OraRefer：分割引导的多照片融合用于口腔癌筛查，具备患者级灵敏度保证
OraRefer: Segmentation-Guided Multi-Photograph Fusion for Oral Cancer Screening with Guaranteed Patient-Level Sensitivity
Rupam Mukherjee · Mehakdeep Singh · Shirin Dasgupta · Subhamoy Mandal
中文摘要
口腔癌及其可能先行的口腔潜在恶性疾患（OPMD）在口腔黏膜上可见，使智能手机照片成为有前景的筛查工具。用于此任务的深度学习模型对单张照片进行分类。然而，筛查转诊的是患者而非照片：一名患者通常被拍摄多次，转诊决策必须漏诊不超过设定比例的患者。提出 OraRefer，一种患者级筛查模型，用分割引导的基础模型读取每张照片，其类别决策以预测病变为条件，结合患者所有照片与风险因素，并通过具有有限样本灵敏度保证的阈值进行转诊。模型定位口腔本身，因此推理无需标注。在 714 名患者的 3000 张照片上，患者级融合达到 AUROC 0.919，优于基于注意力的多实例学习。在 90% 灵敏度目标下，OraRefer 检测出 95% 的 OPMD 或癌症患者，61 名癌症患者中漏诊 3 名，特异性为 0.610。由于阈值无需重新训练即可校准，每家诊所可在本地校准样本上设定自己的保证操作点。
关键要点
- 01问题：现有深度学习模型对单张照片分类，但筛查转诊对象是患者，需保证患者级灵敏度。
- 02方法：分割引导基础模型以预测病变为条件分类每张照片，融合患者多照片与风险因素，阈值具有限样本灵敏度保证。
- 03结果：714 名患者 3000 张照片上 AUROC 0.919，优于注意力多实例学习；90% 灵敏度目标下检出 95% OPMD/癌症患者，61 名癌症患者漏诊 3 名，特异性 0.610。
- 04局限：特异性仅 0.610，阈值需本地校准样本设定操作点。
解读
尚无解读。
原始英文摘要
arXiv:2611.08360v1 Announce Type: new Abstract: Oral cancer and the oral potentially malignant disorders (OPMD) that can precede it are visible on the oral mucosa, which makes smartphone photographs a promising tool for screening. Deep learning models for this task classify individual photographs. However, screening refers patients, not photographs: a patient is usually photographed several times, and the referral decision must miss no more than a set share of patients with disease. We propose OraRefer, a patient-level screening model that reads each photograph with a segmentation-guided foundation model that conditions its class decision on the predicted lesion, combines all photographs of a patient with risk factors, and refers through a threshold with a finite-sample sensitivity guarantee. The model localizes the mouth itself, so inference needs no annotation. On 3000 photographs of 714 patients, patient-level fusion reaches an AUROC of 0.919 and outperforms attention-based multiple-instance learning. At a 90% sensitivity target, OraRefer detects 95% of patients with OPMD or cancer and misses 3 of 61 patients with cancer, at a specificity of 0.610. Because the threshold is calibrated without retraining, each clinic can set its own guaranteed operating point on a local calibration sample.