BreastCancerSwinTransformer.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "6f8e8653",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "6f8e8653",
    "outputId": "353edb87-2262-4ac6-99a7-c4f1dd7be060"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/rphorsterphages_gmail_com/miniconda3/lib/python3.8/site-packages/scipy/__init__.py:146: UserWarning: A NumPy version >=1.16.5 and <1.23.0 is required for this version of SciPy (detected version 1.23.5\n",
      "  warnings.warn(f\"A NumPy version >={np_minversion} and <{np_maxversion}\"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: kaggle in ./miniconda3/lib/python3.8/site-packages (1.5.13)\n",
      "Requirement already satisfied: urllib3 in ./miniconda3/lib/python3.8/site-packages (from kaggle) (1.26.13)\n",
      "Requirement already satisfied: six>=1.10 in ./miniconda3/lib/python3.8/site-packages (from kaggle) (1.16.0)\n",
      "Requirement already satisfied: python-slugify in ./miniconda3/lib/python3.8/site-packages (from kaggle) (8.0.1)\n",
      "Requirement already satisfied: tqdm in ./miniconda3/lib/python3.8/site-packages (from kaggle) (4.64.1)\n",
      "Requirement already satisfied: requests in ./miniconda3/lib/python3.8/site-packages (from kaggle) (2.28.1)\n",
      "Requirement already satisfied: python-dateutil in ./miniconda3/lib/python3.8/site-packages (from kaggle) (2.8.2)\n",
      "Requirement already satisfied: certifi in ./miniconda3/lib/python3.8/site-packages (from kaggle) (2022.12.7)\n",
      "Requirement already satisfied: text-unidecode>=1.3 in ./miniconda3/lib/python3.8/site-packages (from python-slugify->kaggle) (1.3)\n",
      "Requirement already satisfied: idna<4,>=2.5 in ./miniconda3/lib/python3.8/site-packages (from requests->kaggle) (3.4)\n",
      "Requirement already satisfied: charset-normalizer<3,>=2 in ./miniconda3/lib/python3.8/site-packages (from requests->kaggle) (2.0.4)\n",
      "\u001b[33mWARNING: Error parsing requirements for biopython: [Errno 2] No such file or directory: '/home/rphorsterphages_gmail_com/miniconda3/lib/python3.8/site-packages/biopython-1.79.dist-info/METADATA'\u001b[0m\u001b[33m\n",
      "\u001b[0m"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import torchvision.models as models\n",
    "import torchvision.transforms as transforms\n",
    "from torchvision.datasets import ImageFolder\n",
    "from torch.utils.data import DataLoader, Dataset, WeightedRandomSampler\n",
    "import glob\n",
    "import fnmatch \n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.image as mpimg\n",
    "from sklearn.metrics import roc_curve, auc, f1_score, confusion_matrix, balanced_accuracy_score\n",
    "from sklearn.model_selection import train_test_split\n",
    "import os\n",
    "from tqdm import tqdm\n",
    "from collections import Counter\n",
    "import concurrent.futures\n",
    "from functools import partial\n",
    "from PIL import Image\n",
    "from os import listdir\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "!pip install kaggle\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "5049e69c",
   "metadata": {
    "id": "5049e69c"
   },
   "outputs": [],
   "source": [
    "# Define custom dataset\n",
    "class BreastCancerDataset(Dataset):\n",
    "    def __init__(self, image_paths, labels, transform=None):\n",
    "        self.image_paths = image_paths\n",
    "        self.labels = labels\n",
    "        self.transform = transform\n",
    "\n",
    "    def __len__(self):\n",
    "        return len(self.image_paths)\n",
    "\n",
    "    def __getitem__(self, index):\n",
    "        image_path = self.image_paths[index]\n",
    "        label = self.labels[index]\n",
    "\n",
    "        image = Image.open(image_path)\n",
    "\n",
    "        if self.transform:\n",
    "            image = self.transform(image)\n",
    "\n",
    "        return image, label"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c9507df3",
   "metadata": {},
   "source": [
    "The below section has three code blocks to 1) download the dataset and extracts labels. 2) download the dataset and load the extracted labels and 3) just extract the labels. This is so you don't have to re-download or extract labels."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0caf1bed",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "0caf1bed",
    "outputId": "efc5ed12-44de-43c1-ef13-866133451767"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Warning: Your Kaggle API key is readable by other users on this system! To fix this, you can run 'chmod 600 /root/.kaggle/kaggle.json'\n",
      "Downloading breast-histopathology-images.zip to /content\n",
      "100% 3.10G/3.10G [02:27<00:00, 24.1MB/s]\n",
      "100% 3.10G/3.10G [02:27<00:00, 22.6MB/s]\n",
      "555048\n",
      "dataset_path/13401/1/13401_idx5_x1701_y801_class1.png\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 555048/555048 [34:05<00:00, 271.35it/s]\n"
     ]
    }
   ],
   "source": [
    "\n",
    "# Set the paths for the Kaggle data\n",
    "dataset_zip_path = 'breast-histopathology-images.zip'\n",
    "dataset_extracted_path = 'dataset_path'\n",
    "\n",
    "# Check if the dataset directory already exists\n",
    "if not os.path.exists(dataset_extracted_path):\n",
    "    # Check if the dataset zip file already exists\n",
    "    if not os.path.exists(dataset_zip_path):\n",
    "        # Move the Kaggle API file\n",
    "        !mkdir -p ~/.kaggle\n",
    "        !mv kaggle.json ~/.kaggle/\n",
    "        # Download the Kaggle dataset\n",
    "        !kaggle datasets download -d paultimothymooney/breast-histopathology-images\n",
    "        # Extract the dataset\n",
    "        !unzip -q breast-histopathology-images.zip -d $dataset_extracted_path\n",
    "    else:\n",
    "        print(\"Dataset zip file already exists.\")\n",
    "else:\n",
    "    print(\"Dataset directory already exists.\")\n",
    "\n",
    "# Set the paths to the downloaded dataset\n",
    "image_paths = glob.glob('dataset_path/*/*/*.png', recursive=True)\n",
    "print(len(image_paths))\n",
    "print(image_paths[0])\n",
    "\n",
    "# get labels\n",
    "patternZero = '*class0.png'\n",
    "patternOne = '*class1.png'\n",
    "classZero = fnmatch.filter(image_paths, patternZero)\n",
    "classOne = fnmatch.filter(image_paths, patternOne)\n",
    "\n",
    "\n",
    "labels = []\n",
    "\n",
    "def process_image(img, classZero, classOne):\n",
    "    if img in classZero:\n",
    "        return 0\n",
    "    elif img in classOne:\n",
    "        return 1\n",
    "\n",
    "def process_images_parallel(image_paths, classZero, classOne):\n",
    "    with concurrent.futures.ThreadPoolExecutor() as executor:\n",
    "        process_func = partial(process_image, classZero=classZero, classOne=classOne)\n",
    "        labels = list(tqdm(executor.map(process_func, image_paths), total=len(image_paths)))\n",
    "    return labels\n",
    "\n",
    "labels = process_images_parallel(image_paths, classZero, classOne)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "qxiHnHAqA42-",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "qxiHnHAqA42-",
    "outputId": "c69782d8-2088-45ee-906c-e833d661a40d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Warning: Your Kaggle API key is readable by other users on this system! To fix this, you can run 'chmod 600 /home/rphorsterphages_gmail_com/.kaggle/kaggle.json'\n",
      "Downloading breast-histopathology-images.zip to /home/rphorsterphages_gmail_com\n",
      "100%|█████████████████████████████████████▉| 3.09G/3.10G [00:14<00:00, 78.6MB/s]\n",
      "100%|███████████████████████████████████████| 3.10G/3.10G [00:14<00:00, 224MB/s]\n",
      "555048\n",
      "dataset_path/12818/0/12818_idx5_x2551_y2051_class0.png\n"
     ]
    }
   ],
   "source": [
    "# Set the paths for the Kaggle data\n",
    "dataset_zip_path = 'breast-histopathology-images.zip'\n",
    "dataset_extracted_path = 'dataset_path'\n",
    "\n",
    "# Check if the dataset directory already exists\n",
    "if not os.path.exists(dataset_extracted_path):\n",
    "    # Check if the dataset zip file already exists\n",
    "    if not os.path.exists(dataset_zip_path):\n",
    "        # Move the Kaggle API file\n",
    "        !mkdir -p ~/.kaggle\n",
    "        !mv kaggle.json ~/.kaggle/\n",
    "        # Download the Kaggle dataset\n",
    "        !kaggle datasets download -d paultimothymooney/breast-histopathology-images\n",
    "        # Extract the dataset\n",
    "        !unzip -q breast-histopathology-images.zip -d $dataset_extracted_path\n",
    "    else:\n",
    "        print(\"Dataset zip file already exists.\")\n",
    "else:\n",
    "    print(\"Dataset directory already exists.\")\n",
    "\n",
    "# Set the paths to the downloaded dataset\n",
    "image_paths = glob.glob('dataset_path/*/*/*.png', recursive=True)\n",
    "print(len(image_paths))\n",
    "print(image_paths[0])\n",
    "labels = []\n",
    "with open('labels.txt', 'r') as file:\n",
    "    for line in file:\n",
    "        label = int(line.strip())\n",
    "        labels.append(label)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "PqJ1swfaTE-W",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "PqJ1swfaTE-W",
    "outputId": "81162fd1-ba72-44fa-a1ef-d4bc5e9c75d8"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "277524\n",
      "dataset_path/12818/0/12818_idx5_x2551_y2051_class0.png\n",
      "277524\n"
     ]
    }
   ],
   "source": [
    "# Set the paths to the downloaded dataset\n",
    "dataset_extracted_path = 'dataset_path'\n",
    "image_paths = glob.glob('dataset_path/*/*/*.png', recursive=True)\n",
    "print(len(image_paths))\n",
    "print(image_paths[0])\n",
    "labels = []\n",
    "with open('labels.txt', 'r') as file:\n",
    "    for line in file:\n",
    "        label = int(line.strip())\n",
    "        labels.append(label)\n",
    "print(len(labels))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "35787b8d",
   "metadata": {},
   "source": [
    "The block below uses the images with no patient context to grab positive and negative samples for invasive ductal carcinoma. What you will see is that it is very hard to tell the difference and may require patient context, also we might have leakage into test set and ideally we want to test on a different distributions to truely validate the model. https://datascience.stanford.edu/news/splitting-data-randomly-can-ruin-your-model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "a3552b4d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Assuming 'image_paths' contains the paths to the images\n",
    "# Assuming 'labels' contains the corresponding labels (0 for negative, 1 for positive)\n",
    "\n",
    "# Find the indices of positive and negative examples\n",
    "positive_indices = [i for i, label in enumerate(labels) if label == 1]\n",
    "negative_indices = [i for i, label in enumerate(labels) if label == 0]\n",
    "\n",
    "# Display a positive image\n",
    "positive_image_path = image_paths[positive_indices[1]]\n",
    "positive_image = mpimg.imread(positive_image_path)\n",
    "plt.imshow(positive_image)\n",
    "plt.title(\"Positive Example\")\n",
    "plt.axis('off')\n",
    "plt.show()\n",
    "\n",
    "# Display a negative image\n",
    "negative_image_path = image_paths[negative_indices[200]]\n",
    "negative_image = mpimg.imread(negative_image_path)\n",
    "plt.imshow(negative_image)\n",
    "plt.title(\"Negative Example\")\n",
    "plt.axis('off')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1539d27b",
   "metadata": {},
   "source": [
    "To provide pateint context I will create a dataframe with patient ids, image paths and target/labels that I can use to split data by patient."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "bdf13b6d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>patient_id</th>\n",
       "      <th>path</th>\n",
       "      <th>target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>12818</td>\n",
       "      <td>dataset_path/12818/0/12818_idx5_x2551_y2051_cl...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>12818</td>\n",
       "      <td>dataset_path/12818/0/12818_idx5_x851_y1451_cla...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>12818</td>\n",
       "      <td>dataset_path/12818/0/12818_idx5_x1251_y901_cla...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>12818</td>\n",
       "      <td>dataset_path/12818/0/12818_idx5_x3651_y1651_cl...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>12818</td>\n",
       "      <td>dataset_path/12818/0/12818_idx5_x1651_y1951_cl...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  patient_id                                               path target\n",
       "0      12818  dataset_path/12818/0/12818_idx5_x2551_y2051_cl...      0\n",
       "1      12818  dataset_path/12818/0/12818_idx5_x851_y1451_cla...      0\n",
       "2      12818  dataset_path/12818/0/12818_idx5_x1251_y901_cla...      0\n",
       "3      12818  dataset_path/12818/0/12818_idx5_x3651_y1651_cl...      0\n",
       "4      12818  dataset_path/12818/0/12818_idx5_x1651_y1951_cl...      0"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset_patient_path = \"dataset_path/\"\n",
    "folder = os.listdir(dataset_patient_path)\n",
    "\n",
    "num_images = len(image_paths)\n",
    "data = pd.DataFrame(index=range(num_images), columns=[\"patient_id\", \"path\", \"target\"])\n",
    "\n",
    "k = 0\n",
    "for n in range(len(folder)):\n",
    "    patient_id = folder[n]\n",
    "    if patient_id == \"IDC_regular_ps50_idx5\":\n",
    "        continue  # Skip the IDC_regular_ps50_idx5 folder\n",
    "        \n",
    "    patient_path = os.path.join(dataset_patient_path, patient_id)\n",
    "    \n",
    "    for c in [0, 1]:\n",
    "        class_path = os.path.join(patient_path, str(c))\n",
    "        subfiles = os.listdir(class_path)\n",
    "        \n",
    "        data.loc[k:k+len(subfiles)-1, \"path\"] = [os.path.join(class_path, subfile) for subfile in subfiles]\n",
    "        data.loc[k:k+len(subfiles)-1, \"target\"] = c\n",
    "        data.loc[k:k+len(subfiles)-1, \"patient_id\"] = patient_id\n",
    "        \n",
    "        k += len(subfiles)\n",
    "\n",
    "data.reset_index(drop=True, inplace=True)\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f871b791",
   "metadata": {},
   "source": [
    "To prevent overfitting I flipped the photos randomly and I normalized the data using imagenet to train relative to the distribution of the pretrained model's."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "17e1fe5b",
   "metadata": {},
   "outputs": [],
   "source": [
    "##########################Training with patient context:\n",
    "\n",
    "patients = data.patient_id.unique()\n",
    "\n",
    "\n",
    "train_ids, sub_test_ids = train_test_split(patients,\n",
    "                                           test_size=0.3,\n",
    "                                           random_state=42)\n",
    "test_ids, dev_ids = train_test_split(sub_test_ids, test_size=0.5, random_state=42)\n",
    "\n",
    "\n",
    "train_df = data[data.patient_id.isin(train_ids)]\n",
    "dev_df = data[data.patient_id.isin(dev_ids)]\n",
    "test_df = data[data.patient_id.isin(test_ids)]\n",
    "\n",
    "train_image_paths = list(train_df['path'])\n",
    "train_labels = list(train_df['target'])\n",
    "dev_image_paths = list(dev_df['path'])\n",
    "dev_labels = list(dev_df['target'])\n",
    "test_image_paths = list(test_df['path'])\n",
    "test_labels = list(test_df['target'])\n",
    "\n",
    "transform = transforms.Compose([\n",
    "    transforms.Resize((50,50)),\n",
    "    transforms.RandomHorizontalFlip(),\n",
    "    transforms.RandomVerticalFlip(),\n",
    "    transforms.ToTensor(),\n",
    "    transforms.Normalize(mean=[0.485, 0.456, 0.406], std= [0.229, 0.224, 0.225])\n",
    "])\n",
    "\n",
    "class_freq = Counter(train_df['target'])\n",
    "total_samples = len(train_df['target'])\n",
    "class_weights = [total_samples / (class_freq[i] * len(class_freq)) for i in range(len(class_freq))]\n",
    "class_weights = torch.tensor(class_weights, dtype=torch.double)\n",
    "sample_weights = [class_weights[label] for label in train_df['target']]\n",
    "sampler = WeightedRandomSampler(sample_weights, len(train_labels), replacement=True)\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "train_dataset = BreastCancerDataset(train_image_paths, train_labels, transform=transform)\n",
    "dev_dataset = BreastCancerDataset(dev_image_paths, dev_labels, transform=transform)\n",
    "test_dataset = BreastCancerDataset(test_image_paths, test_labels, transform=transform)\n",
    "\n",
    "\n",
    "batch_size = 32\n",
    "#train_dataloader = DataLoader(train_dataset, batch_size=batch_size, sampler=sampler, num_workers=32 ,pin_memory=True)\n",
    "train_dataloader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True, num_workers=32 ,pin_memory=True)#########################"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cc050af3",
   "metadata": {},
   "source": [
    "I am using Swin Transformer, the model is likely overkill, however I want to play with the latest model architecture for computer vision tasks and compare to the model orignally used on this dataset. SwinTransformer:https://openaccess.thecvf.com/content/CVPR2022/papers/Liu_Swin_Transformer_V2_Scaling_Up_Capacity_and_Resolution_CVPR_2022_paper.pdf "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "e2bbeb67",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "e2bbeb67",
    "outputId": "6d64b096-f141-49ec-9237-b4016b4b8409"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_27882/2846083555.py:22: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).\n",
      "  class_weights_tensor = torch.tensor(class_weights, dtype=torch.float32)\n"
     ]
    }
   ],
   "source": [
    "# Load the pre-trained model\n",
    "model = models.swin_v2_t(weights='DEFAULT')\n",
    "# Freeze all layers except the last one\n",
    "for param in model.parameters():\n",
    "    param.requires_grad = False\n",
    "\n",
    "# Replace the last fully connected layer with a new one\n",
    "num_ftrs = model.head.in_features\n",
    "model.head = nn.Sequential(\n",
    "    nn.Linear(num_ftrs, 1024),\n",
    "    nn.BatchNorm1d(1024),  # BatchNorm layer added\n",
    "    nn.ReLU(),\n",
    "    nn.Dropout(0.5),\n",
    "    nn.Linear(1024, 512),\n",
    "    nn.BatchNorm1d(512),  # BatchNorm layer added\n",
    "    nn.ReLU(),\n",
    "    nn.Dropout(0.5),\n",
    "    nn.Linear(512, 2)\n",
    ")\n",
    "\n",
    "# Define the loss function and optimizer (uses frequency weights)\n",
    "class_weights_tensor = torch.tensor(class_weights, dtype=torch.float32)\n",
    "criterion = nn.CrossEntropyLoss(weight=class_weights_tensor)\n",
    "#criterion = nn.CrossEntropyLoss()\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr=0.0001, weight_decay = 0.0001)\n",
    "scheduler = torch.optim.lr_scheduler.StepLR(optimizer, step_size=5, gamma=0.1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "d96e4f75",
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "d96e4f75",
    "outputId": "f64bb86b-babc-4675-dcbd-be4338d31214"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  3%|████▏                                                                                                                        | 1/30 [16:26<7:56:37, 986.14s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [1/30], Loss: 0.4334\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  7%|████████▎                                                                                                                    | 2/30 [32:56<7:41:22, 988.68s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [2/30], Loss: 0.3981\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 10%|████████████▍                                                                                                               | 3/30 [49:56<7:31:25, 1003.16s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [3/30], Loss: 0.3915\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 13%|████████████████▍                                                                                                          | 4/30 [1:06:17<7:10:52, 994.32s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [4/30], Loss: 0.3850\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 17%|████████████████████▌                                                                                                      | 5/30 [1:22:20<6:49:33, 982.94s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [5/30], Loss: 0.3820\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 20%|████████████████████████▌                                                                                                  | 6/30 [1:38:34<6:31:56, 979.85s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [6/30], Loss: 0.3803\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 23%|████████████████████████████▋                                                                                              | 7/30 [1:54:45<6:14:33, 977.11s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [7/30], Loss: 0.3766\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 27%|████████████████████████████████▊                                                                                          | 8/30 [2:10:48<5:56:38, 972.67s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [8/30], Loss: 0.3744\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 30%|████████████████████████████████████▉                                                                                      | 9/30 [2:26:55<5:39:45, 970.76s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [9/30], Loss: 0.3729\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 33%|████████████████████████████████████████▋                                                                                 | 10/30 [2:43:05<5:23:31, 970.60s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [10/30], Loss: 0.3710\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 37%|████████████████████████████████████████████▋                                                                             | 11/30 [2:59:19<5:07:37, 971.44s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [11/30], Loss: 0.3695\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 40%|████████████████████████████████████████████████▊                                                                         | 12/30 [3:15:08<4:49:24, 964.68s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [12/30], Loss: 0.3693\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 43%|████████████████████████████████████████████████████▊                                                                     | 13/30 [3:30:48<4:31:14, 957.35s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [13/30], Loss: 0.3681\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 47%|████████████████████████████████████████████████████████▉                                                                 | 14/30 [3:46:52<4:15:50, 959.40s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [14/30], Loss: 0.3681\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 50%|█████████████████████████████████████████████████████████████                                                             | 15/30 [4:02:25<3:57:49, 951.31s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [15/30], Loss: 0.3662\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 53%|█████████████████████████████████████████████████████████████████                                                         | 16/30 [4:18:38<3:43:30, 957.89s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [16/30], Loss: 0.3664\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 57%|█████████████████████████████████████████████████████████████████████▏                                                    | 17/30 [4:34:41<3:27:49, 959.22s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [17/30], Loss: 0.3649\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 60%|█████████████████████████████████████████████████████████████████████████▏                                                | 18/30 [4:50:59<3:13:00, 965.07s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [18/30], Loss: 0.3657\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 63%|█████████████████████████████████████████████████████████████████████████████▎                                            | 19/30 [5:07:02<2:56:48, 964.40s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [19/30], Loss: 0.3655\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 67%|█████████████████████████████████████████████████████████████████████████████████▎                                        | 20/30 [5:23:01<2:40:26, 962.70s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [20/30], Loss: 0.3627\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 70%|█████████████████████████████████████████████████████████████████████████████████████▍                                    | 21/30 [5:39:27<2:25:28, 969.84s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [21/30], Loss: 0.3616\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 73%|█████████████████████████████████████████████████████████████████████████████████████████▍                                | 22/30 [5:55:24<2:08:46, 965.77s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [22/30], Loss: 0.3637\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 77%|█████████████████████████████████████████████████████████████████████████████████████████████▌                            | 23/30 [6:11:28<1:52:38, 965.51s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [23/30], Loss: 0.3633\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 80%|█████████████████████████████████████████████████████████████████████████████████████████████████▌                        | 24/30 [6:27:58<1:37:15, 972.64s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [24/30], Loss: 0.3618\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 83%|█████████████████████████████████████████████████████████████████████████████████████████████████████▋                    | 25/30 [6:44:47<1:21:57, 983.54s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [25/30], Loss: 0.3630\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 87%|█████████████████████████████████████████████████████████████████████████████████████████████████████████▋                | 26/30 [7:01:04<1:05:27, 981.75s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [26/30], Loss: 0.3616\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 90%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████▌            | 27/30 [7:17:18<48:57, 979.27s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [27/30], Loss: 0.3614\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 93%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████▋        | 28/30 [7:33:41<32:40, 980.34s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [28/30], Loss: 0.3622\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 97%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████▊    | 29/30 [7:50:25<16:27, 987.58s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [29/30], Loss: 0.3612\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 30/30 [8:07:00<00:00, 974.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch [30/30], Loss: 0.3608\n",
      "Final Accuracy: 84.48%\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "# Train the model\n",
    "num_epochs = 30\n",
    "if torch.cuda.device_count() > 1:\n",
    "    model = nn.DataParallel(model)\n",
    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "model = model.to(device)\n",
    "model.train()\n",
    "\n",
    "# Lists to store training and test loss\n",
    "train_losses = []\n",
    "train_accuracies = []\n",
    "\n",
    "for epoch in tqdm(range(num_epochs)):\n",
    "    running_loss = 0.0\n",
    "    correct = 0\n",
    "    total = 0\n",
    "\n",
    "    for images, labels in train_dataloader:\n",
    "        images = images.to(device)\n",
    "        labels = labels.to(device)\n",
    "\n",
    "        optimizer.zero_grad()\n",
    "\n",
    "        # Forward pass\n",
    "        outputs = model(images)\n",
    "        loss = criterion(outputs, labels)\n",
    "\n",
    "        # Backward pass and optimization\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "\n",
    "        running_loss += loss.item()\n",
    "        # Calculate accuracy\n",
    "        _, predicted = torch.max(outputs.data, 1)\n",
    "        total += labels.size(0)\n",
    "        correct += (predicted == labels).sum().item()\n",
    "        \n",
    "    epoch_loss = running_loss / len(train_dataloader)\n",
    "    train_losses.append(epoch_loss)\n",
    "    print(f\"Epoch [{epoch+1}/{num_epochs}], Loss: {epoch_loss:.4f}\")\n",
    "    epoch_accuracy = 100 * correct / total\n",
    "    train_accuracies.append(epoch_accuracy)\n",
    "# Calculate final accuracy\n",
    "final_accuracy = train_accuracies[-1]\n",
    "print(f\"Final Accuracy: {final_accuracy:.2f}%\")\n",
    "\n",
    "# Save the trained model\n",
    "torch.save(model.state_dict(), 'breast_cancer_model_swin.pth')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "86dfddf0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(range(num_epochs), train_accuracies)\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Accuracy (%)')\n",
    "plt.title('Training Accuracy')\n",
    "plt.savefig('Acurve_BCMs.pdf')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "6cbd14d1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot training over time \n",
    "plt.figure()\n",
    "plt.plot(range(1, num_epochs+1), train_losses, label='Training Loss')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Loss')\n",
    "plt.title('Training Loss')\n",
    "plt.legend()\n",
    "plt.savefig('Lcurve_BCM.pdf')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "4b8f53dc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dev Accuracy: 83.33%\n",
      "Dev Loss: 0.3770\n",
      "F1 Score: 0.7318\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Evaluate the model\n",
    "dev_dataloader = DataLoader(dev_dataset, batch_size=batch_size, shuffle=False, num_workers=32 ,pin_memory=True)\n",
    "model.eval()\n",
    "correct_predictions = 0\n",
    "total_predictions = 0\n",
    "dev_losses = []\n",
    "predictions = []\n",
    "true_labels = []\n",
    "\n",
    "with torch.no_grad():\n",
    "    for images, labels in dev_dataloader:\n",
    "        images = images.to(device)\n",
    "        labels = labels.to(device)\n",
    "\n",
    "        # Forward pass\n",
    "        outputs = model(images)\n",
    "        _, predicted = torch.max(outputs.data, 1)\n",
    "\n",
    "        total_predictions += labels.size(0)\n",
    "        correct_predictions += (predicted == labels).sum().item()\n",
    "\n",
    "        # Calculate loss\n",
    "        loss = criterion(outputs, labels)\n",
    "        dev_losses.append(loss.item())\n",
    "\n",
    "        predictions.extend(predicted.cpu().numpy())\n",
    "        true_labels.extend(labels.cpu().numpy())\n",
    "\n",
    "    dev_accuracy = 100 * correct_predictions / total_predictions\n",
    "    dev_loss = sum(dev_losses) / len(dev_losses)\n",
    "    print(f\"Dev Accuracy: {dev_accuracy:.2f}%\")\n",
    "    print(f\"Dev Loss: {dev_loss:.4f}\")\n",
    "    # Calculate F1 score\n",
    "    f1 = f1_score(true_labels, predictions)\n",
    "    print(f\"F1 Score: {f1:.4f}\")\n",
    "\n",
    "# Calculate ROC curve\n",
    "fpr, tpr, thresholds = roc_curve(true_labels, predictions)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "\n",
    "# Plot ROC curve\n",
    "plt.figure()\n",
    "plt.plot(fpr, tpr, color='darkorange', lw=2, label='ROC curve (area = %0.2f)' % roc_auc)\n",
    "plt.plot([0, 1], [0, 1], color='navy', lw=2, linestyle='--')\n",
    "plt.xlim([0.0, 1.0])\n",
    "plt.ylim([0.0, 1.05])\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.legend(loc=\"lower right\")\n",
    "plt.savefig('ROC_BCMS_dev.pdf')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "85bf97f0",
   "metadata": {
    "id": "85bf97f0",
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Accuracy: 82.32%\n",
      "Test Loss: 0.4011\n",
      "F1 Score: 0.7534\n",
      "Confusion Matrix:\n",
      "[[22144  5429]\n",
      " [ 1652 10816]]\n",
      "Balanced Accuracy: 0.8353\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Evaluate the model\n",
    "test_dataloader = DataLoader(test_dataset, batch_size=batch_size, shuffle=False, num_workers=32 ,pin_memory=True)\n",
    "model.eval()\n",
    "correct_predictions = 0\n",
    "total_predictions = 0\n",
    "test_losses = []\n",
    "predictions = []\n",
    "true_labels = []\n",
    "\n",
    "with torch.no_grad():\n",
    "    for images, labels in test_dataloader:\n",
    "        images = images.to(device)\n",
    "        labels = labels.to(device)\n",
    "\n",
    "        # Forward pass\n",
    "        outputs = model(images)\n",
    "        _, predicted = torch.max(outputs.data, 1)\n",
    "\n",
    "        total_predictions += labels.size(0)\n",
    "        correct_predictions += (predicted == labels).sum().item()\n",
    "\n",
    "        # Calculate loss\n",
    "        loss = criterion(outputs, labels)\n",
    "        test_losses.append(loss.item())\n",
    "\n",
    "        predictions.extend(predicted.cpu().numpy())\n",
    "        true_labels.extend(labels.cpu().numpy())\n",
    "\n",
    "    test_accuracy = 100 * correct_predictions / total_predictions\n",
    "    test_loss = sum(test_losses) / len(test_losses)\n",
    "    print(f\"Test Accuracy: {test_accuracy:.2f}%\")\n",
    "    print(f\"Test Loss: {test_loss:.4f}\")\n",
    "    \n",
    "    # Calculate F1 score\n",
    "    f1 = f1_score(true_labels, predictions)\n",
    "    print(f\"F1 Score: {f1:.4f}\")\n",
    "    \n",
    "    # Calculate confusion matrix\n",
    "    cm = confusion_matrix(true_labels, predictions)\n",
    "    print(\"Confusion Matrix:\")\n",
    "    print(cm)\n",
    "\n",
    "    # Calculate balanced accuracy\n",
    "    balanced_acc = balanced_accuracy_score(true_labels, predictions)\n",
    "    print(f\"Balanced Accuracy: {balanced_acc:.4f}\")\n",
    "\n",
    "# Calculate ROC curve\n",
    "fpr, tpr, thresholds = roc_curve(true_labels, predictions)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "\n",
    "# Plot ROC curve\n",
    "plt.figure()\n",
    "plt.plot(fpr, tpr, color='darkorange', lw=2, label='ROC curve (area = %0.2f)' % roc_auc)\n",
    "plt.plot([0, 1], [0, 1], color='navy', lw=2, linestyle='--')\n",
    "plt.xlim([0.0, 1.0])\n",
    "plt.ylim([0.0, 1.05])\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.legend(loc=\"lower right\")\n",
    "plt.savefig('ROC_BCMS_test.pdf')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a182564d",
   "metadata": {},
   "source": [
    "Automatic detection of invasive ductal carcinoma in whole slide images with convolutional neural networks: https://www.spiedigitallibrary.org/conference-proceedings-of-spie/9041/904103/Automatic-detection-of-invasive-ductal-carcinoma-in-whole-slide-images/10.1117/12.2043872.short\n",
    "\n",
    "In this paper they used a 3-layer conv.net and achieved in terms of F1 score and balanced accuracy of (71.80%, 84.23%). Our implementation of the Swin Transformer model achieved an F1 score and balanced accuracy of (75.34%, 83.53%) on the test set. "
   ]
  }
 ],
 "metadata": {
  "colab": {
   "gpuType": "T4",
   "provenance": []
  },
  "gpuClass": "standard",
  "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": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}