Benchmarking the Domain Gap: Model Selection Instability Under Domain Shift in Video Capsule Endoscopy

Gastrointestinal cancers cause approximately 3.4 million deaths annually, and early small-bowel lesions are easily missed at wireless capsule endoscopy (WCE). RGB-trained WCE classifiers conflate hemoglobin contrast with bile staining and illumination falloff, limiting sensitivity to small-vessel vascular findings such as Lymphangiectasia. We introduce a physics-informed framework that injects an analytic, Monte-Carlo-inspired hemoglobin prior into a standard classifier purely at training time -- to our knowledge the first use of an explicit optical light-transport prior in WCE classification. On Kvasir-Capsule (47,238 frames, 43 patients, 11 evaluable classes; patient-disjoint split) we evaluate, across six seeds against an RGB-only EfficientNet-B0 baseline, a five-channel input-fusion variant feeding the prior alongside RGB, a distillation variant that runs on plain three-channel RGB at inference, and a three-stream extension adding a temporal Transformer and an autoencoder-residual stream; we replicate across ResNet-18 and ConvNeXt-Tiny and assess cross-vendor zero-shot transfer on the public Galar cohort. Input fusion lifts cross-seed macro-AUC from 0.760 to 0.783 (5/6 seeds positive); distillation reaches 0.773; the three-stream model reaches 0.804 (+0.044 over baseline, paired DeLong p < 0.0001). Lymphangiectasia AUC rises from 0.238 to 0.337, sign-consistent across all six seeds. A four-variant ablation reveals a parameterization-mechanism boundary: only the spatial-channel form lifts. Cross-vendor zero-shot on Galar retains about 60% of the lift. The distillation variant deploys on plain RGB with a free interpretability heatmap, and we release GalKva-2026, a paired cross-vendor benchmark.

Wireless capsule endoscopy (WCE) enables non-invasive visual assessment of the small bowel, but its clinical utility is constrained by the large volume of frames generated per examination and the difficulty of recognising subtle abnormalities under highly variable imaging conditions. Existing learning-based approaches for WCE are predominantly vision-only, often confined to narrow pathology sets, and show limited transfer across datasets and centres. To address these limitations, this study introduces CapCLIP, a domain-specific vision-language representation learning framework for WCE. CapCLIP aligns capsule endoscopy frames with clinically grounded textual descriptions derived from standardised nomenclature and pathology-aware caption templates, thereby learning embeddings that are both semantically informed and transferable. The proposed framework is evaluated against relevant open-source vision and vision-language foundation models under strict zero-shot conditions using unseen WCE datasets. Evaluation covers three downstream tasks: K-nearest neighbour classification, CLIP-style image-text classification, and text-to-image retrieval. Across these settings, CapCLIP consistently outperforms the compared baselines, with particularly strong gains in zero-shot image-text classification and cross-modal retrieval on out-of-distribution datasets. The results indicate that language-guided representation learning can improve both generalisation and semantic interpretability in WCE analysis. These findings position CapCLIP as a step toward foundation models tailored to capsule endoscopy and support the use of language-grounded WCE analysis.

Capsule endoscopy (CE) enables non-invasive gastrointestinal screening, but current CE research remains largely limited to frame-level classification and detection, leaving video-level analysis underexplored. To bridge this gap, we introduce and formally define a new task, diagnosis-driven CE video summarization, which requires extracting key evidence frames that covers clinically meaningful findings and making accurate diagnoses from those evidence frames. This setting is challenging because diagnostically relevant events are extremely sparse and can be overwhelmed by tens of thousands of redundant normal frames, while individual observations are often ambiguous due to motion blur, debris, specular highlights, and rapid viewpoint changes. To facilitate research in this direction, we introduce VideoCAP, the first CE dataset with diagnosis-driven annotations derived from real clinical reports. VideoCAP comprises 240 full-length videos and provides realistic supervision for both key evidence frame extraction and diagnosis. To address this task, we further propose DiCE, a clinician-inspired framework that mirrors the standard CE reading workflow. DiCE first performs efficient candidate screening over the raw video, then uses a Context Weaver to organize candidates into coherent diagnostic contexts that preserve distinct lesion events, and an Evidence Converger to aggregate multi-frame evidence within each context into robust clip-level judgments. Experiments show that DiCE consistently outperforms state-of-the-art methods, producing concise and clinically reliable diagnostic summaries. These results highlight diagnosis-driven contextual reasoning as a promising paradigm for ultra-long CE video summarization.