df_val = df_val.groupby(['eye_id']).mean().reset_index()   # Average across bags\n",
    "    ns_auc, ns_fpr, ns_tpr = 0, 0, 0\n",
    "\n",
    "    if no_skill:\n",
    "        ns_probs = [0 for _ in range(len(df_val))]\n",
    "        ns_auc = roc_auc_score(df_val['label'], ns_probs)\n",
    "        ns_fpr, ns_tpr, _ = roc_curve(df_val['label'], ns_probs)\n",
    "    \n",
    "    lr_auc = roc_auc_score(df_val['label'], df_val['probability'])\n",
    "    lr_fpr, lr_tpr, ths = roc_curve(df_val['label'], df_val['probability'])\n",
    "    \n",
    "    print('ROC AUC: ', lr_auc)\n",
    "    return ((lr_auc, ns_auc), lr_fpr, lr_tpr, ns_fpr, ns_tpr) if no_skill else ((lr_auc, None), lr_fpr, lr_tpr, None, None)\n",
    "\n",
    "\n",
    "def get_prroc_values(path, no_skill = False, group_by_patient=False):\n",
    "    # df_att = pd.read_csv(path)\n",
    "    df_val = df_att.iloc[:, :4]\n",
    "\n",
    "    if group_by_patient:\n",
    "        df_val = df_val.groupby(['eye_id']).agg(\n",
    "            label=pd.NamedAgg(column='label', aggfunc='first'),\n",
    "            prediction=pd.NamedAgg(column='prediction', aggfunc=lambda x: 1 if np.mean(x) >= 0.5 else 0),\n",
    "            probability=pd.NamedAgg(column='probability', aggfunc='sum')\n",
    "        ).reset_index()\n",
    "\n",
    "    precision, recall, ths = precision_recall_curve(df_val['label'], df_val['probability'])\n",
    "    pr_auc = metrics.auc(recall, precision)\n",
    "    ns = 0\n",
    "    \n",
    "    if no_skill:\n",
    "        ns = len(df_val['label'][df_val['label']==1]) / len(df_val['label'])\n",
    "        print(ns)\n",
    "    \n",
    "    print('PR AUC: ', pr_auc)\n",
    "    return (pr_auc, precision, recall, ns if no_skill else None)\n",
    "\n",
    "\n",
    "#auc_mil, lr_fpr, lr_tpr, ns_fpr, ns_tpr = get_roc_values('/home/simon/Code/krepelschrofen/pad_classification/20201210_1703_fold6_PAD/49_last_pad_model_0.5.csv', no_skill=True)\n",
    "auc_mil, lr_fpr, lr_tpr, ns_fpr, ns_tpr = get_roc_values(WORKING_FILE, no_skill=True, group_by_patient=False)\n",
    "#get_roc_values('../docs/training_mil_avg_0.744_0.700.csv', no_skill=True)\n",
    "#auc_nopre, lr_fpr_nopre, lr_tpr_nopre, _, _ = get_roc_values('../docs/training_mil_avg_svm_nopretraining.csv')\n",
    "#auc_nosvm, lr_fpr_nosvm, lr_tpr_nosvm, _, _ = get_roc_values('../docs/training_mil_avg_nosvm_pretrained.csv')\n",
    "\n",
    "\n",
    "plt.figure(figsize=(16, 10))\n",
    "plt.plot(ns_fpr, ns_tpr, linestyle='--', label='No Skill')\n",
    "plt.plot(lr_fpr, lr_tpr, marker='.', label=f'MIL AUC: {auc_mil[0]:0.3f}')\n",
    "#plt.plot(lr_fpr_nopre, lr_tpr_nopre, linestyle='-.', marker='.', label=f'w/o KAGGLE AUC: {auc_nopre[0]:0.3f}', linewidth=4)\n",
    "#plt.plot(lr_fpr_nosvm, lr_tpr_nosvm, linestyle='-.', marker='.', label=f'w/o SVM AUC: {auc_nosvm[0]:0.3f}', linewidth=4)\n",
    "plt.xlabel('False Positive Rate') # axis labels\n",
    "plt.ylabel('True Positive Rate')\n",
    "# plt.title(f'ROC AUC = {lr_auc:0.3f}')\n",
    "plt.title(f'ROC Curves')\n",
    "plt.legend() # show the legend\n",
    "plt.savefig(f'/home/simon/Pictures/{CUR_DATE}_pad_mil_roc_curve')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.7481481481481481\n",
      "PR AUC:  0.8390030142157039\n",
      "Precision recall AUC:  0.8390030142157039\n"
     ]
    },
    {
     "data": {
      "text/plain": "<Figure size 1152x720 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Precision-recall curve\n",
    "\n",
    "pr_auc_mil, pre, rec, ns = get_prroc_values(WORKING_FILE, no_skill=True, group_by_patient=False)\n",
    "#pr_auc_nopre, pre_nopre, rec_nopre, _ = get_prroc_values('../docs/training_mil_avg_svm_nopretraining.csv')\n",
    "#pr_auc_nosvm, pre_nosvm, rec_nosvm, _ = get_prroc_values('../docs/training_mil_avg_nosvm_pretrained.csv')\n",
    "\n",
    "plt.figure(figsize=(16, 10))\n",
    "plt.plot([0, 1], [ns, ns], linestyle='--', label='No Skill')\n",
    "plt.plot(rec, pre, marker='.', label=f'MIL AUC: {pr_auc_mil:0.3f}')\n",
    "#plt.plot(rec_nopre, pre_nopre, linestyle='-.', marker='.', linewidth=4, label=f'w/o KAGGLE AUC: {pr_auc_nopre:0.3f}')\n",
    "#plt.plot(rec_nosvm, pre_nosvm, linestyle='-.', marker='.', linewidth=4, label=f'w/o SVM AUC: {pr_auc_nosvm:0.3f}')\n",
    "\n",
    "plt.xlabel('Recall') # axis labels\n",
    "plt.ylabel('Precision')\n",
    "plt.title(f'Precision-Recall AUC')\n",
    "#plt.title(f'Precision-Recall AUC = {pr_auc:0.3f}')\n",
    "plt.legend() # show the legend\n",
    "plt.savefig(f'/home/simon/Pictures/{CUR_DATE}_pad_mil_pr_curve.png')\n",
    "print('Precision recall AUC: ', pr_auc_mil)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Metrics for /home/simon/Documents/20210719_299_ablation/fused.csv\n",
      " Accuracy: 0.6592592592592592\n",
      " F1: 0.7578947368421052\n",
      " Precision: 0.8089887640449438\n",
      " Recall: 0.7128712871287128\n",
      " ROC AUC: 0.628130460104834\n",
      " PR AUC: 0.8404775301080869\n"
     ]
    }
   ],
   "source": [
    "#df_att = df_att.groupby(['eye_id']).agg(\n",
    "#            label=pd.NamedAgg(column='label', aggfunc='first'),\n",
    "#            prediction=pd.NamedAgg(column='prediction', aggfunc=lambda x: 1 if np.mean(x) >= 0.5 else 0),\n",
    "#            probability=pd.NamedAgg(column='probability', aggfunc='sum')\n",
    "#        ).reset_index()\n",
    "print(f'Metrics for {WORKING_FILE}')\n",
    "print(f' Accuracy: {metrics.accuracy_score(df_att[\"label\"], df_att[\"prediction\"])}')\n",
    "print(f' F1: {metrics.f1_score(df_att[\"label\"], df_att[\"prediction\"])}')\n",
    "print(f' Precision: {metrics.precision_score(df_att[\"label\"], df_att[\"prediction\"])}')\n",
    "print(f' Recall: {metrics.recall_score(df_att[\"label\"], df_att[\"prediction\"])}')\n",
    "print(f' ROC AUC: {metrics.roc_auc_score(df_att[\"label\"], df_att[\"probability\"])}')\n",
    "print(f' PR AUC: {metrics.average_precision_score(df_att[\"label\"], df_att[\"probability\"])}')\n"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": "HBox(children=(HTML(value=''), FloatProgress(value=0.0, max=135.0), HTML(value='')))",
      "application/vnd.jupyter.widget-view+json": {
       "version_major": 2,
       "version_minor": 0,
       "model_id": "d2573901dfa74e23ba828c528109d0c5"
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def plot_attention_map(attention_weights_path, image_path, black_th=0.5, ceil_th=3.56, segment_size=399, silent=False, grid_color=(0, 255, 0)):\n",
    "    example_examination = os.path.splitext(os.path.basename(image_path))[0]\n",
    "    df_att = pd.read_csv(attention_weights_path)\n",
    "    #df_att['prediction'] = df_att['prediction'].apply(lambda d: eval(d)[0])\n",
    "    att_w = get_normalized_weights(df_att, mode='append')\n",
    "\n",
    "    example_idx = df_att[df_att['eye_id'] == example_examination].index\n",
    "    example_img = io.imread(image_path[:-4] + '.jpg')\n",
    "    example_img[:,::segment_size,:] = list(grid_color)\n",
    "    example_img[::segment_size,:,:] = list(grid_color)\n",
    "    example_row = df_att[df_att['eye_id'] == example_examination]\n",
    "    prediction, label = example_row['prediction'].values[0], int(example_row['label'].values[0])\n",
    "\n",
    "    count, seg_count = 0, 0\n",
    "    weights = att_w[example_idx]\n",
    "    weights[weights >= ceil_th] = ceil_th\n",
    "    weights = (weights - np.min(weights)) / (np.max(weights) - np.min(weights))\n",
    "    for y in range(0, example_img.shape[0], segment_size):\n",
    "        for x in range(0, example_img.shape[1], segment_size):\n",
    "            segment = example_img[y:y+segment_size, x:x+segment_size]\n",
    "            if segment.shape[0] * segment.shape[1] != segment_size**2:\n",
    "                continue\n",
    "            count_non_black_px = cv2.countNonZero(cv2.cvtColor(segment, cv2.COLOR_BGR2GRAY))\n",
    "            if count_non_black_px / segment_size**2 < black_th:\n",
    "                count += 1\n",
    "                weights = np.insert(weights, seg_count, 0.0)\n",
    "            seg_count += 1\n",
    "    if not silent: print(f'Inserted attention weights for {count} black segements')\n",
    "\n",
    "    att_heatmap = weights.reshape((example_img.shape[0] // segment_size, example_img.shape[1] // segment_size))\n",
    "    example_img = example_img[:segment_size * att_heatmap.shape[0], :segment_size * att_heatmap.shape[1]]\n",
    "    att_heatmap_scale = cv2.resize(att_heatmap, (example_img.shape[1], example_img.shape[0]))\n",
    "    att_heatmap_scale = (att_heatmap_scale * 256).astype(np.uint8)\n",
    "    att_heatmap_scale = cv2.merge((att_heatmap_scale, np.zeros_like(att_heatmap_scale), np.zeros_like(att_heatmap_scale)))\n",
    "\n",
    "    plt.axis('off')\n",
    "    fig = plt.figure(figsize=(10., 10.))\n",
    "    fin = cv2.addWeighted(att_heatmap_scale, 0.5, example_img, 0.5, 0)\n",
    "    plt.imshow(fin)\n",
    "    output_path = f'Attention_map_label{label}_pred{prediction}_{example_examination}.png'\n",
    "    plt.savefig(os.path.join(OUTPUT_FOLDER, output_path))\n",
    "    plt.close(fig)\n",
    "\n",
    "#att_path = '~/Documents/Fused validation results/pad_mil_fused_results.csv'\n",
    "img_path = '~/Data/pAVK_dataset/Color_anonym/OR1248R.jpg'\n",
    "#plot_attention_map(att_path, img_path)\n",
    "\n",
    "df_eyes = pd.read_csv(WORKING_FILE)\n",
    "for i, rows in tqdm(df_eyes.iterrows(), total=len(df_eyes)):\n",
    "    eye_path = os.path.join('~/Data/pAVK_dataset/Color_anonym/', rows['eye_id'] + '.jpg')\n",
    "    plot_attention_map(WORKING_FILE, eye_path, silent=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}