![](\"https://www.mdpi.com/applsci/applsci-10-07274/article_deploy/html/images/applsci-10-07274-g001.png\")
![](\"https://1.bp.blogspot.com/-DjZT_TLYZok/XO3BYqpxCJI/AAAAAAAAEKM/BvV53klXaTUuQHCkOXZZGywRMdU9v9T_wCLcBGAs/s1600/image2.png\")
![](\"https://i.imgur.com/pzwzeQj.png\"/)
![](\"https://miro.medium.com/max/1400/1*8oE4jOMfOXeEzgsHjSB5ww.png\"/)
Diabetic retinopathy (DR) is one of the leading causes of vision loss. According to a recent study from International Diabetes Federation, the global prevalence of DR among the individuals with diabetes for the period from 2015 to 2019 was at more than 25%. According to the World Health Organization, more than 300 million people worldwide have diabetes, and the disease\nprevalence has been rising rapidly in developing countries.
\nEarly detection and treatment are crucial steps towards preventing DR. Currently, detecting DR is a time-consuming process. The screening procedure requires a trained clinical expert to examine the fundus photographs of the patient’s retina. This creates delays in diagnosis and treatment of the disease. Automated evaluation of retina photographs can speed up the efficiency and coverage of the DR screening programs. This is especially relevant for developing countries, which often lack qualified medical stuff to perform the diagnosis.
\nThe project aims at developing a deep learning model for predicting the severity of DR disease based on the patient’s retina photograph. Previous research has explored the usage of deep learning for detecting DR and concluded that convolutional neural networks (CNNs) have high potential in this task . The Asia Pacific Tele-Ophthalmology Society (APTOS) has launched two Kaggle competitions with a goal of promoting the use of deep learning for DR detection and boosting the development of automated detection systems.
\n\n1. main data set used for modeling and evaluation. This data set is provided by APTOS. It has been employed in the APTOS 2019 Blindness Detection competition on Kaggle and is available for the download at the competition website: https://www.kaggle.com/c/aptos2019-blindness-detection/data. The data set includes 3,662 labeled retina images of clinical patients. The images are taken using a fundus photography technique.
\n2. supplementary data set for pre-training. This data set features 35,126 retina images labeled by a clinician using the same scale as the main data set. The data set has been used in the 2015 Diabetic Retinopathy Detection competition and is available for the download at the corresponding website: https://www.kaggle.com/c/diabetic-retinopathy-detection/data.
","metadata":{}},{"cell_type":"markdown","source":"\n## 1.1 Import Packages","metadata":{}},{"cell_type":"code","source":"#importing libraries \nimport numpy as np\nimport pandas as pd\n\nfrom PIL import Image, ImageFile\nImageFile.LOAD_TRUNCATED_IMAGES = True\nimport cv2\n\nfrom tqdm import tqdm_notebook as tqdm\nfrom functools import partial\nimport scipy as sp\n\nimport random\nimport time\nimport sys\nimport os\n\nimport matplotlib.pyplot as plt\nimport seaborn as sns\n%matplotlib inline\n\nfrom sklearn import metrics\nfrom sklearn.metrics import confusion_matrix\nimport torch\nimport torchvision\n\nimport torch.nn as nn\nimport torch.optim as optim\nfrom torch.optim import lr_scheduler\nfrom torchvision import transforms, models, datasets\nfrom torch.utils.data import Dataset\nfrom torch.autograd import Variable\n\n!pip install efficientnet_pytorch\nfrom efficientnet_pytorch import EfficientNet\n\nimport warnings\nwarnings.filterwarnings('ignore')\n!mkdir models","metadata":{"execution":{"iopub.status.busy":"2024-12-17T18:26:10.540571Z","iopub.execute_input":"2024-12-17T18:26:10.540848Z","iopub.status.idle":"2024-12-17T18:26:26.519769Z","shell.execute_reply.started":"2024-12-17T18:26:10.540786Z","shell.execute_reply":"2024-12-17T18:26:26.518392Z"},"trusted":true},"outputs":[{"name":"stdout","text":"Collecting efficientnet_pytorch\n Downloading efficientnet_pytorch-0.7.1.tar.gz (21 kB)\n Preparing metadata (setup.py) ... \u001b[?25ldone\n\u001b[?25hRequirement already satisfied: torch in /opt/conda/lib/python3.7/site-packages (from efficientnet_pytorch) (1.11.0)\nRequirement already satisfied: typing-extensions in /opt/conda/lib/python3.7/site-packages (from torch->efficientnet_pytorch) (4.1.1)\nBuilding wheels for collected packages: efficientnet_pytorch\n Building wheel for efficientnet_pytorch (setup.py) ... \u001b[?25ldone\n\u001b[?25h Created wheel for efficientnet_pytorch: filename=efficientnet_pytorch-0.7.1-py3-none-any.whl size=16446 sha256=316152771bf1ac2d571c2567ee65488fa17008d0baa1f5001f2e7fe122518edf\n Stored in directory: /root/.cache/pip/wheels/0e/cc/b2/49e74588263573ff778da58cc99b9c6349b496636a7e165be6\nSuccessfully built efficientnet_pytorch\nInstalling collected packages: efficientnet_pytorch\nSuccessfully installed efficientnet_pytorch-0.7.1\n\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\u001b[33m\n\u001b[0m","output_type":"stream"}],"execution_count":1},{"cell_type":"markdown","source":"\n\n## 1.2 Create Classes and Function","metadata":{}},{"cell_type":"code","source":"# seed function\ndef seed_everything(seed = 23):\n # tests\n assert isinstance(seed, int), 'seed has to be an integer'\n \n # randomness\n os.environ['PYTHONHASHSEED'] = str(seed)\n random.seed(seed)\n np.random.seed(seed)\n torch.manual_seed(seed)\n torch.cuda.manual_seed(seed)\n torch.backends.cudnn.deterministic = True\n torch.backends.cudnn.benchmark = False","metadata":{"execution":{"iopub.status.busy":"2024-11-11T04:37:44.940852Z","iopub.execute_input":"2024-11-11T04:37:44.941653Z","iopub.status.idle":"2024-11-11T04:37:44.948956Z","shell.execute_reply.started":"2024-11-11T04:37:44.941618Z","shell.execute_reply":"2024-11-11T04:37:44.947922Z"},"trusted":true},"outputs":[],"execution_count":2},{"cell_type":"code","source":"image_size = 256\n#IMAGE PREPROCESSING\n\ndef prepare_image(path, \n sigmaX = 10, \n do_random_crop = False):\n \n '''\n Preprocess image\n '''\n \n # import image\n image = cv2.imread(path)\n image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)\n \n # perform smart crops\n image = crop_black(image, tol = 7)\n if do_random_crop == True:\n image = random_crop(image, size = (0.9, 1))\n \n # resize and color\n image = cv2.resize(image, (int(image_size), int(image_size)))\n image = cv2.addWeighted(image, 4, cv2.GaussianBlur(image, (0, 0), sigmaX), -4, 128)\n \n # circular crop\n image = circle_crop(image, sigmaX = sigmaX)\n\n # convert to tensor \n image = torch.tensor(image)\n image = image.permute(2, 1, 0)\n return image","metadata":{"execution":{"iopub.status.busy":"2024-12-17T18:26:39.290805Z","iopub.execute_input":"2024-12-17T18:26:39.291476Z","iopub.status.idle":"2024-12-17T18:26:39.299637Z","shell.execute_reply.started":"2024-12-17T18:26:39.291438Z","shell.execute_reply":"2024-12-17T18:26:39.298578Z"},"trusted":true},"outputs":[],"execution_count":2},{"cell_type":"code","source":"#CROP FUNCTIONS\n\ndef crop_black(img, \n tol = 7):\n \n '''\n Perform automatic crop of black areas\n '''\n \n if img.ndim == 2:\n mask = img > tol\n return img[np.ix_(mask.any(1),mask.any(0))]\n \n elif img.ndim == 3:\n gray_img = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)\n mask = gray_img > tol\n check_shape = img[:,:,0][np.ix_(mask.any(1),mask.any(0))].shape[0]\n \n if (check_shape == 0): \n return img \n else:\n img1 = img[:,:,0][np.ix_(mask.any(1),mask.any(0))]\n img2 = img[:,:,1][np.ix_(mask.any(1),mask.any(0))]\n img3 = img[:,:,2][np.ix_(mask.any(1),mask.any(0))]\n img = np.stack([img1, img2, img3], axis = -1)\n return img\n","metadata":{"execution":{"iopub.status.busy":"2024-12-17T18:26:42.575465Z","iopub.execute_input":"2024-12-17T18:26:42.575816Z","iopub.status.idle":"2024-12-17T18:26:42.583573Z","shell.execute_reply.started":"2024-12-17T18:26:42.575787Z","shell.execute_reply":"2024-12-17T18:26:42.582648Z"},"trusted":true},"outputs":[],"execution_count":3},{"cell_type":"code","source":"def circle_crop(img, \n sigmaX = 10): \n \n '''\n Perform circular crop around image center\n '''\n \n height, width, depth = img.shape\n \n largest_side = np.max((height, width))\n img = cv2.resize(img, (largest_side, largest_side))\n\n height, width, depth = img.shape\n \n x = int(width / 2)\n y = int(height / 2)\n r = np.amin((x,y))\n \n circle_img = np.zeros((height, width), np.uint8)\n cv2.circle(circle_img, (x,y), int(r), 1, thickness = -1)\n \n img = cv2.bitwise_and(img, img, mask = circle_img)\n return img \n","metadata":{"execution":{"iopub.status.busy":"2024-12-17T18:26:45.682534Z","iopub.execute_input":"2024-12-17T18:26:45.682893Z","iopub.status.idle":"2024-12-17T18:26:45.689887Z","shell.execute_reply.started":"2024-12-17T18:26:45.682863Z","shell.execute_reply":"2024-12-17T18:26:45.688845Z"},"trusted":true},"outputs":[],"execution_count":4},{"cell_type":"code","source":"def random_crop(img, \n size = (0.9, 1)):\n \n '''\n Random crop\n '''\n\n height, width, depth = img.shape\n \n cut = 1 - random.uniform(size[0], size[1])\n \n i = random.randint(0, int(cut * height))\n j = random.randint(0, int(cut * width))\n h = i + int((1 - cut) * height)\n w = j + int((1 - cut) * width)\n\n img = img[i:h, j:w, :] \n \n return img","metadata":{"execution":{"iopub.status.busy":"2024-12-17T18:27:07.091851Z","iopub.execute_input":"2024-12-17T18:27:07.092773Z","iopub.status.idle":"2024-12-17T18:27:07.098303Z","shell.execute_reply.started":"2024-12-17T18:27:07.092736Z","shell.execute_reply":"2024-12-17T18:27:07.09744Z"},"trusted":true},"outputs":[],"execution_count":5},{"cell_type":"code","source":"class EyeData(Dataset):\n \n # initialize\n def __init__(self, data, directory, transform = None, do_random_crop = True, itype = '.png'):\n self.data = data\n self.directory = directory\n self.transform = transform\n self.do_random_crop = do_random_crop\n self.itype = itype\n # length\n def __len__(self):\n return len(self.data)\n \n # get items \n def __getitem__(self, idx):\n img_name = os.path.join(self.directory, self.data.loc[idx, 'id_code'] + self.itype)\n image = prepare_image(img_name, do_random_crop = self.do_random_crop)\n image = self.transform(image)\n label = torch.tensor(self.data.loc[idx, 'diagnosis'])\n return {'image': image, 'label': label}","metadata":{"execution":{"iopub.status.busy":"2024-12-17T18:27:10.939812Z","iopub.execute_input":"2024-12-17T18:27:10.940455Z","iopub.status.idle":"2024-12-17T18:27:10.946953Z","shell.execute_reply.started":"2024-12-17T18:27:10.940421Z","shell.execute_reply":"2024-12-17T18:27:10.946026Z"},"trusted":true},"outputs":[],"execution_count":6},{"cell_type":"code","source":"class Data(Dataset):\n \n # initialize\n def __init__(self, data, directory, transform = None, do_random_crop = True, itype = '.png'):\n self.data = data\n self.directory = directory\n self.transform = transform\n self.do_random_crop = do_random_crop\n self.itype = itype\n # length\n def __len__(self):\n return len(self.data)\n \n # get items \n def __getitem__(self, idx):\n img_name = os.path.join(self.directory, self.data.loc[idx, 'id_code'] + self.itype)\n image = cv2.imread(img_name)\n image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)\n image = crop_black(image, tol = 7)\n image = cv2.resize(image, (int(image_size), int(image_size)))\n image = circle_crop(image, sigmaX = 10)\n image = torch.tensor(image)\n image = image.permute(2, 1, 0)\n image = self.transform(image)\n label = torch.tensor(self.data.loc[idx, 'diagnosis'])\n return {'image': image, 'label': label}","metadata":{"execution":{"iopub.status.busy":"2024-12-17T18:27:13.408471Z","iopub.execute_input":"2024-12-17T18:27:13.409391Z","iopub.status.idle":"2024-12-17T18:27:13.416511Z","shell.execute_reply.started":"2024-12-17T18:27:13.409355Z","shell.execute_reply":"2024-12-17T18:27:13.415662Z"},"trusted":true},"outputs":[],"execution_count":7},{"cell_type":"code","source":"def init_model(train= True, \n trn_layers = 2,\n model_name = 'enet_b7'):\n \n '''\n Initialize the model\n '''\n \n ### training modea\n if train == True:\n \n # load pre-trained model\n model = EfficientNet.from_pretrained('efficientnet-b7', num_classes = 5)\n model.load_state_dict(torch.load('../input/diabetic-retinopathy-pre-training/models/model_{}.bin'.format(model_name, 1))) \n \n # freeze first layers\n for child in list(model.children())[:-trn_layers]:\n for param in child.parameters():\n param.requires_grad = False\n \n \n #inference mode\n if train == False:\n \n # load pre-trained model\n model = EfficientNet.from_pretrained('efficientnet-b7', num_classes = 5)\n model.load_state_dict(torch.load('../input/diabetic-retinopathy-pre-training/models/model_{}.bin'.format(model_name, 1))) \n\n # freeze all layers\n for param in model.parameters():\n param.requires_grad = False\n \n \n ### return model\n return model","metadata":{"execution":{"iopub.status.busy":"2024-12-17T18:27:16.077163Z","iopub.execute_input":"2024-12-17T18:27:16.077788Z","iopub.status.idle":"2024-12-17T18:27:16.08435Z","shell.execute_reply.started":"2024-12-17T18:27:16.077758Z","shell.execute_reply":"2024-12-17T18:27:16.083395Z"},"trusted":true},"outputs":[],"execution_count":8},{"cell_type":"code","source":"#GPU CHECK\ntrain_on_gpu = torch.cuda.is_available()\nif not train_on_gpu:\n print('CUDA is not available. Training on CPU...')\n device = torch.device('cpu')\nelse:\n print('CUDA is available. Training on GPU...')\n device = torch.device('cuda:0')","metadata":{"execution":{"iopub.status.busy":"2024-12-17T18:27:20.55011Z","iopub.execute_input":"2024-12-17T18:27:20.550751Z","iopub.status.idle":"2024-12-17T18:27:20.646433Z","shell.execute_reply.started":"2024-12-17T18:27:20.550717Z","shell.execute_reply":"2024-12-17T18:27:20.645454Z"},"trusted":true},"outputs":[{"name":"stdout","text":"CUDA is available. Training on GPU...\n","output_type":"stream"}],"execution_count":9},{"cell_type":"code","source":"#RANDOMNESS\n\nseed = 23\nseed_everything(seed)","metadata":{"execution":{"iopub.status.busy":"2024-12-17T18:27:23.593579Z","iopub.execute_input":"2024-12-17T18:27:23.594477Z","iopub.status.idle":"2024-12-17T18:27:23.730041Z","shell.execute_reply.started":"2024-12-17T18:27:23.594441Z","shell.execute_reply":"2024-12-17T18:27:23.728854Z"},"trusted":true},"outputs":[{"traceback":["\u001b[0;31m---------------------------------------------------------------------------\u001b[0m","\u001b[0;31mNameError\u001b[0m Traceback (most recent call last)","\u001b[0;32m/tmp/ipykernel_27/2570612563.py\u001b[0m in \u001b[0;36m