{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"# *Multi Step Approach*\n\n- Use the Task 1 labeled data to create a model to predict tumor presence or absence on axial T1wCE images\n- The tumor slice predictor is then used on Task 2 data after converting the data to 128x128x128 volumes to account for orientation differences in the dicom data\n- Positive tumor slices are used to create a tiled image\n- The image is composed of 3 channels in order to use info from three scan types at once\n- Finally, MGMT status is trained with the 3 channel, tumor positive tiles and predicted on the test set\n\n\n## *Building on the prior work from:*\n\n- \"Load Task 1 Dataset & Comparison w/ Task 2 Dataset\" by Darien Schettler  https://www.kaggle.com/dschettler8845/load-task-1-dataset-comparison-w-task-2-dataset which was used to create the data in '../input/task1-output/data'\n\n- \"Normalized Voxels: Align Planes and Crop\" by yu4u  https://www.kaggle.com/ren4yu/normalized-voxels-align-planes-and-crop/data\n\n\nThis approach is new because the tiled approach flattens the 3D volume into a 2D array of images. Only the images with tumor are used and the RGB channels of the jpg images are used to store info from three different scan types.\n\nThis gets about as bad results as everything else :-) but it's a different way to look at the data. \n\n","metadata":{}},{"cell_type":"markdown","source":"## Example of a Tiled image with one scan type in each of 3 channels","metadata":{}},{"cell_type":"markdown","source":"# 4x4 \n\n![00022.jpg](attachment:f58436f2-2145-4559-ba3f-7d19339a4543.jpg)\n\n# 5x5\n\n![00011.jpg](attachment:20b715ac-1ed6-4951-bbc7-7ddd1d660dc9.jpg)","metadata":{},"attachments":{"f58436f2-2145-4559-ba3f-7d19339a4543.jpg":{"image/jpeg":"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"},"20b715ac-1ed6-4951-bbc7-7ddd1d660dc9.jpg":{"image/jpeg":"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"}}},{"cell_type":"markdown","source":"## **Imports**","metadata":{}},{"cell_type":"code","source":"import sys\n!{sys.executable} -m pip install ../input/pydicomfastaihelper/torch-1.9.0-cp37-cp37m-manylinux1_x86_64.whl -q\n!{sys.executable} -m pip install ../input/pydicomfastaihelper/torchvision-0.10.0-cp37-cp37m-manylinux1_x86_64.whl -q\n!{sys.executable} -m pip install ../input/pydicomfastaihelper/fastcore-1.3.26-py3-none-any.whl -q\n!{sys.executable} -m pip install ../input/pydicomfastaihelper/fastai-2.5.2-py3-none-any.whl -q","metadata":{"execution":{"iopub.status.busy":"2021-09-11T02:25:24.2552Z","iopub.execute_input":"2021-09-11T02:25:24.255569Z","iopub.status.idle":"2021-09-11T02:25:59.852688Z","shell.execute_reply.started":"2021-09-11T02:25:24.255531Z","shell.execute_reply":"2021-09-11T02:25:59.851566Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# from Awsaf pydicom_conda_helper\n!conda install '/kaggle/input/pydicomfastaihelper/libjpeg-turbo-2.1.0-h7f98852_0.tar.bz2' -c conda-forge -y\n!conda install '/kaggle/input/pydicomfastaihelper/libgcc-ng-9.3.0-h2828fa1_19.tar.bz2' -c conda-forge -y\n!conda install '/kaggle/input/pydicomfastaihelper/gdcm-2.8.9-py37h500ead1_1.tar.bz2' -c conda-forge -y\n!conda install '/kaggle/input/pydicomfastaihelper/conda-4.10.1-py37h89c1867_0.tar.bz2' -c conda-forge -y\n!conda install '/kaggle/input/pydicomfastaihelper/certifi-2020.12.5-py37h89c1867_1.tar.bz2' -c conda-forge -y\n!conda install '/kaggle/input/pydicomfastaihelper/openssl-1.1.1k-h7f98852_0.tar.bz2' -c conda-forge -y","metadata":{"execution":{"iopub.status.busy":"2021-09-11T02:23:17.088299Z","iopub.execute_input":"2021-09-11T02:23:17.088627Z","iopub.status.idle":"2021-09-11T02:24:23.299313Z","shell.execute_reply.started":"2021-09-11T02:23:17.088591Z","shell.execute_reply":"2021-09-11T02:24:23.298007Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import os\nimport PIL\nimport cv2\nimport tarfile\nimport numpy as np\nimport pydicom\nimport pandas as pd\nfrom glob import glob\nimport nibabel as nib\nfrom PIL import Image\nimport matplotlib.pyplot as plt\nfrom scipy import ndimage as ndi\nfrom tqdm.notebook import tqdm\nimport shutil\n\nimport random\nrandom.seed(42)\n\nfrom pathlib import Path","metadata":{"execution":{"iopub.status.busy":"2021-09-16T00:50:56.850789Z","iopub.execute_input":"2021-09-16T00:50:56.851082Z","iopub.status.idle":"2021-09-16T00:50:56.857453Z","shell.execute_reply.started":"2021-09-16T00:50:56.851055Z","shell.execute_reply":"2021-09-16T00:50:56.856485Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Identify Slices with Tumor Labels","metadata":{}},{"cell_type":"markdown","source":"- Label 1 shows necrotic and non-enhancing tumor core and label 4 shows contrast enhancing tumor. We'll combine those and call those slices tumor positive. Other slices are negative.\n\n- Only one volume is used for tumor slice prediction - T1wCE. The contrast enhanced study is used to take advantage of label 4.\n\n- The volumes are large, so we will only use every third slice. There are also many slices with minimal or no brain, only slices with more than the minimum number of non-zero pixels will be included.\n\n- Create a csv for training labels identifying slices with and without tumor.\n\n- Save chosen slices to jpg for training.","metadata":{}},{"cell_type":"code","source":"def normalize_contrast(voxel):\n    if voxel.sum() == 0:\n        return voxel\n    voxel = voxel - np.min(voxel)\n    voxel = voxel / np.max(voxel)\n    voxel = (voxel * 255).astype(np.uint8)\n    return voxel\n\ndef load_nii(study_id, scan_type):\n    fn = f'{data_folder}/BraTS2021_{study_id}/BraTS2021_{study_id}_{scan_type}.nii.gz'\n    if Path(fn).is_file():\n        nii = nib.load(fn).get_fdata()\n        nii = normalize_contrast(nii)\n        return nii\n    else:\n        print(fn)\n        return None","metadata":{"execution":{"iopub.status.busy":"2021-09-16T00:51:00.414197Z","iopub.execute_input":"2021-09-16T00:51:00.414468Z","iopub.status.idle":"2021-09-16T00:51:00.423090Z","shell.execute_reply.started":"2021-09-16T00:51:00.414442Z","shell.execute_reply":"2021-09-16T00:51:00.421824Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"min_brain = 1000\ndata_folder = '../input/task1-output/data'\n\nif not os.path.isdir(\"/kaggle/working/slices\"):\n    os.makedirs(\"/kaggle/working/slices\", exist_ok=True)\n    \ncollect = []\ncase_dirs = Path(data_folder).iterdir()\ncase_dirs = [_ for _ in case_dirs]\n\nfor case in tqdm(case_dirs, total =len(case_dirs)):\n    seg_files = [_ for _ in case.glob(\"*_seg.nii.gz\")]\n    \n    if len(seg_files) > 0:\n        seg = nib.load(seg_files[0]).get_fdata()\n        #Labels 1 and 4 are tumor labels\n        labels = [1* (((seg[:,:,o] == 1).sum() + (seg[:,:,o] == 4).sum()) > 0) for o in range(seg.shape[-1])]\n\n        study_id= str(case).split('_')[1]\n        nii = load_nii(study_id, 't1ce')\n\n        for i in range(0,nii.shape[-1],3):\n            if (nii[:,:,i] > 0).sum() > min_brain:\n                save_fn = f'slices/{study_id}_{i}.jpg'\n                collect.append({'filename':save_fn, 'study_id': study_id, 'slice': i, 'has_tumor': labels[i]})\n                #Rotate\n                rot = np.rot90(nii[:,:,i],3)\n                img = Image.fromarray(rot)\n                img = img.convert(\"L\")\n                img.save(save_fn)\n    \n\ncollect = pd.DataFrame(collect)\ncollect.to_csv('./tumor_slices_brats_axial_thirds.csv',index=False)\ncollect.head()","metadata":{"execution":{"iopub.status.busy":"2021-09-10T19:07:40.744988Z","iopub.execute_input":"2021-09-10T19:07:40.745361Z","iopub.status.idle":"2021-09-10T19:16:16.391686Z","shell.execute_reply.started":"2021-09-10T19:07:40.745327Z","shell.execute_reply":"2021-09-10T19:16:16.390703Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"collect.has_tumor.value_counts()","metadata":{"execution":{"iopub.status.busy":"2021-09-10T19:16:16.393082Z","iopub.execute_input":"2021-09-10T19:16:16.393449Z","iopub.status.idle":"2021-09-10T19:16:16.40188Z","shell.execute_reply.started":"2021-09-10T19:16:16.393401Z","shell.execute_reply":"2021-09-10T19:16:16.400848Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Train on Tumor Slices from Part 1","metadata":{}},{"cell_type":"code","source":"#!pip install fastai --upgrade -q","metadata":{"execution":{"iopub.status.busy":"2021-09-10T19:16:16.403261Z","iopub.execute_input":"2021-09-10T19:16:16.403652Z","iopub.status.idle":"2021-09-10T19:16:24.47018Z","shell.execute_reply.started":"2021-09-10T19:16:16.403617Z","shell.execute_reply":"2021-09-10T19:16:24.46919Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"models_data_root = Path('models/')\nmodels_data_root.mkdir(parents=True, exist_ok=True)\n\ncwd = '/kaggle/working/'","metadata":{"execution":{"iopub.status.busy":"2021-09-16T00:51:11.471936Z","iopub.execute_input":"2021-09-16T00:51:11.472246Z","iopub.status.idle":"2021-09-16T00:51:11.477029Z","shell.execute_reply.started":"2021-09-16T00:51:11.472215Z","shell.execute_reply":"2021-09-16T00:51:11.476346Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from fastai.vision.all import *\n\nbs = 64\n\ndf = collect[['filename', 'has_tumor']]\n\ndls = ImageDataLoaders.from_df(df, path=cwd, bs=bs, item_tfms=Resize(224), batch_tfms=[ *aug_transforms()])\ndls.vocab","metadata":{"execution":{"iopub.status.busy":"2021-09-11T02:27:26.127627Z","iopub.execute_input":"2021-09-11T02:27:26.128399Z","iopub.status.idle":"2021-09-11T02:27:28.533068Z","shell.execute_reply.started":"2021-09-11T02:27:26.128335Z","shell.execute_reply":"2021-09-11T02:27:28.531575Z"},"collapsed":true,"jupyter":{"outputs_hidden":true},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"learn = cnn_learner(dls, resnet50,loss_func=LabelSmoothingCrossEntropyFlat(), metrics=[accuracy])\nlearn.fine_tune(20)","metadata":{"execution":{"iopub.status.busy":"2021-09-10T19:16:30.071883Z","iopub.execute_input":"2021-09-10T19:16:30.072249Z","iopub.status.idle":"2021-09-10T19:46:51.489929Z","shell.execute_reply.started":"2021-09-10T19:16:30.072212Z","shell.execute_reply":"2021-09-10T19:46:51.488768Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"tumor_slice_model_name = Path.joinpath(models_data_root, 'tumor_slice.pkl')\nlearn.export(tumor_slice_model_name)","metadata":{"execution":{"iopub.status.busy":"2021-09-10T19:46:51.491691Z","iopub.execute_input":"2021-09-10T19:46:51.492079Z","iopub.status.idle":"2021-09-10T19:46:51.931778Z","shell.execute_reply.started":"2021-09-10T19:46:51.492039Z","shell.execute_reply":"2021-09-10T19:46:51.930812Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#remove training folder\nshutil.rmtree('slices') ","metadata":{"execution":{"iopub.status.busy":"2021-09-10T19:46:51.933031Z","iopub.execute_input":"2021-09-10T19:46:51.933366Z","iopub.status.idle":"2021-09-10T19:46:53.397826Z","shell.execute_reply.started":"2021-09-10T19:46:51.93333Z","shell.execute_reply":"2021-09-10T19:46:53.396869Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Voxel class","metadata":{}},{"cell_type":"code","source":"# Voxel from dicom from \"Normalized Voxels: Align Planes and Crop\" by yu4u\n    # https://www.kaggle.com/ren4yu/normalized-voxels-align-planes-and-crop/data\n    # https://www.kaggle.com/arnabs007/part-1-rsna-miccai-btrc-understanding-the-data\n    # https://www.kaggle.com/davidbroberts/determining-mr-image-planes\n    \nclass Voxel:\n    def __init__(self,study_id, dicom_folder, save_data_folder, icon_size):\n\n        self.study_id=study_id\n        self.dicom_folder=dicom_folder\n        self.save_data_folder=save_data_folder\n        self.icon_size=icon_size\n\n    def get_voxel_data_root(self):\n        return Path(self.save_data_folder).joinpath(f'voxels_{self.icon_size}')\n\n    def get_voxel_prepared(self):\n        #/kaggle/working/data/voxels_128/{study_id}/{scan_type}.npy\n        fn = f'{self.get_voxel_data_root()}/{self.study_id}/{Path(self.dicom_folder).parts[-1]}.npy'\n        if os.path.isfile(fn) == False:\n            self.create_and_save_voxels()\n        voxel = np.load(fn)\n        return voxel\n\n    def get_image_plane(self, data):\n        x1, y1, _, x2, y2, _ = [round(j) for j in data.ImageOrientationPatient]\n        cords = [x1, y1, x2, y2]\n\n        if cords == [1, 0, 0, 0]:\n            return 'Coronal'\n        elif cords == [1, 0, 0, 1]:\n            return 'Axial'\n        elif cords == [0, 1, 0, 0]:\n            return 'Sagittal'\n        else:\n            return 'Unknown'\n        \n    def get_dicom_paths(self):\n        dicom_paths = sorted(self.dicom_folder.glob(\"*.dcm\"), key=lambda x: int(x.stem.split(\"-\")[-1]))\n        return dicom_paths\n\n    def get_voxel(self):\n        imgs = []\n        positions = []\n        dcm_paths = self.get_dicom_paths()\n        for dcm_path in dcm_paths:\n            img = pydicom.dcmread(str(dcm_path))\n            imgs.append(img.pixel_array)\n            positions.append(img.ImagePositionPatient)\n            \n        plane = self.get_image_plane(img)\n        voxel = np.stack(imgs)\n        \n        # reorder planes if needed and rotate voxel\n        if plane == \"Coronal\":\n            if positions[0][1] < positions[-1][1]:\n                voxel = voxel[::-1]\n                #print(f\"{study_id} {scan_type} {plane} reordered\")\n            voxel = voxel.transpose((1, 0, 2))\n        elif plane == \"Sagittal\":\n            if positions[0][0] < positions[-1][0]:\n                voxel = voxel[::-1]\n                #print(f\"{study_id} {scan_type} {plane} reordered\")\n            voxel = voxel.transpose((1, 2, 0))\n            voxel = np.rot90(voxel, 2, axes=(1, 2))\n        elif plane == \"Axial\":\n            if positions[0][2] > positions[-1][2]:\n                voxel = voxel[::-1]\n                #print(f\"{study_id} {scan_type} {plane} reordered\")\n            voxel = np.rot90(voxel, 2)\n        else:\n            if positions[0][2] > positions[-1][2]:\n                voxel = voxel[::-1]\n                #print(f\"{study_id} {scan_type} {plane} reordered\")\n            voxel = np.rot90(voxel, 2)\n            #raise ValueError(f\"Unknown plane {plane}\")\n        return voxel, plane\n\n    def normalize_contrast(self, voxel):\n        if voxel.sum() == 0:\n            return voxel\n        voxel = voxel - np.min(voxel)\n        voxel = voxel / np.max(voxel)\n        voxel = (voxel * 255).astype(np.uint8)\n        return voxel\n\n    def crop_voxel(self, voxel):\n        if voxel.sum() == 0:\n            return voxel\n        keep = (voxel.mean(axis=(0, 1)) > 0)\n        voxel = voxel[:, :, keep]\n        keep = (voxel.mean(axis=(0, 2)) > 0)\n        voxel = voxel[:, keep]\n        keep = (voxel.mean(axis=(1, 2)) > 0)\n        voxel = voxel[keep]\n        return voxel\n\n    def resize_voxel(self, voxel):\n        sz = int(self.icon_size)\n        output = np.zeros((sz, sz, sz), dtype=np.uint8)\n\n        if np.argmax(voxel.shape) == 0:\n            for i, s in enumerate(np.linspace(0, voxel.shape[0] - 1, sz)):\n                output[i] = cv2.resize(voxel[int(s)], (sz, sz))\n        elif np.argmax(voxel.shape) == 1:\n            for i, s in enumerate(np.linspace(0, voxel.shape[1] - 1, sz)):\n                output[:, i] = cv2.resize(voxel[:, int(s)], (sz, sz))\n        elif np.argmax(voxel.shape) == 2:\n            for i, s in enumerate(np.linspace(0, voxel.shape[2] - 1, sz)):\n                output[:, :, i] = cv2.resize(voxel[:, :, int(s)], (sz, sz))\n\n        return output\n\n    def create_voxel(self):\n        voxel, _ = self.get_voxel()\n        voxel = self.normalize_contrast(voxel)\n        voxel = self.crop_voxel(voxel)\n        voxel = self.resize_voxel(voxel)\n        return voxel\n\n    def create_and_save_voxels(self):\n        p = self.get_voxel_data_root().joinpath(self.study_id)\n        p.mkdir(parents=True, exist_ok=True)\n        #create and save voxels\n        fn = f'{self.get_voxel_data_root()}/{self.study_id}/{Path(self.dicom_folder).parts[-1]}.npy'\n        \n        if Path(fn).is_file() == False:\n            voxel = self.create_voxel()\n            np.save(fn, voxel)","metadata":{"execution":{"iopub.status.busy":"2021-09-16T00:51:20.059292Z","iopub.execute_input":"2021-09-16T00:51:20.059723Z","iopub.status.idle":"2021-09-16T00:51:20.091161Z","shell.execute_reply.started":"2021-09-16T00:51:20.059683Z","shell.execute_reply":"2021-09-16T00:51:20.090554Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# specific function to get ids\ndef get_study_ids(dicom_data_root):\n    study_ids = dicom_data_root.iterdir()\n    return [o.name for o in study_ids]\n","metadata":{"execution":{"iopub.status.busy":"2021-09-16T00:51:21.681577Z","iopub.execute_input":"2021-09-16T00:51:21.681984Z","iopub.status.idle":"2021-09-16T00:51:21.686680Z","shell.execute_reply.started":"2021-09-16T00:51:21.681955Z","shell.execute_reply":"2021-09-16T00:51:21.685380Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Create voxels for T1w, FLAIR, T1wCE","metadata":{}},{"cell_type":"code","source":"scan_types= ['T1w','FLAIR','T1wCE']\n# Size of one image/tile in tiles\nicon_size=128\n# Tiles is an n x n array but can be made larger\nicons_per_side=5\n\norientation='axial'\n#dataset = 'train'\n\ntrain_dicom_data_root = Path(f\"../input/rsna-miccai-brain-tumor-radiogenomic-classification/train\")\ntest_dicom_data_root = Path(f\"../input/rsna-miccai-brain-tumor-radiogenomic-classification/test\")\n\nsave_data_folder=Path('data/')\nsave_data_folder.mkdir(parents=True, exist_ok=True)\n\nslices_data_root= Path('slices/')\nslices_data_root.mkdir(parents=True, exist_ok=True)\n\nvoxel_data_root = Path(save_data_folder).joinpath(f'voxels_{icon_size}')\nvoxel_data_root.mkdir(parents=True, exist_ok=True)\n\ntiles_data_root = Path('tiles/')\ntiles_data_root.mkdir(parents=True, exist_ok=True)","metadata":{"execution":{"iopub.status.busy":"2021-09-16T00:51:37.946692Z","iopub.execute_input":"2021-09-16T00:51:37.946997Z","iopub.status.idle":"2021-09-16T00:51:37.958508Z","shell.execute_reply.started":"2021-09-16T00:51:37.946968Z","shell.execute_reply":"2021-09-16T00:51:37.957497Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def create_voxels_for_dataset(dicom_data_root):\n    study_ids = get_study_ids(dicom_data_root)\n    print('Creating voxels - ', scan_types)\n    for study_id in tqdm(study_ids):\n        for scan_type in scan_types:\n            #create voxels\n            dicom_folder = Path(f'{dicom_data_root}/{study_id}/{scan_type}')\n            if dicom_folder.is_dir():\n                v = Voxel(study_id, dicom_folder, save_data_folder, icon_size)\n\n                p = v.get_voxel_data_root().joinpath(study_id)\n                p.mkdir(parents=True, exist_ok=True)\n                    #create and save voxels\n                fn = f'{p}/{scan_type}.npy'\n                #print(fn)\n                if Path(fn).is_file() == False:\n                    voxel = v.create_voxel()\n                    np.save(fn, voxel)","metadata":{"execution":{"iopub.status.busy":"2021-09-16T00:51:42.274898Z","iopub.execute_input":"2021-09-16T00:51:42.275758Z","iopub.status.idle":"2021-09-16T00:51:42.284414Z","shell.execute_reply.started":"2021-09-16T00:51:42.275690Z","shell.execute_reply":"2021-09-16T00:51:42.283082Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"create_voxels_for_dataset(train_dicom_data_root)","metadata":{"execution":{"iopub.status.busy":"2021-09-10T19:46:53.468093Z","iopub.execute_input":"2021-09-10T19:46:53.468668Z","iopub.status.idle":"2021-09-10T20:27:06.27233Z","shell.execute_reply.started":"2021-09-10T19:46:53.468632Z","shell.execute_reply":"2021-09-10T20:27:06.271511Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Use model to evaluate training slices on Task 2 - T1wCE","metadata":{}},{"cell_type":"code","source":"# Create slices\ndef create_slices(dicom_folder):\n    #Load voxel\n    if dicom_folder.is_dir():\n        study_id = dicom_folder.parts[-2]\n        # Load T1wCE voxel\n        v = Voxel(study_id, dicom_folder, save_data_folder, icon_size)\n        voxel = v.get_voxel_prepared()\n        \n        # Create directory for study_id\n        if not os.path.isdir(f\"/kaggle/working/slices/{study_id}\"):\n            os.makedirs(f\"/kaggle/working/slices/{study_id}\", exist_ok=True)\n    \n        # Convert to jpg\n        step = 3\n        if voxel.shape[0] < 100:\n            step = 2\n        for i in range(0,voxel.shape[0],step):\n            if (voxel[i] > 0).sum() > min_brain:\n                im = Image.fromarray(np.array(np.flipud(voxel[i])))\n                save_fn = f'/kaggle/working/slices/{study_id}/{study_id}_{i}.jpg'\n                im.save(save_fn)\n                \ndef create_slices_for_dataset(dicom_data_root, scan_type = 'T1wCE'):\n    study_ids = get_study_ids(dicom_data_root)\n    for study_id in tqdm(study_ids):\n        dicom_folder = Path(f'{dicom_data_root}/{study_id}/{scan_type}')\n        create_slices(dicom_folder)","metadata":{"execution":{"iopub.status.busy":"2021-09-16T00:51:45.124942Z","iopub.execute_input":"2021-09-16T00:51:45.125247Z","iopub.status.idle":"2021-09-16T00:51:45.135216Z","shell.execute_reply.started":"2021-09-16T00:51:45.125218Z","shell.execute_reply":"2021-09-16T00:51:45.134562Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"create_slices_for_dataset(train_dicom_data_root)","metadata":{"execution":{"iopub.status.busy":"2021-09-10T20:52:58.339201Z","iopub.execute_input":"2021-09-10T20:52:58.339524Z","iopub.status.idle":"2021-09-10T20:53:12.977009Z","shell.execute_reply.started":"2021-09-10T20:52:58.339495Z","shell.execute_reply":"2021-09-10T20:53:12.975894Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Predict tumor slices\n#load model\nlearn = load_learner(tumor_slice_model_name)\n\ndef predict(image_paths):\n    dl = learn.dls.test_dl(image_paths)\n    preds, _ = learn.get_preds(dl=dl)\n    class_idxs = preds.argmax(dim=1)\n    predictions =  [learn.dls.vocab[c] for c in class_idxs]\n    confidence = preds.max(dim=1)[0]\n    return predictions, confidence\n\ndef predict_dataset(dicom_data_root, scan_type = 'T1wCE'):\n    collect = []\n    study_ids = get_study_ids(dicom_data_root)\n\n    for idx, study_id in enumerate(study_ids):\n        if idx%250==0 and idx > 0:\n            print(idx)\n            temp = pd.DataFrame(collect)\n            temp.to_csv(f'predict_slices_{dicom_data_root.parts[-1]}_{idx}.csv', index=False)\n\n        #get image_paths\n        study_folder = Path(f\"/kaggle/working/slices/{study_id}\")\n        if study_folder.is_dir():\n            image_paths = study_folder.glob('*.jpg')\n            image_paths = [_ for _ in image_paths]\n            #predict\n            predictions, confidence = predict(image_paths)\n            tmp = []\n            for i,image_path in enumerate(image_paths):\n                slice_number = image_path.stem.split('_')[-1]\n                tmp.append({'filename':image_path,'scan_type':scan_type, 'study_id': study_id, 'slice': int(slice_number), 'has_tumor': predictions[i], 'confidence':float(confidence[i])})\n\n            collect.extend(tmp)\n    if len(collect) == 0:\n        print(\"No predictions\")\n        return pd.DataFrame(columns=['filename','scan_type', 'study_id', 'slice', 'has_tumor', 'confidence'])\n    else:\n        return pd.DataFrame(collect)","metadata":{"execution":{"iopub.status.busy":"2021-09-10T22:12:02.964363Z","iopub.execute_input":"2021-09-10T22:12:02.964743Z","iopub.status.idle":"2021-09-10T22:12:03.095809Z","shell.execute_reply.started":"2021-09-10T22:12:02.96471Z","shell.execute_reply":"2021-09-10T22:12:03.094872Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df = predict_dataset(train_dicom_data_root)\ndf.to_csv(f'predict_slices_train.csv', index=False)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df.head()","metadata":{"execution":{"iopub.status.busy":"2021-09-10T21:57:12.821917Z","iopub.execute_input":"2021-09-10T21:57:12.822254Z","iopub.status.idle":"2021-09-10T21:57:12.83235Z","shell.execute_reply.started":"2021-09-10T21:57:12.822222Z","shell.execute_reply":"2021-09-10T21:57:12.831254Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Tiles","metadata":{}},{"cell_type":"markdown","source":"## Use positive predictions to create tiles","metadata":{}},{"cell_type":"code","source":"def get_tumor_slices(study_id, df):\n    studydf = df[df.study_id == study_id].sort_values('confidence',ascending=False).head(icons_per_side**2)\n    if studydf.shape[0] == 0:\n        #each voxel has 128 slices\n        return list(range(15,icon_size-15))\n    return studydf.slice.unique()\n\ndef get_idxs(study_id, df):\n    slices = get_tumor_slices(study_id, df)\n    if len(slices) >= icons_per_side**2:\n        return random.sample(list(slices), icons_per_side**2)\n    else:\n        #fill in tiles with duplicates\n        return random.choices(slices, k=icons_per_side**2)","metadata":{"execution":{"iopub.status.busy":"2021-09-16T00:51:58.588729Z","iopub.execute_input":"2021-09-16T00:51:58.589744Z","iopub.status.idle":"2021-09-16T00:51:58.597056Z","shell.execute_reply.started":"2021-09-16T00:51:58.589708Z","shell.execute_reply":"2021-09-16T00:51:58.596416Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"class Tiles:\n    def __init__(self,study_id, input_array, tiles_data_root, icon_size, icons_per_side):\n   \n        self.study_id=study_id\n        self.input_array=input_array\n        self.tiles_data_root=tiles_data_root\n        self.icon_size=icon_size\n        self.icons_per_side=icons_per_side\n      \n   \n    def create_tile_channel(self, i):\n        inputs = self.input_array[i]\n        tile = Image.new(\"RGB\", (self.icons_per_side * self.icon_size, self.icons_per_side * self.icon_size))\n        for index, im in enumerate(inputs):\n            img = Image.fromarray(im)\n            y = index // self.icons_per_side * self.icon_size\n            x = index % self.icons_per_side * self.icon_size\n            w, h = self.icon_size, self.icon_size\n            box = (int(x), int(y), int(x + w), int(y + h))\n            tile.paste(img, box)\n        return tile.split()[0]\n\n    def create_tiles(self):\n        channels = []\n        for i in range(len(self.input_array)): \n            if len(self.input_array) == 0:\n                return None\n            channel = self.create_tile_channel(i)\n            channels.append(channel)\n        if len(channels) >=3:\n            return Image.merge('RGB',channels[:3])\n        elif len(channels) == 2:\n            return Image.merge('RGB',(channels[0],channels[1],channels[0]))\n        elif len(channels) ==1 :\n            return Image.merge('RGB',(channels[0],channels[0],channels[0]))\n\n    def get_save_filename(self):\n        save_dir = self.tiles_data_root\n        save_dir.mkdir(parents=True, exist_ok=True)\n        fn = Path(save_dir).joinpath(f'{self.study_id}.jpg')\n        return str(fn)\n\n    def create_and_save_tiles(self):\n        img = self.create_tiles()\n        if img is not None:\n            fn = self.get_save_filename()\n            img.save(fn)","metadata":{"execution":{"iopub.status.busy":"2021-09-16T00:51:59.334910Z","iopub.execute_input":"2021-09-16T00:51:59.335249Z","iopub.status.idle":"2021-09-16T00:51:59.352600Z","shell.execute_reply.started":"2021-09-16T00:51:59.335216Z","shell.execute_reply":"2021-09-16T00:51:59.351223Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def create_tile(df, study_id):\n    input_array = []\n    #use same idxs for all 3 channels, use positive slices\n    idxs = get_idxs(study_id, df[(df.study_id == study_id) & (df.has_tumor == 1)])\n    #create 3 channel input array from each of the 3 scan_types  ['T1w','FLAIR','T1wCE']\n    for scan_type in scan_types:\n        #get voxel\n        fn = f'{voxel_data_root}/{study_id}/{scan_type}.npy'\n        if Path(fn).is_file():\n            voxel = np.load(fn)\n            #axial\n            voxel = np.array([np.flipud(voxel[i]) for i in range(voxel.shape[0])])\n            inputs = [voxel[idx] for idx in idxs]\n            input_array.append(inputs)\n        else:\n            print(fn)\n    #create tile\n    tile = Tiles(study_id, input_array, tiles_data_root, icon_size, icons_per_side)\n    tile.create_and_save_tiles()\n    \n","metadata":{"execution":{"iopub.status.busy":"2021-09-16T00:52:01.757695Z","iopub.execute_input":"2021-09-16T00:52:01.758026Z","iopub.status.idle":"2021-09-16T00:52:01.767361Z","shell.execute_reply.started":"2021-09-16T00:52:01.757994Z","shell.execute_reply":"2021-09-16T00:52:01.766183Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# not using Pool\ndef create_tiles_for_dataset(df, dicom_data_root):\n    study_ids = get_study_ids(dicom_data_root)\n    for study_id in tqdm(study_ids):\n        create_tile(df, study_id)\n\ncreate_tiles_for_dataset(df, train_dicom_data_root)","metadata":{"execution":{"iopub.status.busy":"2021-09-16T00:52:14.625514Z","iopub.execute_input":"2021-09-16T00:52:14.626615Z","iopub.status.idle":"2021-09-16T00:52:14.631740Z","shell.execute_reply.started":"2021-09-16T00:52:14.626556Z","shell.execute_reply":"2021-09-16T00:52:14.630746Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#Using Pool\n# import multiprocessing as mp\n# pool = mp.Pool(mp.cpu_count())\n\n# study_ids = get_study_ids(train_dicom_data_root)\n\n# # Step 1: Init multiprocessing.Pool()\n# pool = mp.Pool(mp.cpu_count())\n\n# # Step 2: `pool.apply` the `howmany_within_range()`\n# results = pool.map(create_tile, [study_id for study_id in study_ids])\n\n# # Step 3: Don't forget to close\n# pool.close()","metadata":{"execution":{"iopub.status.busy":"2021-09-16T00:53:45.211960Z","iopub.execute_input":"2021-09-16T00:53:45.212312Z","iopub.status.idle":"2021-09-16T00:53:45.218980Z","shell.execute_reply.started":"2021-09-16T00:53:45.212275Z","shell.execute_reply":"2021-09-16T00:53:45.217683Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Train MGMT on Tiles","metadata":{}},{"cell_type":"code","source":"# Load Competition Training Dataframe\na = pd.read_csv(\"/kaggle/input/rsna-miccai-brain-tumor-radiogenomic-classification/train_labels.csv\", dtype={'BraTS21ID': str, 'MGMT_value': int})\na = a.rename({'BraTS21ID':'study_id'}, axis=1)\n\nremove = ['00109', '00123', '00709']\na = a[~a.study_id.isin(remove)]\n\na['filename'] = [f'{tiles_data_root}/{study_id}.jpg' for _,study_id in a.study_id.iteritems()]\na =  a[['filename', 'MGMT_value']]\na.head()","metadata":{"execution":{"iopub.status.busy":"2021-09-10T22:00:23.223858Z","iopub.execute_input":"2021-09-10T22:00:23.224216Z","iopub.status.idle":"2021-09-10T22:00:23.245103Z","shell.execute_reply.started":"2021-09-10T22:00:23.224185Z","shell.execute_reply":"2021-09-10T22:00:23.244358Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"bs = 32\n\ndls = ImageDataLoaders.from_df(a, path=cwd, bs=bs, item_tfms=Resize(224), batch_tfms=[ *aug_transforms(mult=1.0, do_flip=True, flip_vert=False, max_rotate=10.0, min_zoom=1.0, max_zoom=1.1, max_lighting=0.2, max_warp=0.05, p_affine=0.1, p_lighting=0.75, xtra_tfms=None, size=None, mode='bilinear', pad_mode='reflection', align_corners=True, batch=False, min_scale=1.0)])\ndls.vocab","metadata":{"execution":{"iopub.status.busy":"2021-09-10T22:00:24.378859Z","iopub.execute_input":"2021-09-10T22:00:24.3792Z","iopub.status.idle":"2021-09-10T22:00:24.494807Z","shell.execute_reply.started":"2021-09-10T22:00:24.379169Z","shell.execute_reply":"2021-09-10T22:00:24.493754Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"dls.show_batch()","metadata":{"execution":{"iopub.status.busy":"2021-09-10T22:00:26.815748Z","iopub.execute_input":"2021-09-10T22:00:26.816079Z","iopub.status.idle":"2021-09-10T22:00:27.649596Z","shell.execute_reply.started":"2021-09-10T22:00:26.816048Z","shell.execute_reply":"2021-09-10T22:00:27.647035Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"learn = cnn_learner(dls, resnet50, loss_func=LabelSmoothingCrossEntropy(), metrics=[accuracy, RocAucBinary()])\nlearn.fine_tune(50)","metadata":{"execution":{"iopub.status.busy":"2021-09-10T22:00:35.551499Z","iopub.execute_input":"2021-09-10T22:00:35.551852Z","iopub.status.idle":"2021-09-10T22:00:49.113297Z","shell.execute_reply.started":"2021-09-10T22:00:35.55182Z","shell.execute_reply":"2021-09-10T22:00:49.112467Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"mgmt_model_name = Path.joinpath(models_data_root, 'miccai.pkl')\nlearn.export(mgmt_model_name)","metadata":{"execution":{"iopub.status.busy":"2021-09-10T22:35:54.422406Z","iopub.execute_input":"2021-09-10T22:35:54.422774Z","iopub.status.idle":"2021-09-10T22:35:54.741535Z","shell.execute_reply.started":"2021-09-10T22:35:54.422743Z","shell.execute_reply":"2021-09-10T22:35:54.740599Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#cleanup voxels, slices, tiles for training data\nfor dir in ['slices','data','tiles']:\n    path = os.path.join(cwd, dir)\n    # removing directory\n    if Path(path).is_dir():\n        shutil.rmtree(path) ","metadata":{"execution":{"iopub.status.busy":"2021-09-12T23:44:47.168832Z","iopub.execute_input":"2021-09-12T23:44:47.169084Z","iopub.status.idle":"2021-09-12T23:44:47.174201Z","shell.execute_reply.started":"2021-09-12T23:44:47.169058Z","shell.execute_reply":"2021-09-12T23:44:47.173402Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Prepare Test set","metadata":{}},{"cell_type":"markdown","source":"## Create Test set Tiles","metadata":{}},{"cell_type":"code","source":"#reset directories\nsave_data_folder=Path('data/')\nsave_data_folder.mkdir(parents=True, exist_ok=True)\n\nslices_data_root= Path('slices/')\nslices_data_root.mkdir(parents=True, exist_ok=True)\n\nvoxel_data_root = Path(save_data_folder).joinpath(f'voxels_{icon_size}')\nvoxel_data_root.mkdir(parents=True, exist_ok=True)\n\ntiles_data_root = Path('tiles/')\ntiles_data_root.mkdir(parents=True, exist_ok=True)","metadata":{"execution":{"iopub.status.busy":"2021-09-13T19:08:31.622961Z","iopub.execute_input":"2021-09-13T19:08:31.623504Z","iopub.status.idle":"2021-09-13T19:08:31.632121Z","shell.execute_reply.started":"2021-09-13T19:08:31.623466Z","shell.execute_reply":"2021-09-13T19:08:31.631219Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Dicom -> Voxels -> Slices -> Tiles","metadata":{}},{"cell_type":"code","source":"#create voxels\nprint('Creating voxels...')\ncreate_voxels_for_dataset(test_dicom_data_root)\n#create slices\nprint('Creating slices...')\ncreate_slices_for_dataset(test_dicom_data_root)\n#predict on slices to get those with tumor\nprint('Evaluating slices...')\nlearn = load_learner(tumor_slice_model_name)\ndf = predict_dataset(test_dicom_data_root)\n#create tiles\nprint('Creating tiles...')\ncreate_tiles_for_dataset(df, test_dicom_data_root)\nprint('Done')","metadata":{"execution":{"iopub.status.busy":"2021-09-10T22:01:30.464532Z","iopub.execute_input":"2021-09-10T22:01:30.464871Z","iopub.status.idle":"2021-09-10T22:07:39.551191Z","shell.execute_reply.started":"2021-09-10T22:01:30.46484Z","shell.execute_reply":"2021-09-10T22:07:39.549948Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Predict MGMT_value on Test set Tiles","metadata":{}},{"cell_type":"code","source":"learn = load_learner(mgmt_model_name)\n\nimage_paths = tiles_data_root.glob('*.jpg')\nimage_paths = [_ for _ in image_paths]\n\ndl = learn.dls.test_dl(image_paths)\npreds, _ = learn.get_preds(dl=dl)\nclass_idxs = preds.argmax(dim=1)\npredictions =  [learn.dls.vocab[c] for c in class_idxs]\nconfidence = preds.max(dim=1)[0]\n\nfinal = []\nfor i,image_path in enumerate(image_paths):\n    study_id = image_path.stem\n    final.append({'study_id': study_id, 'prediction': predictions[i], 'confidence':confidence[i]})\n\nfinal = pd.DataFrame(final)\nfinal.head()","metadata":{"execution":{"iopub.status.busy":"2021-09-10T22:36:17.986476Z","iopub.execute_input":"2021-09-10T22:36:17.98683Z","iopub.status.idle":"2021-09-10T22:36:34.076157Z","shell.execute_reply.started":"2021-09-10T22:36:17.986802Z","shell.execute_reply":"2021-09-10T22:36:34.07505Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"final.prediction.value_counts()","metadata":{"execution":{"iopub.status.busy":"2021-09-10T22:36:38.896644Z","iopub.execute_input":"2021-09-10T22:36:38.897003Z","iopub.status.idle":"2021-09-10T22:36:38.906982Z","shell.execute_reply.started":"2021-09-10T22:36:38.896967Z","shell.execute_reply":"2021-09-10T22:36:38.904586Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"### Cleanup","metadata":{}},{"cell_type":"code","source":"#cleanup voxels, slices, tiles\ndr = '/kaggle/working'\n# path\nfor dir in ['slices','data','tiles']:\n    path = os.path.join(dr, dir)\n    # removing directory\n    if Path(path).is_dir():\n        shutil.rmtree(path) ","metadata":{"execution":{"iopub.status.busy":"2021-09-10T22:36:44.462849Z","iopub.execute_input":"2021-09-10T22:36:44.463225Z","iopub.status.idle":"2021-09-10T22:36:44.628161Z","shell.execute_reply.started":"2021-09-10T22:36:44.463193Z","shell.execute_reply":"2021-09-10T22:36:44.627221Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Submission file","metadata":{}},{"cell_type":"code","source":"final = final[['study_id', 'prediction']]\nfinal.columns = ['BraTS21ID','MGMT_value']\nfinal.to_csv('submission.csv', index=False)","metadata":{"execution":{"iopub.status.busy":"2021-09-10T22:36:46.284658Z","iopub.execute_input":"2021-09-10T22:36:46.284994Z","iopub.status.idle":"2021-09-10T22:36:46.294637Z","shell.execute_reply.started":"2021-09-10T22:36:46.284962Z","shell.execute_reply":"2021-09-10T22:36:46.293838Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}