发表机构
School of Computing, Newcastle University, UK（计算学院，新卡斯尔大学，英国）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究针对视网膜 OCT 分类中模型置信度不可信问题，提出将混合卷积-Transformer 编码器与 XGBoost 分类头及临床安全层配对的方法，在四类 OCT 上达 95.4%准确率，显著降低校准误差，是首个联合验证三种安全机制的 OCT 分类器。
AI 中文摘要
用于视网膜光学相干断层扫描（OCT）分类的深度模型报告了高精度，但很少报告其置信度是否可信——当错误但自信的读数延迟挽救视力的治疗时，这一差距至关重要。我们将混合卷积-Transformer 编码器与梯度提升（XGBoost）分类头以及三部分临床安全层配对：置信度校准、分布外（OOD）拒绝和逐预测不确定性标记。在四类 OCT（84,495 次扫描）上，该模型达到了 95.4%的准确率，同时将校准误差降低了 12 倍（预期校准误差，ECE = 0.0024），因此其报告的置信度跟踪其真实准确率。据我们所知，这是第一个联合验证所有三种安全机制的 OCT 分类器，具有公开权重和可重复的多种子评估。
英文摘要
Deep models for retinal optical coherence tomography (OCT) classification report high accuracy but rarely report whether their confidence can be trusted -- a gap that matters when a wrong-but-confident reading delays sight-saving treatment. We pair a hybrid convolutional-Transformer encoder with a gradient-boosting (XGBoost) classification head and a three-part clinical safety layer: confidence calibration, out-of-distribution (OOD) rejection, and per-prediction uncertainty flagging. On four-class OCT (84,495 scans) the model reaches 95.4% accuracy while cutting calibration error twelve-fold (expected calibration error, ECE = 0.0024), so the confidence it reports tracks its true accuracy. To our knowledge this is the first OCT classifier to validate all three safety mechanisms jointly, with public weights and reproducible multi-seed evaluation.
Comments4 pages, 2 figures, 4 tables. Code, model weights, and REST inference API are available on GitHub and HuggingFace
