fix: update README to trigger clean rebuild

README.md

@@ -10,44 +10,20 @@ app_file: app.py
 pinned: false
 ---
 
-# WoundNetB7 — Automated DFU Assessment Pipeline
+# WoundNetB7 — DFU Analysis Pipeline
 
-End-to-end pipeline for Diabetic Foot Ulcer (DFU) analysis:
+Complete pipeline for Diabetic Foot Ulcer analysis:
+1. **Binary segmentation** (ulcer detection, Dice: 0.927)
+2. **Multiclass segmentation** (background / foot / perilesion / ulcer)
+3. **Fitzpatrick/ITA** skin type estimation (86.9% accuracy)
+4. **PWAT scores** with Fitzpatrick debiasing (46.6% group gap reduction)
 
-**Image** -> **Multiclass Segmentation** -> **PWAT Scoring** + **Fitzpatrick/ITA Estimation**
+## Features
+- **Guided camera capture** with foot silhouette overlay for healthcare workers
+- **PDF clinical report** downloadable with all results
+- **JSON output** for system integration
 
-## Pipeline
+## Model
+EfficientNet-B7 + ASPP + CBAM + CoordAttention + TAM (Topological Attention Module)
 
-```
-Input Image (DFU photograph)
-    |
-    v
-[1] WoundNetB7 Segmentation (EfficientNet-B7 + ASPP + CBAM)
-    -> 4-class masks: background, foot, perilesion, ulcer
-    -> Ulcer Dice: 0.927 (95% CI: [0.917, 0.936])
-    |
-    +---> [2] PWAT Estimation (XGBoost per item)
-    |         -> Items 3-8 scores (0-4 ordinal)
-    |         -> 5-fold CV: MAE 0.61-0.87, Adjacent Match 75-89%
-    |
-    +---> [3] Fitzpatrick/ITA Estimation
-    |         -> Healthy skin = foot - perilesion - ulcer
-    |         -> ITA angle + Fitzpatrick I-VI classification
-    |         -> Calibrated on 61 DFU images (86.9% exact match)
-    |
-    +---> [4] Debiasing
-              -> PWAT scores adjusted by Fitzpatrick type
-              -> 18% max group gap reduction (p < 10^-27)
-```
-
-## Models
-
-| Component | Architecture | Metric | Value |
-|-----------|-------------|--------|-------|
-| Segmentation | EfficientNet-B7 + UNet + ASPP + CBAM | Ulcer Dice | 0.927 |
-| PWAT Item 7 | XGBoost + SMOTE | MAE / Adjacent | 0.61 / 89.3% |
-| Fitzpatrick | ITA calibrated thresholds | Exact match | 86.9% |
-
-## License
-
-Research use only. Clinical deployment requires regulatory approval.
+Trained with Combo Loss + Small Object Focal Loss. 6-fold TTA at inference.