TopoPult-SSL: Gland-Mask-Free Cross-Device Meibomian Gland Segmentation via Self-Distilled Weak Clinical Priors
Quick Answer
TopoPult-SSL introduces a two-stage framework for meibomian gland segmentation across devices, achieving a Dice score of 0.716 on the MGD-1k to CAMG benchmark.
Quick Take
TopoPult-SSL introduces a two-stage framework for meibomian gland segmentation across devices, achieving a Dice score of 0.716 on the MGD-1k to CAMG benchmark. The model outperforms UA-MT and ensemble teachers while enabling deployment without dense gland contouring, leveraging weak clinical signals.
Key Points
- Stage 1 uses weak clinical signals for gland segmentation without target masks.
- Stage 2 employs self-distillation to enhance model performance when target masks are available.
- Achieved Dice score of 0.716, surpassing UA-MT's 0.710 and ensemble teacher's 0.720.
- Gland-mask-free variant reaches Precision of 0.694, significantly better than SAM/MedSAM.
- Code and reproducibility scripts are publicly available for further research.
Article Excerpt
From source RSS / original summaryarXiv:2606. 05347v1 Announce Type: new Abstract: Every new clinical imaging device creates a domain shift where dense gland masks are expensive yet cheap clinical signals -- eyelid outlines, Pult grades, morphometric ratios -- are routinely recorded. We present TopoPult-SSL, a two-stage framework for cross-device meibomian gland segmentation. Stage 1 adapts a source-trained model without target gland masks in the training loss, using four weak-prior anchors driven by target eyelid masks and clinical metadata only.
Stage 2, when target gland masks are available, distils complementary Stage-1 teachers into a single compact student via supervised self-distillation. We develop and validate the technique on the public MGD-1k to CAMG research benchmark (1,000 to 100 images, different device), where the distilled model achieves Dice 0. 716+/-0. 006 (best 0. 726), surpassing UA-MT (0. 710) and the ensemble teacher (0. 720) -- with a single pass. The gland-mask-free Stage-1 variant reaches Precision 0. 694 vs. 0. 30-0.
34 for SAM/MedSAM (p<0. 001), enabling deployment without dense gland contouring. Code and reproducibility scripts are released.
Reader Mode unavailable (could not extract clean content).
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CV
See more →LLM-Guided ANN Index Optimization for Human-Object Interaction Retrieval
A phase-aware LLM agent optimizes human-object interaction retrieval, outperforming Optuna TPE by 33.3% and VDTuner by 34.2% on the HICO-DET benchmark. This method enhances throughput by 15.3x over UniIR and demonstrates strong transferability across vector database management systems.