How ytashan/brain-tumor-segmentation works
This notebook implements a BraTS 2020 MRI segmentation pipeline, from dataset discovery and preparation through model training and patient-level reporting. The primary user workflow supplies a patient’s MRI path, obtains a tumor segmentation, then derives size and heuristic risk results and produces a 3D report. Separate documented branches cover model evaluation and batch prediction for the unlabeled validation set. The repository contains no sampled source excerpts, so relationships below rely on the README’s documented stages and capabilities; implementation details and internal call ownership cannot be verified.
Main entry points: brain-tumor-segmentation-era-v4.ipynb Tumor Segmentation
Components 18
Data Preparation
- Dataset Discovery
brain-tumor-segmentation-era-v4.ipynb - MRI Exploration
brain-tumor-segmentation-era-v4.ipynb - MRI Preprocessing
brain-tumor-segmentation-era-v4.ipynb - Preprocessed Cache
brain-tumor-segmentation-era-v4.ipynb - Patch Dataset
brain-tumor-segmentation-era-v4.ipynb
Segmentation Model
- Residual Attention U-Net
brain-tumor-segmentation-era-v4.ipynb - Model Training
brain-tumor-segmentation-era-v4.ipynb - Sliding-Window Inference
brain-tumor-segmentation-era-v4.ipynb
Analysis and Reports
- Tumor Segmentation
brain-tumor-segmentation-era-v4.ipynb - Test Evaluation
brain-tumor-segmentation-era-v4.ipynb - Tumor Quantification
brain-tumor-segmentation-era-v4.ipynb - Research Risk Score
brain-tumor-segmentation-era-v4.ipynb - 3D Visualization
brain-tumor-segmentation-era-v4.ipynb
User Outputs
- Patient 3D Report
brain-tumor-segmentation-era-v4.ipynb - Validation Batch Prediction
brain-tumor-segmentation-era-v4.ipynb - Prediction and Metric Files
brain-tumor-segmentation-era-v4.ipynb
Other components
- Research User
- BraTS MRI Data
Connections 21
- Research User → Dataset Discovery: configures paths
README.md - BraTS MRI Data → Dataset Discovery: provides MRI data
README.md - Dataset Discovery → MRI Exploration: enables exploration
README.md - Dataset Discovery → MRI Preprocessing: locates data for
README.md - MRI Preprocessing → Preprocessed Cache: writes cache
README.md - Preprocessed Cache → Patch Dataset: supplies volumes
README.md - Patch Dataset → Model Training: provides samples
README.md - Model Training → Residual Attention U-Net: optimizes
README.md - Residual Attention U-Net → Sliding-Window Inference: provides predictions
README.md - Sliding-Window Inference → Tumor Segmentation: produces labels
README.md - Tumor Segmentation → Test Evaluation: evaluated on test set
README.md - Tumor Segmentation → Tumor Quantification: measured for size
README.md - Tumor Quantification → Research Risk Score: informs score
README.md - Tumor Segmentation → 3D Visualization: overlaid in
README.md - 3D Visualization → Patient 3D Report: renders report
README.md - Research Risk Score → Patient 3D Report: included in
README.md - Research User → Sliding-Window Inference: requests patient prediction
README.md - Sliding-Window Inference → Validation Batch Prediction: used for batch prediction
README.md - Validation Batch Prediction → Prediction and Metric Files: writes NIfTIs
README.md - Test Evaluation → Prediction and Metric Files: produces metrics
README.md - Patient 3D Report → Prediction and Metric Files: exports reports
README.md