{"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":"3.10.12"},"kaggle":{"accelerator":"gpu","dataSources":[{"sourceId":71549,"databundleVersionId":8561470,"sourceType":"competition"},{"sourceId":992,"sourceType":"modelInstanceVersion","modelInstanceId":846,"modelId":101}],"dockerImageVersionId":30823,"isInternetEnabled":false,"language":"python","sourceType":"notebook","isGpuEnabled":true},"papermill":{"default_parameters":{},"duration":1513.513706,"end_time":"2024-12-18T23:30:52.064621","environment_variables":{},"exception":null,"input_path":"__notebook__.ipynb","output_path":"__notebook__.ipynb","parameters":{},"start_time":"2024-12-18T23:05:38.550915","version":"2.6.0"}},"nbformat_minor":4,"nbformat":4,"cells":[{"id":"c7e248bf-a120-4fd8-bf2f-c1823f33891b","cell_type":"markdown","source":"https://www.kaggle.com/code/brahimztrk15/notebookbe8a4e1bc9 with confusion matrix ","metadata":{}},{"id":"b3dcae93","cell_type":"code","source":"# This Python 3 environment comes with many helpful analytics libraries installed\n# It is defined by the kaggle/python Docker image: https://github.com/kaggle/docker-python\n# For example, here's several helpful packages to load\n\nimport numpy as np # linear algebra\nimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n\n# Input data files are available in the read-only \"../input/\" directory\n# For example, running this (by clicking run or pressing Shift+Enter) will list all files under the input directory\n\nimport os\ncounter = 0  # Sayaç başlat\nfor dirname, _, filenames in os.walk('/kaggle/input'):\n    for filename in filenames:\n        print(os.path.join(dirname, filename))\n        counter += 1\n        if counter == 15:  # 5 dosya yazdırdıktan sonra dur\n            break\n    if counter == 15:  # İç döngü kırıldığında dış döngüyü de kır\n        break\n\n# You can write up to 20GB to the current directory (/kaggle/working/) that gets preserved as output when you create a version using \"Save & Run All\" \n# You can also write temporary files to /kaggle/temp/, but they won't be saved outside of the current session","metadata":{"_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","execution":{"iopub.status.busy":"2025-01-09T00:19:36.915274Z","iopub.execute_input":"2025-01-09T00:19:36.915477Z","iopub.status.idle":"2025-01-09T00:19:38.479798Z","shell.execute_reply.started":"2025-01-09T00:19:36.915457Z","shell.execute_reply":"2025-01-09T00:19:38.478962Z"},"papermill":{"duration":2.123612,"end_time":"2024-12-18T23:05:43.119961","exception":false,"start_time":"2024-12-18T23:05:40.996349","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"ad66157f","cell_type":"code","source":"import seaborn as sns\n\nimport matplotlib.pyplot as plt\nimport os\nimport time\nimport numpy as np\nimport glob\nimport json\nimport collections\nimport torch\nimport torch.nn as nn\n\nimport pydicom as dicom\nimport matplotlib.patches as patches\n\nfrom matplotlib import animation, rc\nimport pandas as pd\n\nimport pydicom as dicom # dicom\nimport pydicom\nfrom pydicom.pixel_data_handlers.util import apply_voi_lut","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:19:38.480773Z","iopub.execute_input":"2025-01-09T00:19:38.481258Z","iopub.status.idle":"2025-01-09T00:19:42.665559Z","shell.execute_reply.started":"2025-01-09T00:19:38.481220Z","shell.execute_reply":"2025-01-09T00:19:42.664902Z"},"papermill":{"duration":5.369569,"end_time":"2024-12-18T23:05:48.500684","exception":false,"start_time":"2024-12-18T23:05:43.131115","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"666a13a8","cell_type":"code","source":"# read data\ntrain_path = '/kaggle/input/rsna-2024-lumbar-spine-degenerative-classification/'\n\ntrain  = pd.read_csv(train_path + 'train.csv')\nlabel = pd.read_csv(train_path + 'train_label_coordinates.csv')\ntrain_desc  = pd.read_csv(train_path + 'train_series_descriptions.csv')\ntest_desc   = pd.read_csv(train_path + 'test_series_descriptions.csv')\nsub         = pd.read_csv(train_path + 'sample_submission.csv')\nlen(test_desc) #number of test_description.csv rows ","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:19:42.666268Z","iopub.execute_input":"2025-01-09T00:19:42.666629Z","iopub.status.idle":"2025-01-09T00:19:42.806367Z","shell.execute_reply.started":"2025-01-09T00:19:42.666604Z","shell.execute_reply":"2025-01-09T00:19:42.805693Z"},"papermill":{"duration":0.182016,"end_time":"2024-12-18T23:05:48.693173","exception":false,"start_time":"2024-12-18T23:05:48.511157","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"87eecc32","cell_type":"code","source":"test_desc.head(5)","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:19:42.808369Z","iopub.execute_input":"2025-01-09T00:19:42.808627Z","iopub.status.idle":"2025-01-09T00:19:42.822616Z","shell.execute_reply.started":"2025-01-09T00:19:42.808606Z","shell.execute_reply":"2025-01-09T00:19:42.821737Z"},"papermill":{"duration":0.025559,"end_time":"2024-12-18T23:05:48.729456","exception":false,"start_time":"2024-12-18T23:05:48.703897","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"e4c7f9da","cell_type":"code","source":"train_desc.head(5)","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:19:42.823782Z","iopub.execute_input":"2025-01-09T00:19:42.824051Z","iopub.status.idle":"2025-01-09T00:19:42.840030Z","shell.execute_reply.started":"2025-01-09T00:19:42.824023Z","shell.execute_reply":"2025-01-09T00:19:42.839349Z"},"papermill":{"duration":0.019985,"end_time":"2024-12-18T23:05:48.759925","exception":false,"start_time":"2024-12-18T23:05:48.739940","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"f4402489","cell_type":"code","source":"train.head(5)","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:19:42.840743Z","iopub.execute_input":"2025-01-09T00:19:42.840989Z","iopub.status.idle":"2025-01-09T00:19:42.869096Z","shell.execute_reply.started":"2025-01-09T00:19:42.840956Z","shell.execute_reply":"2025-01-09T00:19:42.868381Z"},"papermill":{"duration":0.034305,"end_time":"2024-12-18T23:05:48.804261","exception":false,"start_time":"2024-12-18T23:05:48.769956","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"9cd2d3ac","cell_type":"code","source":"# Function to generate image paths based on directory structure\ndef generate_image_paths(df, data_dir):\n    image_paths = []\n    for study_id, series_id in zip(df['study_id'], df['series_id']):\n        study_dir = os.path.join(data_dir, str(study_id))\n        series_dir = os.path.join(study_dir, str(series_id))\n        images = os.listdir(series_dir)\n        image_paths.extend([os.path.join(series_dir, img) for img in images])\n    return image_paths\n\n# Generate image paths for train and test data\ntrain_image_paths = generate_image_paths(train_desc, f'{train_path}/train_images')\ntest_image_paths = generate_image_paths(test_desc, f'{train_path}/test_images')","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:19:42.869824Z","iopub.execute_input":"2025-01-09T00:19:42.870106Z","iopub.status.idle":"2025-01-09T00:20:29.406060Z","shell.execute_reply.started":"2025-01-09T00:19:42.870076Z","shell.execute_reply":"2025-01-09T00:20:29.405350Z"},"papermill":{"duration":29.629979,"end_time":"2024-12-18T23:06:18.444467","exception":false,"start_time":"2024-12-18T23:05:48.814488","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"78f04cd8","cell_type":"code","source":"len(train_desc)","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:29.406827Z","iopub.execute_input":"2025-01-09T00:20:29.407086Z","iopub.status.idle":"2025-01-09T00:20:29.412041Z","shell.execute_reply.started":"2025-01-09T00:20:29.407064Z","shell.execute_reply":"2025-01-09T00:20:29.411299Z"},"papermill":{"duration":0.018278,"end_time":"2024-12-18T23:06:18.474134","exception":false,"start_time":"2024-12-18T23:06:18.455856","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"f836d685","cell_type":"code","source":"len(train_image_paths)","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:29.412817Z","iopub.execute_input":"2025-01-09T00:20:29.413109Z","iopub.status.idle":"2025-01-09T00:20:29.430659Z","shell.execute_reply.started":"2025-01-09T00:20:29.413080Z","shell.execute_reply":"2025-01-09T00:20:29.430018Z"},"papermill":{"duration":0.018133,"end_time":"2024-12-18T23:06:18.503231","exception":false,"start_time":"2024-12-18T23:06:18.485098","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"34d623b5","cell_type":"code","source":"# Define function to reshape a single row of the DataFrame\ndef reshape_row(row):\n    data = {'study_id': [], 'condition': [], 'level': [], 'severity': []}\n    \n    for column, value in row.items():\n        if column not in ['study_id', 'series_id', 'instance_number', 'x', 'y', 'series_description']:\n            parts = column.split('_')\n            condition = ' '.join([word.capitalize() for word in parts[:-2]])\n            level = parts[-2].capitalize() + '/' + parts[-1].capitalize()\n            data['study_id'].append(row['study_id'])\n            data['condition'].append(condition)\n            data['level'].append(level)\n            data['severity'].append(value)\n    \n    return pd.DataFrame(data)\n\n# Reshape the DataFrame for all rows\nnew_train_df = pd.concat([reshape_row(row) for _, row in train.iterrows()], ignore_index=True)\n\n# Display the first few rows of the reshaped dataframe\nnew_train_df.head(5)","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:29.431487Z","iopub.execute_input":"2025-01-09T00:20:29.431721Z","iopub.status.idle":"2025-01-09T00:20:30.429614Z","shell.execute_reply.started":"2025-01-09T00:20:29.431690Z","shell.execute_reply":"2025-01-09T00:20:30.428855Z"},"papermill":{"duration":1.209763,"end_time":"2024-12-18T23:06:19.723961","exception":false,"start_time":"2024-12-18T23:06:18.514198","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"9a9455e7","cell_type":"code","source":"# Print columns in a neat way\nprint(\"\\nColumns in new_train_df:\")\nprint(\",\".join(new_train_df.columns))\n\nprint(\"\\nColumns in label:\")\nprint(\",\".join(label.columns))\n\nprint(\"\\nColumns in test_desc:\")\nprint(\",\".join(test_desc.columns))\n\nprint(\"\\nColumns in sub:\")\nprint(\",\".join(sub.columns))","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:30.430353Z","iopub.execute_input":"2025-01-09T00:20:30.430564Z","iopub.status.idle":"2025-01-09T00:20:30.436989Z","shell.execute_reply.started":"2025-01-09T00:20:30.430546Z","shell.execute_reply":"2025-01-09T00:20:30.436235Z"},"papermill":{"duration":0.01879,"end_time":"2024-12-18T23:06:19.753457","exception":false,"start_time":"2024-12-18T23:06:19.734667","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"ee686b87","cell_type":"code","source":"# Merge the dataframes on the common columns\nmerged_df = pd.merge(new_train_df, label, on=['study_id', 'condition', 'level'], how='inner')\n# Merge the dataframes on the common column 'series_id'\nfinal_merged_df = pd.merge(merged_df, train_desc, on='series_id', how='inner')","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:30.437841Z","iopub.execute_input":"2025-01-09T00:20:30.438157Z","iopub.status.idle":"2025-01-09T00:20:30.510971Z","shell.execute_reply.started":"2025-01-09T00:20:30.438127Z","shell.execute_reply":"2025-01-09T00:20:30.510317Z"},"papermill":{"duration":0.086193,"end_time":"2024-12-18T23:06:19.850331","exception":false,"start_time":"2024-12-18T23:06:19.764138","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"b373d2f8","cell_type":"code","source":"# Merge the dataframes on the common column 'series_id'\nfinal_merged_df = pd.merge(merged_df, train_desc, on=['series_id','study_id'], how='inner')\n# Display the first few rows of the final merged dataframe\nfinal_merged_df.head(5)","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:30.514292Z","iopub.execute_input":"2025-01-09T00:20:30.514513Z","iopub.status.idle":"2025-01-09T00:20:30.538366Z","shell.execute_reply.started":"2025-01-09T00:20:30.514495Z","shell.execute_reply":"2025-01-09T00:20:30.537588Z"},"papermill":{"duration":0.041332,"end_time":"2024-12-18T23:06:19.902434","exception":false,"start_time":"2024-12-18T23:06:19.861102","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"dfa89a39","cell_type":"code","source":"import pandas as pd\n\n# Create the row_id column\nfinal_merged_df['row_id'] = (\n    final_merged_df['study_id'].astype(str) + '_' +\n    final_merged_df['condition'].str.lower().str.replace(' ', '_') + '_' +\n    final_merged_df['level'].str.lower().str.replace('/', '_')\n)\n\n# Create the image_path column\nfinal_merged_df['image_path'] = (\n    f'{train_path}/train_images/' + \n    final_merged_df['study_id'].astype(str) + '/' +\n    final_merged_df['series_id'].astype(str) + '/' +\n    final_merged_df['instance_number'].astype(str) + '.dcm'\n)\n\n# Note: Check image path, since there's 1 instance id, for 1 image, but there's many more images other than the ones labelled in the instance ID. \n\n# Display the updated dataframe\nfinal_merged_df.head(5)","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:30.539911Z","iopub.execute_input":"2025-01-09T00:20:30.540153Z","iopub.status.idle":"2025-01-09T00:20:30.713445Z","shell.execute_reply.started":"2025-01-09T00:20:30.540134Z","shell.execute_reply":"2025-01-09T00:20:30.712738Z"},"papermill":{"duration":0.229759,"end_time":"2024-12-18T23:06:20.142757","exception":false,"start_time":"2024-12-18T23:06:19.912998","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"45e5a70b","cell_type":"code","source":"final_merged_df[final_merged_df[\"severity\"] == \"Normal/Mild\"].value_counts().sum()","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:30.714093Z","iopub.execute_input":"2025-01-09T00:20:30.714282Z","iopub.status.idle":"2025-01-09T00:20:30.815848Z","shell.execute_reply.started":"2025-01-09T00:20:30.714264Z","shell.execute_reply":"2025-01-09T00:20:30.815113Z"},"papermill":{"duration":0.132002,"end_time":"2024-12-18T23:06:20.286576","exception":false,"start_time":"2024-12-18T23:06:20.154574","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"81004573","cell_type":"code","source":"final_merged_df[final_merged_df[\"severity\"] == \"Moderate\"].value_counts().sum()","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:30.816629Z","iopub.execute_input":"2025-01-09T00:20:30.816843Z","iopub.status.idle":"2025-01-09T00:20:30.849901Z","shell.execute_reply.started":"2025-01-09T00:20:30.816823Z","shell.execute_reply":"2025-01-09T00:20:30.849242Z"},"papermill":{"duration":0.052468,"end_time":"2024-12-18T23:06:20.350445","exception":false,"start_time":"2024-12-18T23:06:20.297977","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"aa5336fa","cell_type":"code","source":"final_merged_df[final_merged_df[\"severity\"] == \"Severe\"].value_counts().sum()","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:30.850724Z","iopub.execute_input":"2025-01-09T00:20:30.851089Z","iopub.status.idle":"2025-01-09T00:20:30.871161Z","shell.execute_reply.started":"2025-01-09T00:20:30.851053Z","shell.execute_reply":"2025-01-09T00:20:30.870424Z"},"papermill":{"duration":0.036552,"end_time":"2024-12-18T23:06:20.398755","exception":false,"start_time":"2024-12-18T23:06:20.362203","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"cb47b024","cell_type":"code","source":"import pandas as pd\n\n# En düşük sınıf sayısını belirleyelim\nmin_class_count = 3081\n\n# Normal/Mild ve Moderate sınıflarını azaltalım\nnormal_mild_df = final_merged_df[final_merged_df[\"severity\"] == \"Normal/Mild\"].sample(n=min_class_count, random_state=42)\nmoderate_df = final_merged_df[final_merged_df[\"severity\"] == \"Moderate\"].sample(n=min_class_count, random_state=42)\nsevere_df = final_merged_df[final_merged_df[\"severity\"] == \"Severe\"]\n\n# İndeksleri sıfırlayalım\nnormal_mild_df = normal_mild_df.reset_index(drop=True)\nmoderate_df = moderate_df.reset_index(drop=True)\nsevere_df = severe_df.reset_index(drop=True)\n\n# Verileri birleştirelim ve final_merged_df'yi güncelleyelim\nfinal_merged_df = pd.concat([normal_mild_df, moderate_df, severe_df])\n\n# Sonuçları kontrol edelim\nprint(final_merged_df[\"severity\"].value_counts())\n","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:30.872027Z","iopub.execute_input":"2025-01-09T00:20:30.872339Z","iopub.status.idle":"2025-01-09T00:20:30.906354Z","shell.execute_reply.started":"2025-01-09T00:20:30.872306Z","shell.execute_reply":"2025-01-09T00:20:30.905685Z"},"papermill":{"duration":0.054813,"end_time":"2024-12-18T23:06:20.465346","exception":false,"start_time":"2024-12-18T23:06:20.410533","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"a0d5aa7a","cell_type":"code","source":"final_merged_df[final_merged_df[\"severity\"] == \"Normal/Mild\"].value_counts().sum()","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:30.907117Z","iopub.execute_input":"2025-01-09T00:20:30.907308Z","iopub.status.idle":"2025-01-09T00:20:30.924602Z","shell.execute_reply.started":"2025-01-09T00:20:30.907290Z","shell.execute_reply":"2025-01-09T00:20:30.924015Z"},"papermill":{"duration":0.036226,"end_time":"2024-12-18T23:06:20.514462","exception":false,"start_time":"2024-12-18T23:06:20.478236","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"11097f0d","cell_type":"code","source":"final_merged_df[final_merged_df[\"severity\"] == \"Moderate\"].value_counts().sum()","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:30.925463Z","iopub.execute_input":"2025-01-09T00:20:30.925735Z","iopub.status.idle":"2025-01-09T00:20:30.944024Z","shell.execute_reply.started":"2025-01-09T00:20:30.925708Z","shell.execute_reply":"2025-01-09T00:20:30.943399Z"},"papermill":{"duration":0.035131,"end_time":"2024-12-18T23:06:20.561603","exception":false,"start_time":"2024-12-18T23:06:20.526472","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"f11360a3","cell_type":"code","source":"final_merged_df[final_merged_df[\"severity\"] == \"Severe\"].value_counts().sum()","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:30.944706Z","iopub.execute_input":"2025-01-09T00:20:30.944995Z","iopub.status.idle":"2025-01-09T00:20:30.962237Z","shell.execute_reply.started":"2025-01-09T00:20:30.944968Z","shell.execute_reply":"2025-01-09T00:20:30.961411Z"},"papermill":{"duration":0.032782,"end_time":"2024-12-18T23:06:20.606101","exception":false,"start_time":"2024-12-18T23:06:20.573319","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"5a4f326f","cell_type":"code","source":"# Define the base path for test images\nbase_path = '/kaggle/input/rsna-2024-lumbar-spine-degenerative-classification/test_images/'\n\n# Function to get image paths for a series\ndef get_image_paths(row):\n    series_path = os.path.join(base_path, str(row['study_id']), str(row['series_id']))\n    if os.path.exists(series_path):\n        return [os.path.join(series_path, f) for f in os.listdir(series_path) if os.path.isfile(os.path.join(series_path, f))]\n    return []\n\n# Mapping of series_description to conditions\ncondition_mapping = {\n    'Sagittal T1': {'left': 'left_neural_foraminal_narrowing', 'right': 'right_neural_foraminal_narrowing'},\n    'Axial T2': {'left': 'left_subarticular_stenosis', 'right': 'right_subarticular_stenosis'},\n    'Sagittal T2/STIR': 'spinal_canal_stenosis'\n}\n\n# Create a list to store the expanded rows\nexpanded_rows = []\n\n# Expand the dataframe by adding new rows for each file path\nfor index, row in test_desc.iterrows():\n    image_paths = get_image_paths(row)\n    conditions = condition_mapping.get(row['series_description'], {})\n    if isinstance(conditions, str):  # Single condition\n        conditions = {'left': conditions, 'right': conditions}\n    for side, condition in conditions.items():\n        for image_path in image_paths:\n            expanded_rows.append({\n                'study_id': row['study_id'],\n                'series_id': row['series_id'],\n                'series_description': row['series_description'],\n                'image_path': image_path,\n                'condition': condition,\n                'row_id': f\"{row['study_id']}_{condition}\"\n            })\n\n# Create a new dataframe from the expanded rows\nexpanded_test_desc = pd.DataFrame(expanded_rows)\n\n# Display the resulting dataframe\nexpanded_test_desc.head(5)","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:30.962877Z","iopub.execute_input":"2025-01-09T00:20:30.963097Z","iopub.status.idle":"2025-01-09T00:20:31.055683Z","shell.execute_reply.started":"2025-01-09T00:20:30.963078Z","shell.execute_reply":"2025-01-09T00:20:31.054833Z"},"papermill":{"duration":0.074643,"end_time":"2024-12-18T23:06:20.692329","exception":false,"start_time":"2024-12-18T23:06:20.617686","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"a25a19b8","cell_type":"code","source":"# change severity column labels\n#Normal/Mild': 'normal_mild', 'Moderate': 'moderate', 'Severe': 'severe'}\nfinal_merged_df['severity'] = final_merged_df['severity'].map({'Normal/Mild': 'normal_mild', 'Moderate': 'moderate', 'Severe': 'severe'})","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:31.056516Z","iopub.execute_input":"2025-01-09T00:20:31.056720Z","iopub.status.idle":"2025-01-09T00:20:31.061467Z","shell.execute_reply.started":"2025-01-09T00:20:31.056701Z","shell.execute_reply":"2025-01-09T00:20:31.060707Z"},"papermill":{"duration":0.019436,"end_time":"2024-12-18T23:06:20.724234","exception":false,"start_time":"2024-12-18T23:06:20.704798","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"95cff6ce","cell_type":"code","source":"test_data = expanded_test_desc\ntrain_data = final_merged_df","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:31.062221Z","iopub.execute_input":"2025-01-09T00:20:31.062408Z","iopub.status.idle":"2025-01-09T00:20:31.076536Z","shell.execute_reply.started":"2025-01-09T00:20:31.062391Z","shell.execute_reply":"2025-01-09T00:20:31.075764Z"},"papermill":{"duration":0.017134,"end_time":"2024-12-18T23:06:20.753257","exception":false,"start_time":"2024-12-18T23:06:20.736123","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"30fc0133","cell_type":"code","source":"train_data.head(5)","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:31.077297Z","iopub.execute_input":"2025-01-09T00:20:31.077484Z","iopub.status.idle":"2025-01-09T00:20:31.102683Z","shell.execute_reply.started":"2025-01-09T00:20:31.077466Z","shell.execute_reply":"2025-01-09T00:20:31.101853Z"},"papermill":{"duration":0.026603,"end_time":"2024-12-18T23:06:20.791537","exception":false,"start_time":"2024-12-18T23:06:20.764934","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"9fa38bd4","cell_type":"code","source":"train_data['series_description'].value_counts()","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:31.103360Z","iopub.execute_input":"2025-01-09T00:20:31.103560Z","iopub.status.idle":"2025-01-09T00:20:31.119346Z","shell.execute_reply.started":"2025-01-09T00:20:31.103539Z","shell.execute_reply":"2025-01-09T00:20:31.118536Z"},"papermill":{"duration":0.02034,"end_time":"2024-12-18T23:06:20.823816","exception":false,"start_time":"2024-12-18T23:06:20.803476","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"f7c2ff8f","cell_type":"code","source":"def load_dicom(path):\n    dicom = pydicom.dcmread(path)\n    data = dicom.pixel_array\n    data = data - np.min(data)\n    if np.max(data) != 0:\n        data = data / np.max(data)\n    data = (data * 255).astype(np.uint8)\n    return data","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:31.120068Z","iopub.execute_input":"2025-01-09T00:20:31.120247Z","iopub.status.idle":"2025-01-09T00:20:31.134255Z","shell.execute_reply.started":"2025-01-09T00:20:31.120231Z","shell.execute_reply":"2025-01-09T00:20:31.133628Z"},"papermill":{"duration":0.01789,"end_time":"2024-12-18T23:06:20.853698","exception":false,"start_time":"2024-12-18T23:06:20.835808","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"fbd78333","cell_type":"code","source":"import random\nimport matplotlib.pyplot as plt\n\n# Yeni sıfırlanmış indekslerle rastgele seçim yapalım\nfinal_merged_df_reset = final_merged_df.reset_index(drop=True)\n\n# Rastgele iki indeks seçelim\nselected_indices = random.sample(range(len(final_merged_df_reset)), 2)\n\nimages = []\nrow_ids = []\n\n# Seçilen indekslerle görselleri yükleyelim\nfor i in selected_indices:\n    image = load_dicom(final_merged_df_reset['image_path'][i])  # Yeni sıfırlanmış indeksi kullan\n    images.append(image)\n    row_ids.append(final_merged_df_reset['row_id'][i])  # Yeni sıfırlanmış indeksi kullan\n\n# Görselleri çizdirelim\nfig, ax = plt.subplots(1, 2, figsize=(8, 4))\nfor i in range(2):\n    ax[i].imshow(images[i], cmap='gray')\n    ax[i].set_title(f'Row ID: {row_ids[i]}', fontsize=8)\n    ax[i].axis('off')\nplt.tight_layout()\nplt.show()\n","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:31.135035Z","iopub.execute_input":"2025-01-09T00:20:31.135297Z","iopub.status.idle":"2025-01-09T00:20:31.607773Z","shell.execute_reply.started":"2025-01-09T00:20:31.135271Z","shell.execute_reply":"2025-01-09T00:20:31.606821Z"},"papermill":{"duration":0.477334,"end_time":"2024-12-18T23:06:21.342510","exception":false,"start_time":"2024-12-18T23:06:20.865176","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"a5122795","cell_type":"code","source":"train_data ","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:31.608599Z","iopub.execute_input":"2025-01-09T00:20:31.608876Z","iopub.status.idle":"2025-01-09T00:20:31.623120Z","shell.execute_reply.started":"2025-01-09T00:20:31.608850Z","shell.execute_reply":"2025-01-09T00:20:31.622366Z"},"papermill":{"duration":0.034675,"end_time":"2024-12-18T23:06:21.395618","exception":false,"start_time":"2024-12-18T23:06:21.360943","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"fb1b62d5","cell_type":"code","source":"train_data = train_data.dropna()","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:31.623766Z","iopub.execute_input":"2025-01-09T00:20:31.623996Z","iopub.status.idle":"2025-01-09T00:20:31.644584Z","shell.execute_reply.started":"2025-01-09T00:20:31.623977Z","shell.execute_reply":"2025-01-09T00:20:31.643900Z"},"papermill":{"duration":0.032005,"end_time":"2024-12-18T23:06:21.445140","exception":false,"start_time":"2024-12-18T23:06:21.413135","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"6d048e72","cell_type":"code","source":"import pandas as pd\nfrom sklearn.model_selection import train_test_split\nfrom torch.utils.data import Dataset, DataLoader\nimport torchvision.transforms as transforms\nimport torch\nimport torch.optim.lr_scheduler as lr_scheduler\nfrom tqdm import tqdm\n\n# Define a custom dataset class\nclass CustomDataset(Dataset):\n    def __init__(self, dataframe, transform=None):\n        self.dataframe = dataframe\n        self.transform = transform\n\n    def __len__(self):\n        return len(self.dataframe)\n\n    def __getitem__(self, index):\n        image_path = self.dataframe['image_path'][index]\n        image = load_dicom(image_path)  # Define this function to load your DICOM images\n        label = self.dataframe['severity'][index]\n        \n        if self.transform:\n            image = self.transform(image)\n\n        return image, label\n\n\"\"\"# Function to create datasets and dataloaders for each series description\ndef create_datasets_and_loaders(df, series_description, transform, batch_size=8):\n    filtered_df = df[df['series_description'] == series_description]\n    \n    train_df, val_df = train_test_split(filtered_df, test_size=0.2, random_state=42)\n    train_df = train_df.reset_index(drop=True)\n    val_df = val_df.reset_index(drop=True)\n\n    train_dataset = CustomDataset(train_df, transform)\n    val_dataset = CustomDataset(val_df, transform)\n\n    trainloader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)\n    valloader = DataLoader(val_dataset, batch_size=batch_size, shuffle=False)\n    \n    return trainloader, valloader, len(train_df), len(val_df)\"\"\"\n# Function to create datasets and dataloaders for each series description\ndef create_datasets_and_loaders(df, series_description, transform, batch_size=8):\n    filtered_df = df[df['series_description'] == series_description]\n    \n    # %5'ini al frac değerini değiştirerek trainde verinin ne kadarını kullanacagını belirlersin\n    filtered_df = filtered_df.sample(frac=1.0, random_state=42)  \n    \n    train_df, val_df = train_test_split(filtered_df, test_size=0.2, random_state=42)\n    train_df = train_df.reset_index(drop=True)\n    val_df = val_df.reset_index(drop=True)\n\n    train_dataset = CustomDataset(train_df, transform)\n    val_dataset = CustomDataset(val_df, transform)\n\n    trainloader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)\n    valloader = DataLoader(val_dataset, batch_size=batch_size, shuffle=False)\n    \n    return trainloader, valloader, len(train_df), len(val_df)\n\n\n# Define the transforms\ntransform = transforms.Compose([\n    transforms.Lambda(lambda x: (x * 255).astype(np.uint8)),  # Convert back to uint8 for PIL\n    transforms.ToPILImage(),\n    transforms.Resize((224, 224)),\n    transforms.Grayscale(num_output_channels=3),\n    transforms.ToTensor(),\n])\n\n# Create dataloaders for each series description\ndataloaders = {}\nlengths = {}\n\ntrainloader_t1, valloader_t1, len_train_t1, len_val_t1 = create_datasets_and_loaders(train_data, 'Sagittal T1', transform)\ntrainloader_t2, valloader_t2, len_train_t2, len_val_t2 = create_datasets_and_loaders(train_data, 'Axial T2', transform)\ntrainloader_t2stir, valloader_t2stir, len_train_t2stir, len_val_t2stir = create_datasets_and_loaders(train_data, 'Sagittal T2/STIR', transform)\n\ndataloaders['Sagittal T1'] = (trainloader_t1, valloader_t1)\ndataloaders['Axial T2'] = (trainloader_t2, valloader_t2)\ndataloaders['Sagittal T2/STIR'] = (trainloader_t2stir, valloader_t2stir)\n\nlengths['Sagittal T1'] = (len_train_t1, len_val_t1)\nlengths['Axial T2'] = (len_train_t2, len_val_t2)\nlengths['Sagittal T2/STIR'] = (len_train_t2stir, len_val_t2stir)\n\n# Dictionary mapping labels to indices\nlabel_map = {'Mild': 0, 'Moderate': 1, 'Severe': 2}","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:31.645373Z","iopub.execute_input":"2025-01-09T00:20:31.645631Z","iopub.status.idle":"2025-01-09T00:20:32.783388Z","shell.execute_reply.started":"2025-01-09T00:20:31.645610Z","shell.execute_reply":"2025-01-09T00:20:32.782712Z"},"papermill":{"duration":1.749743,"end_time":"2024-12-18T23:06:23.212726","exception":false,"start_time":"2024-12-18T23:06:21.462983","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"88cdbc76","cell_type":"code","source":"import matplotlib.pyplot as plt\n\n# Function to visualize a batch of images\ndef visualize_batch(dataloader):\n    images, labels = next(iter(dataloader))\n    fig, axes = plt.subplots(1, len(images), figsize=(20, 5))\n    for i, (img, lbl) in enumerate(zip(images, labels)):\n        ax = axes[i]\n        img = img.permute(1, 2, 0)  # Convert to HWC for visualization\n        ax.imshow(img)\n        ax.set_title(f\"Label: {lbl}\")\n        ax.axis('off')\n    plt.show()\n\n# Visualize samples from each dataloader\nprint(\"Visualizing Sagittal T1 samples\")\nvisualize_batch(trainloader_t1)\nprint(\"Visualizing Axial T2 samples\")\nvisualize_batch(trainloader_t2)\nprint(\"Visualizing Sagittal T2/STIR samples\")\nvisualize_batch(trainloader_t2stir)","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:32.784197Z","iopub.execute_input":"2025-01-09T00:20:32.784556Z","iopub.status.idle":"2025-01-09T00:20:35.148599Z","shell.execute_reply.started":"2025-01-09T00:20:32.784535Z","shell.execute_reply":"2025-01-09T00:20:35.147746Z"},"papermill":{"duration":2.525854,"end_time":"2024-12-18T23:06:25.756109","exception":false,"start_time":"2024-12-18T23:06:23.230255","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"080efbc6","cell_type":"code","source":"import matplotlib.pyplot as plt\n\nimage, label = next(iter(trainloader_t2))\nsample = image[1].permute(1, 2, 0)  #sample\n\n# Plot images\nplt.figsize=(8, 4)\nplt.imshow(images[0], cmap='gray')\nplt.title(label[0])\nplt.axis('off')\nplt.tight_layout()\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:35.149498Z","iopub.execute_input":"2025-01-09T00:20:35.149780Z","iopub.status.idle":"2025-01-09T00:20:35.434596Z","shell.execute_reply.started":"2025-01-09T00:20:35.149754Z","shell.execute_reply":"2025-01-09T00:20:35.433813Z"},"papermill":{"duration":0.436418,"end_time":"2024-12-18T23:06:26.231681","exception":false,"start_time":"2024-12-18T23:06:25.795263","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"01bd85fe","cell_type":"code","source":"import torch\nimport torch.nn as nn\nimport torchvision.models as models\nfrom torchvision import transforms\nfrom torch.utils.data import DataLoader\nfrom sklearn.model_selection import train_test_split\nimport pandas as pd\nfrom tqdm import tqdm\n\ndevice = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:35.435188Z","iopub.execute_input":"2025-01-09T00:20:35.435461Z","iopub.status.idle":"2025-01-09T00:20:35.489700Z","shell.execute_reply.started":"2025-01-09T00:20:35.435428Z","shell.execute_reply":"2025-01-09T00:20:35.488902Z"},"papermill":{"duration":0.099929,"end_time":"2024-12-18T23:06:26.374441","exception":false,"start_time":"2024-12-18T23:06:26.274512","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"82fa3373","cell_type":"code","source":"import torch\nimport torch.nn as nn\nimport torchvision.models as models\nfrom torchvision import transforms\nfrom torch.utils.data import DataLoader\nfrom sklearn.model_selection import train_test_split\nimport pandas as pd\nfrom tqdm import tqdm\n\ndevice = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n\nclass CustomResNet50(nn.Module):\n    def __init__(self, num_classes=3, pretrained_weights=None):\n        super(CustomResNet50, self).__init__()\n        # pretrained=False ile modelin rastgele ağırlıklarla başlatılmasını sağla\n        self.model = models.resnet50(weights=None)  # torchvision 0.13'te 'pretrained' yerine 'weights' kullanılmalı\n        \n        # Eğer manuel ağırlık yolu verilmişse, bu ağırlıkları yükle\n        if pretrained_weights:\n            self.model.load_state_dict(torch.load(pretrained_weights))\n        \n        num_ftrs = self.model.fc.in_features  # Son katmanın özellik sayısını al\n        self.model.fc = nn.Linear(num_ftrs, num_classes)  # Son katmanı değiştir\n\n    def forward(self, x):\n        return self.model(x)\n\n    def unfreeze_middle_layers(self):\n        \"\"\"Orta katmanları çöz.\"\"\"\n        # Orta katmanlar: 3. ve 4. blokları çöz\n        for name, param in self.model.named_parameters():\n            # Bu katmanlar arasında layer3 ve layer4 çözülür\n            if 'layer3' in name or 'layer4' in name:  \n                param.requires_grad = True  # Orta katmanlar için requires_grad=True\n            else:\n                param.requires_grad = False  # Diğer tüm katmanları dondur\n\n\n# Modeli başlat\nsagittal_t1_model = CustomResNet50(num_classes=3).to(device)\naxial_t2_model = CustomResNet50(num_classes=3).to(device)\nsagittal_t2stir_model = CustomResNet50(num_classes=3).to(device)\n\n# Orta katmanları çözmek için\nfor model in [sagittal_t1_model, axial_t2_model, sagittal_t2stir_model]:\n    model.unfreeze_middle_layers()  # Orta katmanları çöz\n\n# Eğitim parametreleri\nweights = torch.tensor([1.0, 2.0, 4.0])\ncriterion = nn.CrossEntropyLoss(weight=weights.to(device))\n\n# Optimizer ayarları\noptimizer_sagittal_t1 = torch.optim.Adam(sagittal_t1_model.parameters(), lr=0.001)\noptimizer_axial_t2 = torch.optim.Adam(axial_t2_model.parameters(), lr=0.001)\noptimizer_sagittal_t2stir = torch.optim.Adam(sagittal_t2stir_model.parameters(), lr=0.001)\n\n# Modelleri ve optimizörleri saklamak için dictionary\nmodels_dict = {\n    'Sagittal T1': sagittal_t1_model,\n    'Axial T2': axial_t2_model,\n    'Sagittal T2/STIR': sagittal_t2stir_model,\n}\n\noptimizers_dict = {\n    'Sagittal T1': optimizer_sagittal_t1,\n    'Axial T2': optimizer_axial_t2,\n    'Sagittal T2/STIR': optimizer_sagittal_t2stir,\n}\n\n# Eğitim yapılabilir parametrelerin sayısını yazdır\nfor model_name, model in models_dict.items():\n    trainable_params = sum(p.numel() for p in model.parameters() if p.requires_grad)\n    print(f\"Trainable parameters for {model_name}: {trainable_params}\")\n","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:20:35.490530Z","iopub.execute_input":"2025-01-09T00:20:35.490752Z","iopub.status.idle":"2025-01-09T00:20:36.872531Z","shell.execute_reply.started":"2025-01-09T00:20:35.490731Z","shell.execute_reply":"2025-01-09T00:20:36.871721Z"},"papermill":{"duration":1.601431,"end_time":"2024-12-18T23:06:28.015414","exception":false,"start_time":"2024-12-18T23:06:26.413983","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"60cd47e6","cell_type":"code","source":"label_map = {'normal_mild': 0, 'moderate': 1, 'severe': 2}","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:21:14.401283Z","iopub.execute_input":"2025-01-09T00:21:14.401576Z","iopub.status.idle":"2025-01-09T00:21:14.405463Z","shell.execute_reply.started":"2025-01-09T00:21:14.401556Z","shell.execute_reply":"2025-01-09T00:21:14.404458Z"},"papermill":{"duration":0.046798,"end_time":"2024-12-18T23:06:28.103843","exception":false,"start_time":"2024-12-18T23:06:28.057045","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"7b218211","cell_type":"code","source":"for images, labels in trainloader_t2:\n    labels = torch.tensor([label_map[label] for label in labels])\n    labels = labels.to(device)\n    print(labels)\n    break","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:21:17.792956Z","iopub.execute_input":"2025-01-09T00:21:17.793257Z","iopub.status.idle":"2025-01-09T00:21:17.957748Z","shell.execute_reply.started":"2025-01-09T00:21:17.793234Z","shell.execute_reply":"2025-01-09T00:21:17.956861Z"},"papermill":{"duration":0.219249,"end_time":"2024-12-18T23:06:28.363534","exception":false,"start_time":"2024-12-18T23:06:28.144285","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"0839b080","cell_type":"code","source":"import torch.optim.lr_scheduler as lr_scheduler\nfrom copy import deepcopy\nfrom sklearn.metrics import (\n    confusion_matrix,\n    f1_score,\n    precision_score,\n    recall_score,\n    cohen_kappa_score,\n    matthews_corrcoef,\n    balanced_accuracy_score,\n    roc_auc_score,\n    classification_report\n)\n\ndef train_model(model, trainloader, valloader, len_train, len_val, optimizer, num_epochs=10, patience=3):\n    # Learning rate scheduler\n    scheduler = lr_scheduler.StepLR(optimizer, step_size=2, gamma=0.1)\n\n    best_val_acc = 0.0\n    best_model_wts = deepcopy(model.state_dict())\n    counter = 0\n\n    for epoch in range(num_epochs):\n        model.train()\n        train_loss = 0\n        correct_train = 0\n\n        with tqdm(trainloader, unit=\"batch\") as tepoch:\n            for images, labels in tepoch:\n                images, labels = images.to(device), torch.tensor([label_map[label] for label in labels]).to(device)\n                optimizer.zero_grad()\n                outputs = model(images)\n                loss = criterion(outputs, labels)\n                loss.backward()\n                optimizer.step()\n                train_loss += loss.item()\n\n                probabilities = torch.softmax(outputs, dim=1)\n                _, predicted = torch.max(probabilities, 1)\n                correct_train += (predicted == labels).sum().item()\n\n                tepoch.set_postfix(epoch=epoch+1)\n\n        scheduler.step()\n\n        train_loss /= len(trainloader)\n        train_acc = 100 * correct_train / len_train\n\n        model.eval()\n        val_loss, correct_val = 0, 0\n        all_labels = []\n        all_predictions = []\n\n        with torch.no_grad():\n            with tqdm(valloader, unit=\"batch\") as vepoch:\n                for images, labels in vepoch:\n                    images, labels = images.to(device), torch.tensor([label_map[label] for label in labels]).to(device)\n                    outputs = model(images)\n                    loss = criterion(outputs, labels)\n                    val_loss += loss.item()\n\n                    probabilities = torch.softmax(outputs, dim=1)\n\n                    if probabilities.dim() == 1:\n                        _, predicted = torch.max(probabilities, 0)\n                    else:\n                        _, predicted = torch.max(probabilities, 1)\n                    correct_val += (predicted == labels).sum().item()\n\n                    all_labels.extend(labels.cpu().numpy())\n                    all_predictions.extend(predicted.cpu().numpy())\n\n                    vepoch.set_postfix(epoch=epoch+1)\n\n        val_loss /= len(valloader)\n        val_acc = 100 * correct_val / len_val\n\n        # Confusion Matrix ve F1 Skoru Hesaplama\n        conf_matrix = confusion_matrix(all_labels, all_predictions)\n        f1 = f1_score(all_labels, all_predictions, average=\"weighted\")\n        class_report = classification_report(all_labels, all_predictions)\n\n        # Precision ve Recall\n        precision = precision_score(all_labels, all_predictions, average=\"weighted\")\n        recall = recall_score(all_labels, all_predictions, average=\"weighted\")\n\n        # Cohen's Kappa\n        kappa = cohen_kappa_score(all_labels, all_predictions)\n\n        # Matthews Correlation Coefficient (MCC)\n        mcc = matthews_corrcoef(all_labels, all_predictions)\n\n        # Balanced Accuracy\n        balanced_acc = balanced_accuracy_score(all_labels, all_predictions)\n\n        # ROC-AUC (Eğer çok sınıflıysa one-vs-rest yaklaşımı kullanılır)\n        try:\n            roc_auc = roc_auc_score(all_labels, torch.nn.functional.one_hot(torch.tensor(all_predictions), num_classes=len(set(all_labels))).numpy(), multi_class=\"ovr\")\n        except ValueError:\n            roc_auc = \"N/A (ROC-AUC çok sınıflı problemde uygun olmayabilir)\"\n\n        # Çıktılar\n        print(f\"Precision: {precision:.4f}\")\n        print(f\"Recall: {recall:.4f}\")\n        print(f\"Cohen's Kappa: {kappa:.4f}\")\n        print(f\"MCC: {mcc:.4f}\")\n        print(f\"Balanced Accuracy: {balanced_acc:.4f}\")\n        print(f\"ROC-AUC Score: {roc_auc}\")\n        print(f\"Epoch {epoch+1}, Train Loss: {train_loss:.4f}, Train Acc: {train_acc:.2f}%, Val Loss: {val_loss:.4f}, Val Acc: {val_acc:.2f}%\")\n        print(f\"Confusion Matrix:\\n{conf_matrix}\")\n        print(f\"F1 Score: {f1:.4f}\")\n        print(f\"Classification Report:\\n{class_report}\")\n\n        # Save the best model and check for early stopping\n        if val_acc > best_val_acc:\n            best_val_acc = val_acc\n            best_model_wts = deepcopy(model.state_dict())\n            counter = 0\n            torch.save(best_model_wts, f'best_model_{epoch+1}.pth')\n        else:\n            counter += 1\n\n        # Early stopping\n        if counter >= patience:\n            print(f\"Early stopping triggered after {epoch+1} epochs\")\n            break\n\n    # Load best model weights\n    model.load_state_dict(best_model_wts)\n    return model, best_val_acc\n","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:21:19.726027Z","iopub.execute_input":"2025-01-09T00:21:19.726341Z","iopub.status.idle":"2025-01-09T00:21:19.741100Z","shell.execute_reply.started":"2025-01-09T00:21:19.726316Z","shell.execute_reply":"2025-01-09T00:21:19.740193Z"},"papermill":{"duration":0.054746,"end_time":"2024-12-18T23:06:28.458854","exception":false,"start_time":"2024-12-18T23:06:28.404108","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"e7f94b42","cell_type":"code","source":"# Eğitim işlemi\nfor desc, model in models_dict.items():  # Burada 'models_dict' kullanılmalı\n    if desc == 'Sagittal T1':\n        trainloader, valloader, len_train, len_val = trainloader_t1, valloader_t1, len_train_t1, len_val_t1\n    elif desc == 'Axial T2':\n        trainloader, valloader, len_train, len_val = trainloader_t2, valloader_t2, len_train_t2, len_val_t2\n    elif desc == 'Sagittal T2/STIR':\n        trainloader, valloader, len_train, len_val = trainloader_t2stir, valloader_t2stir, len_train_t2stir, len_val_t2stir\n    \n    print(f\"Training model for {desc}\")\n    train_model(model, trainloader, valloader, len_train, len_val, optimizers_dict[desc])  # Burada 'optimizers_dict' kullanılmalı","metadata":{"execution":{"iopub.status.busy":"2025-01-09T00:21:25.059604Z","iopub.execute_input":"2025-01-09T00:21:25.059941Z","iopub.status.idle":"2025-01-09T00:22:58.775195Z","shell.execute_reply.started":"2025-01-09T00:21:25.059890Z","shell.execute_reply":"2025-01-09T00:22:58.773995Z"},"papermill":{"duration":1420.023013,"end_time":"2024-12-18T23:30:08.522458","exception":false,"start_time":"2024-12-18T23:06:28.499445","status":"completed"},"tags":[],"trusted":true},"outputs":[],"execution_count":null},{"id":"979bf287","cell_type":"code","source":"train_data['level'].unique()","metadata":{"execution":{"iopub.execute_input":"2024-12-18T23:30:10.621123Z","iopub.status.busy":"2024-12-18T23:30:10.620770Z","iopub.status.idle":"2024-12-18T23:30:10.627809Z","shell.execute_reply":"2024-12-18T23:30:10.626968Z"},"papermill":{"duration":1.070555,"end_time":"2024-12-18T23:30:10.629255","exception":false,"start_time":"2024-12-18T23:30:09.558700","status":"completed"},"tags":[]},"outputs":[],"execution_count":null},{"id":"38eebf4a","cell_type":"code","source":"expanded_test_desc.head(5)","metadata":{"execution":{"iopub.execute_input":"2024-12-18T23:30:12.809073Z","iopub.status.busy":"2024-12-18T23:30:12.808723Z","iopub.status.idle":"2024-12-18T23:30:12.819545Z","shell.execute_reply":"2024-12-18T23:30:12.818615Z"},"papermill":{"duration":1.088996,"end_time":"2024-12-18T23:30:12.820891","exception":false,"start_time":"2024-12-18T23:30:11.731895","status":"completed"},"tags":[]},"outputs":[],"execution_count":null},{"id":"4ccf67b4","cell_type":"code","source":"levels = ['l1_l2', 'l2_l3', 'l3_l4', 'l4_l5', 'l5_s1']\n\n# Function to update row_id with levels\ndef update_row_id(row, levels):\n    level = levels[row.name % len(levels)]\n    return f\"{row['study_id']}_{row['condition']}_{level}\"\n\n# Update row_id in expanded_test_desc to include levels\nexpanded_test_desc['row_id'] = expanded_test_desc.apply(lambda row: update_row_id(row, levels), axis=1)","metadata":{"execution":{"iopub.execute_input":"2024-12-18T23:30:14.950263Z","iopub.status.busy":"2024-12-18T23:30:14.949690Z","iopub.status.idle":"2024-12-18T23:30:14.959177Z","shell.execute_reply":"2024-12-18T23:30:14.958194Z"},"papermill":{"duration":1.086337,"end_time":"2024-12-18T23:30:14.961809","exception":false,"start_time":"2024-12-18T23:30:13.875472","status":"completed"},"tags":[]},"outputs":[],"execution_count":null},{"id":"728cbac1","cell_type":"code","source":"expanded_test_desc","metadata":{"execution":{"iopub.execute_input":"2024-12-18T23:30:17.176161Z","iopub.status.busy":"2024-12-18T23:30:17.175825Z","iopub.status.idle":"2024-12-18T23:30:17.187418Z","shell.execute_reply":"2024-12-18T23:30:17.186458Z"},"papermill":{"duration":1.071205,"end_time":"2024-12-18T23:30:17.188980","exception":false,"start_time":"2024-12-18T23:30:16.117775","status":"completed"},"tags":[]},"outputs":[],"execution_count":null},{"id":"a78721cf","cell_type":"code","source":"# Define a custom test dataset class\nclass TestDataset(Dataset):\n    def __init__(self, dataframe, transform=None):\n        self.dataframe = dataframe\n        self.transform = transform\n\n    def __len__(self):\n        return len(self.dataframe)\n\n    def __getitem__(self, index):\n        image_path = self.dataframe['image_path'][index]\n        image = load_dicom(image_path)  # Define this function to load your DICOM images\n        if self.transform:\n            image = self.transform(image)\n        return image\n\n# Define the transforms\ntransform = transforms.Compose([\n    transforms.ToPILImage(),\n    transforms.Resize((224, 224)),\n    transforms.Grayscale(num_output_channels=3),\n    transforms.ToTensor(),\n])\n\n# Create a test dataset and dataloader\ntest_dataset = TestDataset(expanded_test_desc, transform)\ntestloader = DataLoader(test_dataset, batch_size=1, shuffle=False)","metadata":{"execution":{"iopub.execute_input":"2024-12-18T23:30:19.307810Z","iopub.status.busy":"2024-12-18T23:30:19.307460Z","iopub.status.idle":"2024-12-18T23:30:19.313530Z","shell.execute_reply":"2024-12-18T23:30:19.312753Z"},"papermill":{"duration":1.006233,"end_time":"2024-12-18T23:30:19.314896","exception":false,"start_time":"2024-12-18T23:30:18.308663","status":"completed"},"tags":[]},"outputs":[],"execution_count":null},{"id":"6f68f1a3","cell_type":"code","source":"for image in testloader:\n    print(image.shape)\n    break","metadata":{"execution":{"iopub.execute_input":"2024-12-18T23:30:21.451540Z","iopub.status.busy":"2024-12-18T23:30:21.451191Z","iopub.status.idle":"2024-12-18T23:30:21.485053Z","shell.execute_reply":"2024-12-18T23:30:21.484063Z"},"papermill":{"duration":1.117292,"end_time":"2024-12-18T23:30:21.486541","exception":false,"start_time":"2024-12-18T23:30:20.369249","status":"completed"},"tags":[]},"outputs":[],"execution_count":null},{"id":"60e9b31f","cell_type":"code","source":"# Define models in a dictionary (use a different name to avoid conflicts)\nmodel_dict = {\n    'Sagittal T1': sagittal_t1_model,\n    'Axial T2': axial_t2_model,\n    'Sagittal T2/STIR': sagittal_t2stir_model,\n}\n\n# Function to get the model based on series_description\ndef get_model(series_description):\n    return model_dict.get(series_description, None)\n\n# Function to make predictions on the test data\ndef predict_test_data(testloader, expanded_test_desc):\n    predictions = []\n    normal_mild_probs = []\n    moderate_probs = []\n    severe_probs = []\n    \n    # Set each model to evaluation mode\n    for model in model_dict.values():\n        model.eval()\n\n    with torch.no_grad():  # Disable gradient calculation during inference\n        for idx, images in enumerate(tqdm(testloader)):  # Iterate through the test data\n            images = images.to(device)  # Move images to the device\n            series_description = expanded_test_desc.iloc[idx]['series_description']  # Get description from DataFrame\n            \n            # Get the model corresponding to the series description\n            model = get_model(series_description)\n            \n            if model:  # If a valid model is found\n                outputs = model(images)  # Forward pass through the model\n                probs = torch.softmax(outputs, dim=1).squeeze(0)  # Get the probabilities for each class\n                normal_mild_probs.append(probs[0].item())  # Probability for normal/mild class\n                moderate_probs.append(probs[1].item())  # Probability for moderate class\n                severe_probs.append(probs[2].item())  # Probability for severe class\n                predictions.append(probs)  # Append the full prediction\n            else:  # If no model is found for the description\n                normal_mild_probs.append(None)\n                moderate_probs.append(None)\n                severe_probs.append(None)\n                predictions.append(None)\n\n    return normal_mild_probs, moderate_probs, severe_probs, predictions\n\n# Make predictions on the test data\nnormal_mild_probs, moderate_probs, severe_probs, test_predictions = predict_test_data(testloader, expanded_test_desc)\n","metadata":{"execution":{"iopub.execute_input":"2024-12-18T23:30:23.563187Z","iopub.status.busy":"2024-12-18T23:30:23.562866Z","iopub.status.idle":"2024-12-18T23:30:29.131653Z","shell.execute_reply":"2024-12-18T23:30:29.130535Z"},"papermill":{"duration":6.584052,"end_time":"2024-12-18T23:30:29.133285","exception":false,"start_time":"2024-12-18T23:30:22.549233","status":"completed"},"tags":[]},"outputs":[],"execution_count":null},{"id":"dcad2336","cell_type":"code","source":"test_predictions[0]","metadata":{"execution":{"iopub.execute_input":"2024-12-18T23:30:31.259261Z","iopub.status.busy":"2024-12-18T23:30:31.258899Z","iopub.status.idle":"2024-12-18T23:30:31.426086Z","shell.execute_reply":"2024-12-18T23:30:31.425170Z"},"papermill":{"duration":1.225958,"end_time":"2024-12-18T23:30:31.427581","exception":false,"start_time":"2024-12-18T23:30:30.201623","status":"completed"},"tags":[]},"outputs":[],"execution_count":null},{"id":"52cc1b59","cell_type":"code","source":"# Add predictions and probabilities to the test DataFrame\nexpanded_test_desc['normal_mild'] = normal_mild_probs\nexpanded_test_desc['moderate'] = moderate_probs\nexpanded_test_desc['severe'] = severe_probs","metadata":{"execution":{"iopub.execute_input":"2024-12-18T23:30:33.552928Z","iopub.status.busy":"2024-12-18T23:30:33.552595Z","iopub.status.idle":"2024-12-18T23:30:33.557971Z","shell.execute_reply":"2024-12-18T23:30:33.557058Z"},"papermill":{"duration":1.075096,"end_time":"2024-12-18T23:30:33.559606","exception":false,"start_time":"2024-12-18T23:30:32.484510","status":"completed"},"tags":[]},"outputs":[],"execution_count":null},{"id":"c6f63e21","cell_type":"code","source":"submission = expanded_test_desc[[\"row_id\",\"normal_mild\",\"moderate\",\"severe\"]]","metadata":{"execution":{"iopub.execute_input":"2024-12-18T23:30:35.720184Z","iopub.status.busy":"2024-12-18T23:30:35.719869Z","iopub.status.idle":"2024-12-18T23:30:35.724750Z","shell.execute_reply":"2024-12-18T23:30:35.724016Z"},"papermill":{"duration":1.113379,"end_time":"2024-12-18T23:30:35.726092","exception":false,"start_time":"2024-12-18T23:30:34.612713","status":"completed"},"tags":[]},"outputs":[],"execution_count":null},{"id":"ec7d3cf3","cell_type":"code","source":"submission.head(10)","metadata":{"execution":{"iopub.execute_input":"2024-12-18T23:30:37.808788Z","iopub.status.busy":"2024-12-18T23:30:37.808448Z","iopub.status.idle":"2024-12-18T23:30:37.818343Z","shell.execute_reply":"2024-12-18T23:30:37.817452Z"},"papermill":{"duration":1.014879,"end_time":"2024-12-18T23:30:37.820048","exception":false,"start_time":"2024-12-18T23:30:36.805169","status":"completed"},"tags":[]},"outputs":[],"execution_count":null},{"id":"d02566ad","cell_type":"code","source":"# Group by 'row_id' and sum the values\ngrouped_submission = submission.groupby('row_id').max().reset_index()\n\n# Normalize the columns\ngrouped_submission[['normal_mild', 'moderate', 'severe']] = grouped_submission[['normal_mild', 'moderate', 'severe']].div(grouped_submission[['normal_mild', 'moderate', 'severe']].sum(axis=1), axis=0)\n\n# Check the first 3 rows\ngrouped_submission","metadata":{"execution":{"iopub.execute_input":"2024-12-18T23:30:39.935940Z","iopub.status.busy":"2024-12-18T23:30:39.935428Z","iopub.status.idle":"2024-12-18T23:30:39.956952Z","shell.execute_reply":"2024-12-18T23:30:39.956210Z"},"papermill":{"duration":1.07052,"end_time":"2024-12-18T23:30:39.958444","exception":false,"start_time":"2024-12-18T23:30:38.887924","status":"completed"},"tags":[]},"outputs":[],"execution_count":null},{"id":"c8bc4a5e","cell_type":"code","source":"len(grouped_submission)","metadata":{"execution":{"iopub.execute_input":"2024-12-18T23:30:42.025106Z","iopub.status.busy":"2024-12-18T23:30:42.024485Z","iopub.status.idle":"2024-12-18T23:30:42.030194Z","shell.execute_reply":"2024-12-18T23:30:42.029145Z"},"papermill":{"duration":1.008813,"end_time":"2024-12-18T23:30:42.031754","exception":false,"start_time":"2024-12-18T23:30:41.022941","status":"completed"},"tags":[]},"outputs":[],"execution_count":null},{"id":"42d6f484","cell_type":"code","source":"sub[['normal_mild', 'moderate', 'severe']] = grouped_submission[['normal_mild', 'moderate', 'severe']]","metadata":{"execution":{"iopub.execute_input":"2024-12-18T23:30:44.177404Z","iopub.status.busy":"2024-12-18T23:30:44.177038Z","iopub.status.idle":"2024-12-18T23:30:44.183138Z","shell.execute_reply":"2024-12-18T23:30:44.182142Z"},"papermill":{"duration":1.067373,"end_time":"2024-12-18T23:30:44.184723","exception":false,"start_time":"2024-12-18T23:30:43.117350","status":"completed"},"tags":[]},"outputs":[],"execution_count":null},{"id":"d6c33131","cell_type":"code","source":"import os\n\n# Save the DataFrame to \"submission.csv\" in the desired directory\nsub.to_csv(\"/kaggle/working/submission.csv\", index=False)","metadata":{"execution":{"iopub.execute_input":"2024-12-18T23:30:46.393752Z","iopub.status.busy":"2024-12-18T23:30:46.393389Z","iopub.status.idle":"2024-12-18T23:30:46.400738Z","shell.execute_reply":"2024-12-18T23:30:46.399802Z"},"papermill":{"duration":1.131307,"end_time":"2024-12-18T23:30:46.402354","exception":false,"start_time":"2024-12-18T23:30:45.271047","status":"completed"},"tags":[]},"outputs":[],"execution_count":null},{"id":"2988a45c","cell_type":"code","source":"sub.head(5)","metadata":{"execution":{"iopub.execute_input":"2024-12-18T23:30:48.533873Z","iopub.status.busy":"2024-12-18T23:30:48.533571Z","iopub.status.idle":"2024-12-18T23:30:48.542314Z","shell.execute_reply":"2024-12-18T23:30:48.541449Z"},"papermill":{"duration":1.067178,"end_time":"2024-12-18T23:30:48.543866","exception":false,"start_time":"2024-12-18T23:30:47.476688","status":"completed"},"tags":[]},"outputs":[],"execution_count":null}]}