Presentation Details
RaP-ProtoViT: Efficient Dual-Head Transformers for Robust Gastric Endoscopy Classification and Generalizable Clinical Deployment
ID：164
View Protection：ATTENDEE
Updated Time：2025-12-23 13:29:05 Hits：331
Online
Start Time：2025-12-30 13:45（Asia/Amman）
Duration：15min
Session：S7 Track 7: Pattern Recognition, Computer Vision and Image Processing »
Video
No Permission
Presentation File
Tips: Only the registered participant can access the file. Please sign in first.
Abstract
We introduce RaP-ProtoViT, an end-to-end dual-head transformer for 8-class GI endoscopy (Kvasir-v2). A margin head (ArcFace/AM-Softmax) enforces angular separation, while a prototype head aggregates top-k token–prototype similarities (with M trainable prototypes/class); a lightweight input-adaptive MLP fuses the heads. A leakage-aware pipeline (pHash dedup + GroupKFold) prevents near-duplicate bleed-over. Training uses AdamW(+SAM) with cosine warm-up, DropPath, label smoothing, SWA, and post-hoc temperature scaling; two-stage HPO (MOTPE+ASHA → qEHVI) under Latency@224 ≤ 200 ms and memory constraints selects operating points. On Kvasir-v2 the model attains 99.1% accuracy, Macro-F1 = 0.991, Macro-AUPRC = 0.997, AUROC = 0.998, and ECE ≈ 0.9%, with per-class F1 tightly clustered in 0.988–0.994 and fold stability (±0.2 pp accuracy, ±0.002 Macro-F1). Ablations show margin-only/prototype-only variants reduce Macro-F1 to 0.967/0.975 and raise ECE to 2.8%/2.2%; removing adaptive fusion drops Macro-F1 to 0.984. The proposed HPO converges 2–3× faster and yields better final MF1/AUPRC/ECE than Bayesian TPE or Random+ASHA. The prototype head provides localized, intrinsically interpretable evidence, complementing the margin head’s discrimination, within a single-model deployment footprint. By advancing robust, interpretable, and computationally efficient AI for gastric endoscopy, our approach can improve early detection of gastrointestinal disease and enable reliable clinical deployment across diverse healthcare settings.
Keywords
Endoscopy classification, Vision transformer, Prototype learning, hyperparameter optimization.
Speaker
Mohamadreza Khosravi
Researcher Shiraz University of Medical SciencesSubmission Author
All Comments
Important Date
-
Conference Date
Dec 29
2025
toDec 31
2025
-
Dec 20 2025
Draft paper submission deadline
-
Dec 31 2025
Contribution Submission Deadline
-
Dec 31 2025
Registration deadline
Sponsored By
United Societies of Science
Organized By
Zarqa University
Contact Information
- Miss AsianComNet
- as******@usssociety.org
Previous Conferences
-
2024-10-24 Thailand Bangkok
2024 Asian Conference on Communication and Networks
Submit Comment