How nhahub/nha-5-187 works
BMTS is a research prototype for 3D brain-MRI tumor segmentation and patient-level assessment. Its central workflow takes T1, T1ce, T2 and FLAIR scans through optional standardization and segmentation, then reports tumor-region volumes, longitudinal change, radiomics and confidence maps. Training, exploratory data analysis, fast model deployment and MRI mask viewing are additional product capabilities. Edges are limited to relationships stated in the README or visible in the supplied excerpts and source index; unsampled app and pipeline internals are not inferred.
Main entry points: analyze.py Analysis pipeline · segmenter.py Fast segmenter · brats.py BraTS loader · trainer.py Model training
Components 21
MRI Data
- Scan cleaning
src/bmts/data/cleaning/standardize.py - Patient splits
src/bmts/data/splits.py - BraTS loader
src/bmts/data/datasets/brats.py - Augmentation
src/bmts/data/augmentation/transforms.py
Segmentation Model
- Model training
src/bmts/segmentation/training/trainer.py - 3D U-Net
src/bmts/segmentation/models/unet3d.py - Model export
src/bmts/deploy/export.py - Fast segmenter
src/bmts/deploy/segmenter.py
Patient Analysis
- Analysis pipeline
src/bmts/pipeline/analyze.py - Tumor volumes
src/bmts/quantitative/volumes.py - Longitudinal change
src/bmts/quantitative/longitudinal.py - Radiomics features
src/bmts/radiomics/features.py - TTA prediction
src/bmts/uncertainty/nnunet_tta.py - Confidence maps
src/bmts/uncertainty/maps.py - Analysis results
User Interfaces
- MRI mask viewer
apps/m2_t05_viewer.py - Web interface
apps/web_interface_m4_t03.py
EDA
- EDA statistics
src/bmts/eda/stats.py - EDA figures
src/bmts/eda/figures.py
Other components
- Research user
- MRI scans
Connections 23
- Research user → MRI scans: provides
README.md - MRI scans → Analysis pipeline: analyzes
README.md - Analysis pipeline → Scan cleaning: optionally cleans
src/bmts/pipeline/analyze.py - Analysis pipeline → TTA prediction: predicts mask
src/bmts/pipeline/analyze.py - TTA prediction → Confidence maps: accumulates passes
src/bmts/uncertainty/nnunet_tta.py - Analysis pipeline → Tumor volumes: computes volumes
src/bmts/pipeline/analyze.py - Analysis pipeline → Longitudinal change: computes change
src/bmts/pipeline/analyze.py - Analysis pipeline → Radiomics features: extracts features
src/bmts/pipeline/analyze.py - Analysis pipeline → Confidence maps: summarizes maps
src/bmts/pipeline/analyze.py - Analysis pipeline → Analysis results: produces
README.md - Research user → MRI mask viewer: views scans
README.md - MRI mask viewer → MRI scans: reads cleaned scans
apps/m2_t05_viewer.py - Research user → Web interface: uses
README.md - BraTS loader → Patient splits: checks eligibility
src/bmts/data/datasets/brats.py - BraTS loader → Augmentation: uses transforms
src/bmts/data/datasets/brats.py - Model training → BraTS loader: builds loaders
src/bmts/segmentation/training/trainer.py - Model training → 3D U-Net: trains model
src/bmts/segmentation/training/trainer.py - Model export → Fast segmenter: packages model for
src/bmts/deploy/segmenter.py - Research user → Fast segmenter: requests prediction
README.md - MRI scans → Fast segmenter: provides modalities
README.md - Fast segmenter → Analysis results: returns mask
README.md - Research user → EDA statistics: explores cohorts
README.md - EDA statistics → EDA figures: creates figures
src/bmts/eda/figures.py