CMVFT: A Multiscale Attention-Guided Framework for Enhanced Keratoconus Suspect Classification in Multiview Corneal...

To develop a multiview fusion framework that effectively identifies suspect keratoconus cases and facilitates the possibility of early clinical intervention.

Retrospective cross-sectional study.

A total of 573 corneal topography maps representing eyes classified as normal, suspect, or keratoconus.

We designed the corneal multiview fusion transformer (CMVFT), which integrates features from seven standard corneal topography maps. A pretrained ResNet-50 extracts single-view representations that are further refined by a custom-designed multiscale attention module (MSAM). This integrated design specifically compensates for the representation gap commonly encountered when applying Transformers to small-sample corneal topography datasets by dynamically bridging local convolution-based feature extraction with global self-attention mechanisms. A subsequent fusion Transformer then models long-range dependencies across views for comprehensive multiview feature integration.

The primary measure was the framework's ability to differentiate suspect cases from normal and keratoconus cases, thereby creating a pathway for early clinical intervention.

Experimental evaluation demonstrated that CMVFT effectively distinguishes suspect cases within a feature space characterized by overlapping attributes. Ablation studies confirmed that both the MSAM and the fusion Transformer are essential for robust multiview feature integration, successfully compensating for potential representation shortcomings in small datasets.

This study is the first to apply a Transformer-driven multiview fusion approach in corneal topography analysis. By compensating for the representation gap inherent in small-sample settings, CMVFT shows promise in enabling the identification of suspect keratoconus cases and supporting early intervention strategies, with prospective implications for early clinical intervention.