How surajjha0244/hvit-for-medical-image-diagonosis works
This is a Django research prototype for multi-modal medical-image classification, with confidence-aware interpretation and a web interface for viewing diagnoses and reports. The documented workflow runs from image submission through preprocessing and hybrid CNN/ViT classification to interpretation and presentation; the views also support account, history, reporting, and feedback features. The graph uses the primary Diagnosis app, not the duplicate Traning Code copy. README statements support the model-stage relationships; source excerpts show URL routing and some record operations. Several internals are unsampled, so those links are omitted or marked unverified.
Main entry points: views.py Request handlers · urls.py App routes · explainability.html Explainability pages · models Model weights
Components 16
Web experience
- App routes
medical_image/Diagnosis/urls.py - Request handlers
medical_image/Diagnosis/views.py - Web pages
medical_image/templates/Diagnosis
Diagnosis workflow
- Image preprocessing
medical_image/Diagnosis/views.py - CNN features
medical_image/Diagnosis/views.py - Patch projection
medical_image/Diagnosis/views.py - Transformer encoder
medical_image/Diagnosis/views.py - Classification head
medical_image/Diagnosis/views.py - Model weights
medical_image/models
Interpretation
- Clinical interpretation
medical_image/Diagnosis/views.py - Explainability logic
medical_image/Diagnosis/xai.py - Explainability pages
medical_image/templates/Diagnosis/explainability.html
Records and services
- Diagnosis records
medical_image/Diagnosis/models.py - Reports and history
medical_image/Diagnosis/views.py - Accounts and feedback
medical_image/Diagnosis/models.py
Other components
- Image submitter
Connections 15
- Image submitter → App routes: submits image
README.md - App routes → Request handlers: routes requests
medical_image/Diagnosis/urls.py - Request handlers → Web pages: renders pages
- Request handlers → Image preprocessing: processes image
README.md - Image preprocessing → CNN features: extracts features
README.md - CNN features → Patch projection: tokenizes features
README.md - Patch projection → Transformer encoder: projects tokens
README.md - Transformer encoder → Classification head: classifies features
README.md - Classification head → Clinical interpretation: interprets prediction
README.md - Clinical interpretation → Explainability pages: presents interpretation
README.md - Request handlers → Diagnosis records: creates and deletes
medical_image/Diagnosis/views.py - Request handlers → Reports and history: serves reports
medical_image/Diagnosis/urls.py - Request handlers → Accounts and feedback: handles account flows
medical_image/Diagnosis/views.py - Model weights → Classification head: provides parameters
- Explainability logic → Explainability pages: supports explanations