MAIA seminar (Part 2).ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "ein.tags": "worksheet-0",
    "slideshow": {
     "slide_type": "-"
    }
   },
   "source": [
    "# Performing medical imaging segmentation like a pro (Part 2)\n",
    "*A highly oppiniated and biased tutorial on MRI lesion segmentation using Pytorch*\n",
    "\n",
    "**Sergi Valverde, PhD**\n",
    "*Universitat de Girona, Spain*\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ein.tags": "worksheet-0",
    "slideshow": {
     "slide_type": "-"
    }
   },
   "source": [
    "# Introduction:\n",
    "\n",
    "---\n",
    "\n",
    "During this part of the tutorial, we will focus entirely on MRI image segmentation using the U-NET architecture. This second part is structured as follows:\n",
    "\n",
    "1. The U-NET model\n",
    "2. Data preparation\n",
    "3. Training the network\n",
    "4. Inference\n",
    "5. Evaluation\n",
    "6. Conclusion\n",
    "\n",
    "First of all, import or install all necessary libraries. We will make use of [ANTS](https://github.com/ANTsX/ANTsPy), which is like a super-power to manage and process `nifti` images. \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "autoscroll": false,
    "ein.hycell": false,
    "ein.tags": "worksheet-0",
    "slideshow": {
     "slide_type": "-"
    }
   },
   "outputs": [],
   "source": [
    "# !pip install https://github.com/ANTsX/ANTsPy/releases/download/v0.2.0/antspyx-0.2.0-cp37-cp37m-linux_x86_64.whl\n",
    "# !pip install statsmodels\n",
    "import os\n",
    "import random\n",
    "import torch\n",
    "import ants\n",
    "import numpy as np\n",
    "\n",
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ein.tags": "worksheet-0",
    "slideshow": {
     "slide_type": "-"
    }
   },
   "source": [
    "# 1. The U-NET model: \n",
    "---\n",
    "\n",
    "Let's now introduce the U-NET model. The encoder-decoder U-NET model was proposed by [Ronneberger et al. 2015](https://lmb.informatik.uni-freiburg.de/Publications/2015/RFB15a/) in the context of biomedical image segmentation (microscopy imaging). Encoder-decoder models are very similar to [autoencoders](https://en.wikipedia.org/wiki/Autoencoder), so the encoding part compresses the input image into the latent space, meaning that we end up with a very efficient compressed version of the image with a lot less information (feauture space? 🤔). \n",
    "\n",
    "Based on the latent space, the decoder part reconstructs the image back. Compared to other encoder-decoder architectures like the [SegNET](https://mi.eng.cam.ac.uk/projects/segnet/), the U-NET model introduced the copy and crop elements, so activations for each of the encoder layers were concatenated (or added) to the same level decoding layers in order to help to reconstruct the image structure prior to segmentation. \n",
    "\n",
    "<img src=\"https://miro.medium.com/max/2640/1*J3t2b65ufsl1x6caf6GiBA.png\" width=\"600\" height=\"600\" />\n",
    "\n",
    "Although the original U-NET was proposed for 2D images, very rapidly 3D implementations arised [Çiçek et al. 2016](https://arxiv.org/abs/1606.06650). However, dealing with entire MRI volumes is not always possible, due to hardware memory limitations, etc... So, in most cases, patch-wise implementations of the same 3D models are used for different tasks such as brain tissue segmentation [Bernal et al. 2019](https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8754679). For our problem in hand, we are going to follow the same 3D patch-wise approach. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ein.tags": "worksheet-0",
    "slideshow": {
     "slide_type": "-"
    }
   },
   "source": [
    "Let's now define our first patch-wise `Unet` model in PyTorch:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "autoscroll": false,
    "ein.hycell": false,
    "ein.tags": "worksheet-0",
    "slideshow": {
     "slide_type": "-"
    }
   },
   "outputs": [],
   "source": [
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "\n",
    "class Unet(nn.Module):\n",
    "    \"\"\"\n",
    "    Basic U-net model\n",
    "    \"\"\"\n",
    "\n",
    "    def __init__(self, input_size, output_size):\n",
    "\n",
    "        super(Unet, self).__init__()\n",
    "\n",
    "        # conv1 down\n",
    "        self.conv1 = nn.Conv3d(in_channels=input_size,\n",
    "                               out_channels=32,\n",
    "                               kernel_size=3,\n",
    "                               padding=1)\n",
    "        # max-pool 1\n",
    "        self.pool1 = nn.Conv3d(in_channels=32,\n",
    "                               out_channels=32,\n",
    "                               kernel_size=2,\n",
    "                               stride=2)\n",
    "        # conv2 down\n",
    "        self.conv2 = nn.Conv3d(in_channels=32,\n",
    "                               out_channels=64,\n",
    "                               kernel_size=3,\n",
    "                               padding=1)\n",
    "        # max-pool 2\n",
    "        self.pool2 = nn.Conv3d(in_channels=64,\n",
    "                               out_channels=64,\n",
    "                               kernel_size=2,\n",
    "                               stride=2)\n",
    "        # conv3 down\n",
    "        self.conv3 = nn.Conv3d(in_channels=64,\n",
    "                               out_channels=128,\n",
    "                               kernel_size=3,\n",
    "                               padding=1)\n",
    "        # max-pool 3\n",
    "        self.pool3 = nn.Conv3d(in_channels=128,\n",
    "                               out_channels=128,\n",
    "                               kernel_size=2,\n",
    "                               stride=2)\n",
    "        # conv4 down (latent space)\n",
    "        self.conv4 = nn.Conv3d(in_channels=128,\n",
    "                               out_channels=256,\n",
    "                               kernel_size=3,\n",
    "                               padding=1)\n",
    "        # up-sample conv4\n",
    "        self.up1 = nn.ConvTranspose3d(in_channels=256,\n",
    "                                      out_channels=128,\n",
    "                                      kernel_size=2,\n",
    "                                      stride=2)        \n",
    "        # conv 5 (add up1 + conv3)\n",
    "        self.conv5 = nn.Conv3d(in_channels=128,\n",
    "                               out_channels=128,\n",
    "                               kernel_size=3,\n",
    "                               padding=1)\n",
    "        # up-sample conv5\n",
    "        self.up2 = nn.ConvTranspose3d(in_channels=128,\n",
    "                                      out_channels=64,\n",
    "                                      kernel_size=2,\n",
    "                                      stride=2)\n",
    "        # conv6 (add up2 + conv2) \n",
    "        self.conv6 = nn.Conv3d(in_channels=64,\n",
    "                               out_channels=64,\n",
    "                               kernel_size=3,\n",
    "                               padding=1)\n",
    "        # up 3\n",
    "        self.up3 = nn.ConvTranspose3d(in_channels=64,\n",
    "                                      out_channels=32,\n",
    "                                      kernel_size=2,\n",
    "                                      stride=2)\n",
    "        # conv7 (add up3 + conv1)\n",
    "        self.conv7 = nn.Conv3d(in_channels=32,\n",
    "                               out_channels=32,\n",
    "                               kernel_size=3,\n",
    "                               padding=1)\n",
    "        # conv8 (classification)\n",
    "        self.conv8 = nn.Conv3d(in_channels=32,\n",
    "                               out_channels=output_size,\n",
    "                               kernel_size=1)\n",
    "\n",
    "    def forward(self, x):\n",
    "\n",
    "        # encoder\n",
    "        x1 = F.relu(self.conv1(x))\n",
    "        x1p = self.pool1(x1)\n",
    "        x2 = F.relu(self.conv2(x1p))\n",
    "        x2p = self.pool2(x2)\n",
    "        x3 = F.relu(self.conv3(x2p))\n",
    "        x3p = self.pool3(x3)\n",
    "        \n",
    "        # latent space\n",
    "        x4 = F.relu(self.conv4(x3p))\n",
    "\n",
    "        # decoder\n",
    "        up1 = self.up1(x4)\n",
    "        x5 = F.relu(self.conv5(up1 + x3)) # look how layers are added :o\n",
    "        up2 = self.up2(x5)\n",
    "        x6 = F.relu(self.conv6(up2 + x2))\n",
    "        up3 = self.up3(x6)\n",
    "        x7 = F.relu(self.conv7(up3 + x1))\n",
    "        \n",
    "        # output layer (2 classes)\n",
    "        # we use a softmax layer to return probabilities for each class\n",
    "        out = F.softmax(self.conv8(x7), dim=1) \n",
    "        return out"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ein.tags": "worksheet-0",
    "slideshow": {
     "slide_type": "-"
    }
   },
   "source": [
    "# 2. Data preparation:\n",
    "---\n",
    "\n",
    "Let's now obtain some data to train and test the model. Download the MRI scans and move it to `data`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "autoscroll": false,
    "ein.hycell": false,
    "ein.tags": "worksheet-0",
    "slideshow": {
     "slide_type": "-"
    }
   },
   "outputs": [],
   "source": [
    "!mkdir data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "autoscroll": false,
    "ein.hycell": false,
    "ein.tags": "worksheet-0",
    "slideshow": {
     "slide_type": "-"
    }
   },
   "outputs": [],
   "source": [
    "%%capture\n",
    "!wget https://www.dropbox.com/s/cfhcbkxlu23nzi9/WMH2017_set.zip?dl=0 \n",
    "!unzip WMH2017_set.zip\\?dl\\=0\n",
    "!mv WMH2017 data/"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ein.tags": "worksheet-0",
    "slideshow": {
     "slide_type": "-"
    }
   },
   "source": [
    "We have stored a reduced set of the [MICCAI2017 WM Hyperintensity challenge](https://wmh.isi.uu.nl) at `data/WMH2017` and split it in two sets containing 20 training and 10 testing cases. Each case contains a `FLAIR` and a `T1` image, co-registered to the `FLAIR` space (3mm). Manual annotated masks for white matter intensities are also provided for each case. Manual annotations include white matter lesions `class=1` and other pathologies `class=2` (which are ignored). Let's see a case as example: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {
    "autoscroll": false,
    "ein.hycell": false,
    "ein.tags": "worksheet-0",
    "slideshow": {
     "slide_type": "-"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 648x648 with 25 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "mri_subject = ants.image_read('data/WMH2017/train/41/FLAIR_brain.nii.gz')\n",
    "lesion_mask = ants.image_read('data/WMH2017/train/41/wmh.nii.gz')\n",
    "mri_subject.plot(overlay=lesion_mask, nslices=25, axis=2, title='Subject 41: FLAIR + WMH')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2.1 Loading data efficiently:\n",
    "\n",
    "Similar to other *Deep Learning* toolkits, Pytorch also incorporates some objects to deal with data efficiently. In particular, we will make use of `Datasets` and `Dataloaders`:\n",
    "\n",
    "* `Datasets` objects are used to build customs datasets from our data, permitting us to manage training data very efficiently. For instance, imagine a custom dataset with hundreds of MRI subjects containing several input modalities. Loading and managing this amount of information may be tricky in some situations, so a custom `Dataset` that loads the data when necessary is convenient. \n",
    "\n",
    "\n",
    "* `Dataloaders` take `Datasets` as input and permits to automatically collate individual fetched data samples into batches via arguments. `Dataloaders` yield consecutive batch samples during the training loop, handling all the data requirements via the `Dataset` object. \n",
    "\n",
    "In order to speed up this tutorial, we are going to make use of some helper funcions to deal with MRI data and convert them into PyTorch `dataloaders` (see `utils.py` for more information). "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "Let's now train our `Unet` model using the `WMH2017` data. Before start training, several steps have to be taken into account: \n",
    "1. Split the available data into training, validation and test sets. \n",
    "2. Build the proper training data structures using the PyTorch `Dataset` and `Dataloader` objects.\n",
    "3. Build the training loop. \n",
    "\n",
    "With the aim  to unify all the hyper-parameters and training options, we will define a dictionary `options` that we will populate with new values: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "options = {}\n",
    "\n",
    "# training data path\n",
    "options['training_path'] = 'data/WMH2017/train'\n",
    "\n",
    "# testing data path \n",
    "options['test_path'] = 'data/WMH2017/test'\n",
    "\n",
    "# train/validation split percentage\n",
    "options['train_split']  = 0.3\n",
    "\n",
    "# input modality names\n",
    "options['input_data'] = ['FLAIR_brain.nii.gz', 'T1_brain.nii.gz']\n",
    "\n",
    "# ground-truth name\n",
    "options['gt_data'] = 'wmh.nii.gz'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2.2 Training generators: \n",
    "\n",
    "We are going to use the helper class `MRI_DataPatchLoader` from `utils.py` to generate a `Dataset` object based on our data. `MRI_DataPathLoader` requires an input dictionary with all image paths, image labels, and some other options for patch generation and normalization that we will cover in a few minutes. \n",
    "\n",
    "Additionally, the `MRI_DataPathLoader` also requires a third dictionary containing the paths a ROI mask from within the patch sampling should be performed. This mask is very handful to guide the sampling process, for instance to remove background voxels, deal with class imbalance etc...\n",
    "\n",
    "For our example, let's generate a ROI mask for each subject simply with the binary brain mask:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "options['roi'] = 'brainmask.nii.gz'\n",
    "\n",
    "def mask_image(im):\n",
    "    return (im > 0).astype('float32')\n",
    "\n",
    "for scan_id in os.listdir(options['training_path']):\n",
    "    scan = ants.image_read(os.path.join(options['training_path'], \n",
    "                                        scan_id, \n",
    "                                        'FLAIR_brain.nii.gz')) \n",
    "    brainmask = ants.image_clone(scan).apply(mask_image)\n",
    "    brainmask.to_filename(os.path.join(options['training_path'], \n",
    "                                       scan_id, \n",
    "                                       options['roi']))\n",
    "                                       \n",
    "for scan_id in os.listdir(options['test_path']):\n",
    "    scan = ants.image_read(os.path.join(options['test_path'], \n",
    "                                        scan_id, \n",
    "                                        'FLAIR_brain.nii.gz')) \n",
    "    brainmask = ants.image_clone(scan).apply(mask_image)\n",
    "    brainmask.to_filename(os.path.join(options['test_path'], \n",
    "                                       scan_id, \n",
    "                                       options['roi']))\n",
    "                                       "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, we can define the input dictionaries for training and validation: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "training_scans = os.listdir(options['training_path'])\n",
    "random.shuffle(training_scans)\n",
    "        \n",
    "# load training / validation data\n",
    "t_d = int(len(training_scans) * (1 - options['train_split']))\n",
    "training_data = training_scans[:t_d]\n",
    "validation_data = training_scans[t_d:]\n",
    "\n",
    "input_train_data = {scan: [os.path.join(options['training_path'], scan, d)\n",
    "                         for d in options['input_data']]\n",
    "                  for scan in training_data}\n",
    "\n",
    "input_train_labels = {scan: [os.path.join(options['training_path'], scan, options['gt_data'])] \n",
    "                      for scan in training_data}\n",
    "\n",
    "input_train_rois = {scan: [os.path.join(options['training_path'], scan, options['roi'])] \n",
    "                      for scan in training_data}\n",
    "\n",
    "input_val_data = {scan: [os.path.join(options['training_path'], scan, d)\n",
    "                         for d in options['input_data']]\n",
    "                  for scan in validation_data}\n",
    "\n",
    "input_val_labels = {scan: [os.path.join(options['training_path'], scan, options['gt_data'])] \n",
    "                    for scan in validation_data}\n",
    "\n",
    "input_val_rois = {scan: [os.path.join(options['training_path'], scan, options['roi'])] \n",
    "                      for scan in validation_data}\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Build the `Datasets` for training and validation data. Here, we need to define the patch size (in 3D), and how we are going to sample the images. Also consider image normalization (take it into account on testing also!). We [normalize](https://datascience.stackexchange.com/questions/32109/zero-mean-and-unit-variance) images with `mean=0` and `std=1`. In particular, we are going to use:\n",
    "\n",
    "* Patch size of `32x32x32`\n",
    "* Sampling step of `16x16x16`\n",
    "* Batch size of `32`\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training data: \n",
      "> DATA: Loaded scan 137 roi size: 377694 label_size:  2552\n",
      "> DATA: Loaded scan 63 roi size: 474500 label_size:  12146\n",
      "> DATA: Loaded scan 41 roi size: 506385 label_size:  12535\n",
      "> DATA: Loaded scan 17 roi size: 561936 label_size:  1598\n",
      "> DATA: Loaded scan 65 roi size: 499090 label_size:  16998\n",
      "> DATA: Loaded scan 0 roi size: 634893 label_size:  19371\n",
      "> DATA: Loaded scan 105 roi size: 404143 label_size:  3307\n",
      "> DATA: Loaded scan 23 roi size: 507669 label_size:  332\n",
      "> DATA: Loaded scan 109 roi size: 472917 label_size:  548\n",
      "> DATA: Loaded scan 68 roi size: 401678 label_size:  20068\n",
      "> DATA: Loaded scan 37 roi size: 494037 label_size:  27218\n",
      "> DATA: Loaded scan 61 roi size: 444999 label_size:  5641\n",
      "> DATA: Loaded scan 67 roi size: 397052 label_size:  14788\n",
      "> DATA: Loaded scan 106 roi size: 399971 label_size:  861\n",
      "> DATA: Training sample size: 1598\n",
      "Validation data: \n",
      "> DATA: Loaded scan 64 roi size: 484833 label_size:  6124\n",
      "> DATA: Loaded scan 132 roi size: 386572 label_size:  12381\n",
      "> DATA: Loaded scan 49 roi size: 450583 label_size:  9644\n",
      "> DATA: Loaded scan 25 roi size: 556481 label_size:  9404\n",
      "> DATA: Loaded scan 21 roi size: 614901 label_size:  9021\n",
      "> DATA: Loaded scan 51 roi size: 420407 label_size:  9283\n",
      "> DATA: Training sample size: 710\n"
     ]
    }
   ],
   "source": [
    "from utils import MRI_DataPatchLoader\n",
    "from torch.utils.data import DataLoader\n",
    "\n",
    "# additional options for patch size, sampling step, normalization, etc...\n",
    "options['patch_size'] = (32, 32, 32)\n",
    "options['sampling_step'] = (16, 16, 16)\n",
    "options['normalize'] = True \n",
    "options['batch_size'] = 32\n",
    "\n",
    "print('Training data: ')\n",
    "training_dataset = MRI_DataPatchLoader(input_data=input_train_data,\n",
    "                                       labels=input_train_labels,\n",
    "                                       rois=input_train_rois,\n",
    "                                       patch_size=options['patch_size'],\n",
    "                                       sampling_step=options['sampling_step'],\n",
    "                                       normalize=options['normalize'])\n",
    "\n",
    "training_dataloader = DataLoader(training_dataset, \n",
    "                                 batch_size=options['batch_size'],\n",
    "                                 shuffle=True)\n",
    "\n",
    "print('Validation data: ')\n",
    "validation_dataset = MRI_DataPatchLoader(input_data=input_val_data,\n",
    "                                        labels=input_val_labels,\n",
    "                                        rois=input_val_rois,\n",
    "                                        patch_size=options['patch_size'],\n",
    "                                        sampling_step=options['sampling_step'],\n",
    "                                        normalize=options['normalize'])\n",
    "\n",
    "validation_dataloader = DataLoader(validation_dataset, \n",
    "                                   batch_size=options['batch_size'],\n",
    "                                   shuffle=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3. Training the network:\n",
    "\n",
    "---\n",
    "\n",
    "\n",
    "Finally we can build our training loop to optimize the `Unet` model based on the same principles we covered in the first part of this tutorial. Basically, what we are going to do is to iterate the training dataset for a number of epochs, saving the weights of the model accordingly. \n",
    "\n",
    "For this first example, we are going to use the [cross-entropy](https://pytorch.org/docs/stable/nn.html#torch.nn.CrossEntropyLoss) loss and the [Adadelta](https://pytorch.org/docs/stable/optim.html#torch.optim.Adadelta) optimizer:\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1 train_loss 0.1482 train_acc 2.1052 val_loss 0.0353 val_acc 0.9900\n",
      "Epoch 2 train_loss 0.0627 train_acc 2.1503 val_loss 0.0238 val_acc 0.9924\n",
      "Epoch 3 train_loss 0.0464 train_acc 2.1534 val_loss 0.0192 val_acc 0.9931\n",
      "Epoch 4 train_loss 0.0385 train_acc 2.1550 val_loss 0.0176 val_acc 0.9933\n",
      "Epoch 5 train_loss 0.0337 train_acc 2.1562 val_loss 0.0170 val_acc 0.9934\n",
      "Epoch 6 train_loss 0.0320 train_acc 2.1566 val_loss 0.0181 val_acc 0.9930\n",
      "Epoch 7 train_loss 0.0307 train_acc 2.1569 val_loss 0.0163 val_acc 0.9935\n",
      "Epoch 8 train_loss 0.0300 train_acc 2.1571 val_loss 0.0162 val_acc 0.9935\n",
      "Epoch 21 train_loss 0.0238 train_acc 2.1587 val_loss 0.0140 val_acc 0.9942\n",
      "Epoch 22 train_loss 0.0235 train_acc 2.1587 val_loss 0.0141 val_acc 0.9941\n",
      "Epoch 23 train_loss 0.0231 train_acc 2.1589 val_loss 0.0168 val_acc 0.9934\n",
      "Epoch 24 train_loss 0.0230 train_acc 2.1588 val_loss 0.0146 val_acc 0.9939\n",
      "Epoch 25 train_loss 0.0234 train_acc 2.1587 val_loss 0.0138 val_acc 0.9943\n",
      "Epoch 26 train_loss 0.0223 train_acc 2.1591 val_loss 0.0139 val_acc 0.9942\n",
      "Epoch 27 train_loss 0.0220 train_acc 2.1591 val_loss 0.0140 val_acc 0.9941\n",
      "Epoch 28 train_loss 0.0218 train_acc 2.1592 val_loss 0.0135 val_acc 0.9942\n",
      "Epoch 29 train_loss 0.0216 train_acc 2.1593 val_loss 0.0133 val_acc 0.9944\n",
      "Epoch 30 train_loss 0.0207 train_acc 2.1595 val_loss 0.0135 val_acc 0.9943\n",
      "Epoch 31 train_loss 0.0209 train_acc 2.1594 val_loss 0.0134 val_acc 0.9944\n",
      "Epoch 32 train_loss 0.0217 train_acc 2.1592 val_loss 0.0132 val_acc 0.9944\n",
      "Epoch 33 train_loss 0.0217 train_acc 2.1592 val_loss 0.0156 val_acc 0.9939\n",
      "Epoch 34 train_loss 0.0210 train_acc 2.1594 val_loss 0.0137 val_acc 0.9941\n",
      "Epoch 35 train_loss 0.0206 train_acc 2.1595 val_loss 0.0134 val_acc 0.9943\n",
      "Epoch 36 train_loss 0.0206 train_acc 2.1595 val_loss 0.0145 val_acc 0.9939\n",
      "Epoch 37 train_loss 0.0198 train_acc 2.1597 val_loss 0.0137 val_acc 0.9943\n",
      "Epoch 38 train_loss 0.0203 train_acc 2.1596 val_loss 0.0149 val_acc 0.9936\n",
      "Epoch 39 train_loss 0.0216 train_acc 2.1592 val_loss 0.0134 val_acc 0.9943\n",
      "Epoch 40 train_loss 0.0202 train_acc 2.1596 val_loss 0.0140 val_acc 0.9943\n",
      "Epoch 41 train_loss 0.0196 train_acc 2.1598 val_loss 0.0142 val_acc 0.9941\n",
      "Epoch 42 train_loss 0.0191 train_acc 2.1599 val_loss 0.0130 val_acc 0.9946\n",
      "Epoch 43 train_loss 0.0204 train_acc 2.1595 val_loss 0.0132 val_acc 0.9946\n",
      "Epoch 44 train_loss 0.0202 train_acc 2.1596 val_loss 0.0129 val_acc 0.9945\n",
      "Epoch 45 train_loss 0.0189 train_acc 2.1600 val_loss 0.0155 val_acc 0.9936\n",
      "Epoch 46 train_loss 0.0186 train_acc 2.1601 val_loss 0.0147 val_acc 0.9938\n",
      "Epoch 47 train_loss 0.0193 train_acc 2.1599 val_loss 0.0128 val_acc 0.9945\n",
      "Epoch 48 train_loss 0.0186 train_acc 2.1601 val_loss 0.0129 val_acc 0.9947\n",
      "Epoch 49 train_loss 0.0242 train_acc 2.1585 val_loss 0.0128 val_acc 0.9946\n",
      "Epoch 50 train_loss 0.0183 train_acc 2.1602 val_loss 0.0127 val_acc 0.9946\n",
      "Epoch 51 train_loss 0.0181 train_acc 2.1602 val_loss 0.0135 val_acc 0.9943\n",
      "Epoch 52 train_loss 0.0183 train_acc 2.1601 val_loss 0.0131 val_acc 0.9945\n",
      "Epoch 53 train_loss 0.0179 train_acc 2.1603 val_loss 0.0133 val_acc 0.9945\n",
      "Epoch 54 train_loss 0.0188 train_acc 2.1600 val_loss 0.0133 val_acc 0.9944\n",
      "Epoch 55 train_loss 0.0181 train_acc 2.1602 val_loss 0.0124 val_acc 0.9948\n",
      "Epoch 56 train_loss 0.0178 train_acc 2.1603 val_loss 0.0128 val_acc 0.9947\n",
      "Epoch 57 train_loss 0.0181 train_acc 2.1602 val_loss 0.0148 val_acc 0.9939\n",
      "Epoch 58 train_loss 0.0177 train_acc 2.1603 val_loss 0.0127 val_acc 0.9946\n",
      "Epoch 59 train_loss 0.0173 train_acc 2.1604 val_loss 0.0171 val_acc 0.9930\n",
      "Epoch 60 train_loss 0.0174 train_acc 2.1604 val_loss 0.0127 val_acc 0.9946\n",
      "Epoch 61 train_loss 0.0188 train_acc 2.1600 val_loss 0.0132 val_acc 0.9945\n",
      "Epoch 62 train_loss 0.0172 train_acc 2.1604 val_loss 0.0133 val_acc 0.9946\n",
      "Epoch 63 train_loss 0.0173 train_acc 2.1604 val_loss 0.0131 val_acc 0.9944\n",
      "Epoch 64 train_loss 0.0185 train_acc 2.1601 val_loss 0.0133 val_acc 0.9944\n",
      "Epoch 65 train_loss 0.0164 train_acc 2.1607 val_loss 0.0133 val_acc 0.9945\n",
      "Epoch 66 train_loss 0.0177 train_acc 2.1603 val_loss 0.0126 val_acc 0.9947\n",
      "Epoch 67 train_loss 0.0167 train_acc 2.1606 val_loss 0.0124 val_acc 0.9948\n",
      "Epoch 68 train_loss 0.0177 train_acc 2.1603 val_loss 0.0129 val_acc 0.9947\n",
      "Epoch 69 train_loss 0.0162 train_acc 2.1608 val_loss 0.0129 val_acc 0.9946\n",
      "Epoch 70 train_loss 0.0165 train_acc 2.1606 val_loss 0.0131 val_acc 0.9945\n",
      "Epoch 71 train_loss 0.0213 train_acc 2.1593 val_loss 0.0125 val_acc 0.9947\n",
      "Epoch 72 train_loss 0.0165 train_acc 2.1607 val_loss 0.0129 val_acc 0.9946\n",
      "Epoch 73 train_loss 0.0162 train_acc 2.1608 val_loss 0.0130 val_acc 0.9947\n",
      "Epoch 74 train_loss 0.0162 train_acc 2.1607 val_loss 0.0133 val_acc 0.9944\n",
      "Epoch 75 train_loss 0.0205 train_acc 2.1595 val_loss 0.0126 val_acc 0.9947\n",
      "Epoch 76 train_loss 0.0158 train_acc 2.1609 val_loss 0.0130 val_acc 0.9946\n",
      "Epoch 77 train_loss 0.0158 train_acc 2.1609 val_loss 0.0133 val_acc 0.9944\n",
      "Epoch 78 train_loss 0.0161 train_acc 2.1608 val_loss 0.0151 val_acc 0.9939\n",
      "Epoch 79 train_loss 0.0155 train_acc 2.1610 val_loss 0.0138 val_acc 0.9943\n",
      "Epoch 80 train_loss 0.0160 train_acc 2.1608 val_loss 0.0130 val_acc 0.9947\n",
      "Epoch 81 train_loss 0.0158 train_acc 2.1609 val_loss 0.0133 val_acc 0.9945\n",
      "Epoch 82 train_loss 0.0158 train_acc 2.1609 val_loss 0.0142 val_acc 0.9942\n",
      "Epoch 83 train_loss 0.0158 train_acc 2.1609 val_loss 0.0133 val_acc 0.9945\n",
      "Epoch 84 train_loss 0.0149 train_acc 2.1612 val_loss 0.0129 val_acc 0.9946\n",
      "Epoch 85 train_loss 0.0161 train_acc 2.1608 val_loss 0.0131 val_acc 0.9945\n",
      "Epoch 86 train_loss 0.0159 train_acc 2.1609 val_loss 0.0154 val_acc 0.9937\n",
      "Epoch 87 train_loss 0.0155 train_acc 2.1610 val_loss 0.0130 val_acc 0.9948\n",
      "Epoch 88 train_loss 0.0151 train_acc 2.1611 val_loss 0.0135 val_acc 0.9944\n",
      "Epoch 89 train_loss 0.0152 train_acc 2.1611 val_loss 0.0138 val_acc 0.9946\n",
      "Epoch 90 train_loss 0.0156 train_acc 2.1609 val_loss 0.0129 val_acc 0.9947\n",
      "Epoch 91 train_loss 0.0301 train_acc 2.1581 val_loss 0.0125 val_acc 0.9946\n",
      "Epoch 92 train_loss 0.0167 train_acc 2.1606 val_loss 0.0124 val_acc 0.9947\n",
      "Epoch 93 train_loss 0.0157 train_acc 2.1609 val_loss 0.0138 val_acc 0.9942\n",
      "Epoch 94 train_loss 0.0153 train_acc 2.1610 val_loss 0.0144 val_acc 0.9940\n",
      "Epoch 95 train_loss 0.0156 train_acc 2.1609 val_loss 0.0132 val_acc 0.9946\n",
      "Epoch 96 train_loss 0.0146 train_acc 2.1613 val_loss 0.0136 val_acc 0.9944\n",
      "Epoch 97 train_loss 0.0153 train_acc 2.1611 val_loss 0.0135 val_acc 0.9946\n",
      "Epoch 98 train_loss 0.0158 train_acc 2.1609 val_loss 0.0133 val_acc 0.9946\n",
      "Epoch 99 train_loss 0.0146 train_acc 2.1613 val_loss 0.0137 val_acc 0.9944\n"
     ]
    }
   ],
   "source": [
    "from torch.optim import Adadelta\n",
    "\n",
    "# mkdir for stored models\n",
    "!mkdir models\n",
    "\n",
    "\n",
    "# Define the Unet model \n",
    "# 2 input channels (FLAIR and T1)\n",
    "# 2 output classes (healthy and wmh) (we ignore other pathologies)\n",
    "lesion_model = Unet(input_size=2, output_size=2)\n",
    "model_name = 'test_maia_wmh'\n",
    "\n",
    "# some training options\n",
    "options['gpu_use'] = True\n",
    "options['num_epochs'] = 100\n",
    "\n",
    "# define the torch.device\n",
    "device = torch.device('cuda') if options['gpu_use'] else torch.device('cpu')\n",
    "\n",
    "# define the optimizer\n",
    "optimizer = Adadelta(lesion_model.parameters())\n",
    "\n",
    "# send the model to the device\n",
    "lesion_model = lesion_model.to(device)\n",
    "\n",
    "# training loop\n",
    "training = True\n",
    "epoch = 1\n",
    "try:\n",
    "    while training:\n",
    "    \n",
    "        # epoch specific metrics\n",
    "        train_loss = 0\n",
    "        train_accuracy = 0\n",
    "        val_loss = 0\n",
    "        val_accuracy = 0\n",
    "        \n",
    "        # -----------------------------\n",
    "        # training samples\n",
    "        # -----------------------------\n",
    "        \n",
    "        # set the model into train mode\n",
    "        lesion_model.train()\n",
    "        for a, batch in enumerate(training_dataloader):\n",
    "                # process batches: each batch is composed by training (x) and labels (y)\n",
    "                # x = [32, 2, 32, 32, 32]\n",
    "                # y = [32, 1, 32, 32, 32]\n",
    "    \n",
    "                x = batch[0].to(device)\n",
    "                y = batch[1].to(device)\n",
    "                \n",
    "                # clear gradients\n",
    "                optimizer.zero_grad()\n",
    "                \n",
    "                # infer the current batch \n",
    "                pred = lesion_model(x)\n",
    "                \n",
    "                # compute the loss. \n",
    "                # we ignore the index=2\n",
    "                loss = F.cross_entropy(torch.log(torch.clamp(pred, 1E-7, 1.0)),\n",
    "                                       y.squeeze(dim=1).long(), ignore_index=2)\n",
    "                train_loss += loss.item()\n",
    "                \n",
    "                # backward loss and next step\n",
    "                loss.backward()\n",
    "                optimizer.step()\n",
    "\n",
    "                # compute the accuracy\n",
    "                pred = pred.max(1, keepdim=True)[1]\n",
    "                batch_accuracy = pred.eq(y.view_as(pred).long())\n",
    "                train_accuracy += (batch_accuracy.sum().item() / np.prod(y.shape))\n",
    "                \n",
    "                \n",
    "        # -----------------------------\n",
    "        # validation samples\n",
    "        # -----------------------------\n",
    "    \n",
    "        # set the model into train mode\n",
    "        lesion_model.eval()\n",
    "        for b, batch in enumerate(validation_dataloader):\n",
    "                x = batch[0].to(device)\n",
    "                y = batch[1].to(device)\n",
    "                \n",
    "                # infer the current batch \n",
    "                with torch.no_grad():\n",
    "                    pred = lesion_model(x)\n",
    "                \n",
    "                    # compute the loss. \n",
    "                    # we ignore the index=2\n",
    "                    loss = F.cross_entropy(torch.log(torch.clamp(pred, 1E-7, 1.0)),\n",
    "                                           y.squeeze(dim=1).long(), ignore_index=2)\n",
    "                    val_loss += loss.item()\n",
    "                \n",
    "                    # compute the accuracy \n",
    "                    pred = pred.max(1, keepdim=True)[1]\n",
    "                    batch_accuracy = pred.eq(y.view_as(pred).long())\n",
    "                    val_accuracy += batch_accuracy.sum().item() / np.prod(y.shape)\n",
    "                \n",
    "        \n",
    "        # compute mean metrics\n",
    "        train_loss /= (a + 1)\n",
    "        train_accuracy /= (a + 1)\n",
    "        val_loss /= (b + 1)\n",
    "        val_accuracy /= (b + 1)\n",
    "\n",
    "        print('Epoch {:d} train_loss {:.4f} train_acc {:.4f} val_loss {:.4f} val_acc {:.4f}'.format(\n",
    "            epoch, \n",
    "            train_loss, \n",
    "            train_accuracy,\n",
    "            val_loss,\n",
    "            val_accuracy))\n",
    "        # update epochs\n",
    "        epoch += 1\n",
    "        \n",
    "        # save weights\n",
    "        torch.save(lesion_model.state_dict(), \n",
    "                  os.path.join('models', 'model' + str(epoch) + '.pt'))\n",
    "\n",
    "        if epoch >= options['num_epochs']:\n",
    "            training = False\n",
    "except KeyboardInterrupt:\n",
    "    pass\n",
    "                        "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 4. Inference:\n",
    "\n",
    "---\n",
    "\n",
    "\n",
    "Once we have trained our first `Unet` model, we can infer some testing cases to see how well it performs. In contrast to training, we do not make use of data generators for infering lesions on testing scans. Again, we are  going to use some additional helper functions included in `utils.py` that will help us to extract the image patches to infer the model:\n",
    "\n",
    "* `get_inference_patches` will extract $[num_{samples}, 2, 32, 32, 32]$ patches from the input image along with their coordenates $(x,y,z)$. \n",
    "\n",
    "* `reconstruct_image` will take the predicted patches and reconstruct them back as an image using the coordinates as a  base. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "from utils import get_inference_patches, reconstruct_image\n",
    "\n",
    "scan_path = os.path.join(options['test_path'], '107')\n",
    "scan = ants.image_read(os.path.join(scan_path, 'FLAIR_brain.nii.gz'))\n",
    "\n",
    "infer_patches, coordenates = get_inference_patches(scan_path=scan_path,\n",
    "                                          input_data=options['input_data'],\n",
    "                                          roi=options['roi'],\n",
    "                                          patch_shape=options['patch_size'],\n",
    "                                          step=options['sampling_step'],\n",
    "                                          normalize=options['normalize'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We evaluate the model on the extracted patches. Note that inference is very similar to validation, but we store the results in the `lesion_out` array:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "lesion_out = np.zeros_like(infer_patches).astype('float32')\n",
    "batch_size = options['batch_size']\n",
    "\n",
    "# model\n",
    "lesion_model.eval()\n",
    "with torch.no_grad():\n",
    "    for b in range(0, len(lesion_out), batch_size):\n",
    "        x = torch.tensor(infer_patches[b:b+batch_size]).to(device)\n",
    "        pred = lesion_model(x)\n",
    "        # save the result back from GPU to CPU --> numpy\n",
    "        lesion_out[b:b+batch_size] = pred.cpu().numpy()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Finally, we reconstruct the image based on the inferred patches and binarize the output. Here by default we use threshold of $th=0.5$, but other can be considered:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/sergivalverde/w/maia_seminar/utils.py:461: RuntimeWarning: invalid value encountered in true_divide\n",
      "  out_image /= freq_count\n"
     ]
    }
   ],
   "source": [
    "# reconstruct image takes the inferred patches, the patches coordenates and the image size as inputs\n",
    "lesion_prob = reconstruct_image(lesion_out[:,1], \n",
    "                                coordenates, \n",
    "                                scan.shape)\n",
    "\n",
    "th = 0.5\n",
    "lesion_prob = (lesion_prob > th).astype('uint8')\n",
    "\n",
    "# save as nifti image is necessary\n",
    "lesion_img = ants.from_numpy(lesion_prob)\n",
    "lesion_img = ants.copy_image_info(scan, lesion_img)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 5. Evaluation:\n",
    "---\n",
    "\n",
    "In order to evaluate the results, we use standard metrics for lesion segmentation such as:\n",
    "* `DSC:` Dice **segmentation** coefficient \n",
    "* `TPF:` **Lesion-wise** true positive rate (Sensitivity) \n",
    "* `PPV:` **Lesion-wise** positive predictive value (PPV)(precision)\n",
    "\n",
    "Let's infer all the test images and evaluate the results in average:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/sergivalverde/w/maia_seminar/utils.py:461: RuntimeWarning: invalid value encountered in true_divide\n",
      "  out_image /= freq_count\n"
     ]
    },
    {
     "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>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>DSC</th>\n",
       "      <td>10.0</td>\n",
       "      <td>0.657255</td>\n",
       "      <td>0.196698</td>\n",
       "      <td>0.339001</td>\n",
       "      <td>0.486785</td>\n",
       "      <td>0.728163</td>\n",
       "      <td>0.795888</td>\n",
       "      <td>0.895906</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TPF</th>\n",
       "      <td>10.0</td>\n",
       "      <td>0.494027</td>\n",
       "      <td>0.090632</td>\n",
       "      <td>0.348624</td>\n",
       "      <td>0.457947</td>\n",
       "      <td>0.501969</td>\n",
       "      <td>0.560594</td>\n",
       "      <td>0.607477</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>PPV</th>\n",
       "      <td>10.0</td>\n",
       "      <td>0.798953</td>\n",
       "      <td>0.072415</td>\n",
       "      <td>0.694939</td>\n",
       "      <td>0.768016</td>\n",
       "      <td>0.788570</td>\n",
       "      <td>0.833935</td>\n",
       "      <td>0.955781</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     count      mean       std       min       25%       50%       75%  \\\n",
       "DSC   10.0  0.657255  0.196698  0.339001  0.486785  0.728163  0.795888   \n",
       "TPF   10.0  0.494027  0.090632  0.348624  0.457947  0.501969  0.560594   \n",
       "PPV   10.0  0.798953  0.072415  0.694939  0.768016  0.788570  0.833935   \n",
       "\n",
       "          max  \n",
       "DSC  0.895906  \n",
       "TPF  0.607477  \n",
       "PPV  0.955781  "
      ]
     },
     "execution_count": 95,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "from metrics import DSC_seg, TPF_det, PPV_det\n",
    "\n",
    "\n",
    "# obtain a list of test scans\n",
    "test_scans = os.listdir(options['test_path'])\n",
    "th = 0.5\n",
    "\n",
    "# iterate through the scans and evaluate the results\n",
    "metrics = np.zeros((len(test_scans), 3))\n",
    "\n",
    "for i, scan_name in enumerate(test_scans):\n",
    "    \n",
    "    scan_path = os.path.join(options['test_path'], scan_name)\n",
    "    scan = ants.image_read(os.path.join(scan_path, 'FLAIR_brain.nii.gz'))\n",
    "\n",
    "    infer_patches, coordenates = get_inference_patches(scan_path=scan_path,\n",
    "                                              input_data=options['input_data'],\n",
    "                                              roi=options['roi'],\n",
    "                                              patch_shape=options['patch_size'],\n",
    "                                              step=options['sampling_step'],\n",
    "                                              normalize=options['normalize'])\n",
    "\n",
    "    lesion_out = np.zeros_like(infer_patches).astype('float32')\n",
    "    batch_size = options['batch_size']\n",
    "\n",
    "    # model evaluation \n",
    "    lesion_model.eval()\n",
    "    with torch.no_grad():\n",
    "        for b in range(0, len(lesion_out), batch_size):\n",
    "            x = torch.tensor(infer_patches[b:b+batch_size]).to(device)\n",
    "            pred = lesion_model(x)\n",
    "            lesion_out[b:b+batch_size] = pred.cpu().numpy()\n",
    "        \n",
    "    # reconstruct image \n",
    "    lesion_prob = reconstruct_image(lesion_out[:,1], \n",
    "                                    coordenates, \n",
    "                                    scan.shape)\n",
    "            \n",
    "    # binarize the results\n",
    "    lesion_prob = (lesion_prob > th).astype('uint8')\n",
    "\n",
    "    # evaluate the results    \n",
    "    gt = ants.image_read(os.path.join(scan_path, 'wmh.nii.gz'))\n",
    "    dsc_metric = DSC_seg(gt.numpy() == 1, lesion_prob > 0)\n",
    "    tpf_metric = TPF_det(gt.numpy() == 1, lesion_prob > 0)\n",
    "    ppv_metric = PPV_det(gt.numpy() == 1, lesion_prob > 0)\n",
    "\n",
    "    metrics[i] = [dsc_metric, tpf_metric, ppv_metric]\n",
    "        \n",
    "    # print('SCAN:', scan_name, 'Dice: ', dsc_metric, 'TPF:', tpf_metric, 'PPV:', ppv_metric)\n",
    "\n",
    "# we use PANDAS to describe data :)\n",
    "m = pd.DataFrame(metrics, columns=['DSC', 'TPF', 'PPV'])\n",
    "m.describe().T"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 6. Conclusions:\n",
    "---\n",
    "\n",
    "The presented implementation is fairly basic given that the goal was just to introduce the U-NET model in PyTorch. During this tutorial, we did not cover most of the techniques and \"tricks\" applied in state-of-the-art methods, although these are easily hackable in our basic model. In particular, I would play with:\n",
    "\n",
    "* Updating the architecture, using residual layers, different non-linear activations, etc...\n",
    "* Early stopping.\n",
    "* Patch sampling: how to handle data imbalance. So far, our model was sampling the input image homogeneously without considering the exisiting differences in voxel size between classes. \n",
    "* Training losses: another way to handle data imbalance. There are plenty of losses that are designed to mitigate data imbalance [(see here)](https://github.com/kevinzakka/pytorch-goodies/blob/master/losses.py). \n",
    "* [Dropout](https://pytorch.org/docs/stable/nn.html?highlight=dropout#torch.nn.Dropout)  is a powerful technique to avoid overfitting. \n",
    "* [BatchNorm](https://pytorch.org/docs/stable/nn.html#batchnorm3d) is also a very interesting technique [(see also here)](https://mlexplained.com/2018/01/10/an-intuitive-explanation-of-why-batch-normalization-really-works-normalization-in-deep-learning-part-1/). \n",
    "* [Uncertainty measures](https://arxiv.org/abs/1808.01200).\n",
    "* [Deep supervision](https://arxiv.org/abs/1505.02496).\n",
    "* ...\n"
   ]
  }
 ],
 "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": "3.7.3"
  },
  "name": "MAIA seminar (Part 2).ipynb"
 },
 "nbformat": 4,
 "nbformat_minor": 2
}