EviDent-CBCT: Evidence-Bottlenecked Report Generation from Dental CBCT under Non-Exhaustive Report Supervision

2026-10-01

AI Summary

This study addresses the image–text inconsistency problem in dental CBCT report generation caused by non-exhaustive supervision. To this end, we propose an evidence bottleneck framework that employs an anatomy-aware network to extract discrete evidence records, which are subsequently rectified via logical consistency projection before report synthesis by a language model. Notably, a reliability-aware training strategy is introduced to handle unannotated findings, while metal-sensitive input channels and an auditable discrete interface further enhance robustness. The proposed method achieved second place in automatic evaluation and third place in clinical blind review at the ODIN 2026 Challenge, significantly outperforming baseline approaches.

Abstract

Dento-maxillofacial cone-beam CT (CBCT) reports may contain dozens of tooth-specific, anatomical, and spatial findings from a single 3D scan. Learning to generate such reports from limited clinical data is challenging because routine reports may not exhaustively document image findings, and a non-mention may reflect either absence or non-reporting. We present EviDent-CBCT, an evidence-bottlenecked framework designed for this incomplete supervision. An anatomy-aware network maps each CBCT scan to a discrete record of tooth-level, global, and tooth-IAC evidence. A dental-logic consistency projection reconciles incompatible evidence before a deterministic renderer and an image-blind local language model generate the report using only this record. For tooth-level evidence, reliability-aware training uses eligible non-mentions as reduced-weight negatives, while unreported global and tooth-IAC labels remain unknown. A metal-sensitive input channel preserves intensity cues from dental materials. Across three validation runs, EviDent-CBCT achieves $0.666\pm0.006$ merged evidence set-F1 and $0.402\pm0.003$ RadFact-Lite-Dental logical-F1, versus $0.371\pm0.018$ for the strongest controlled direct baseline. In the ODIN 2026 challenge, it ranked second in automated evaluation and third in blinded clinical Arena comparison on the hidden test set. These results support the discrete evidence record as an effective and auditable interface for CBCT report generation.

Problem

• Dental CBCT
• Report generation
• Non-exhaustive supervision
• Incomplete labels
• 3D medical imaging

Innovation

• Evidence-Bottlenecked Framework
• Non-Exhaustive Supervision
• Reliability-Aware Training
• Dental-Logic Consistency Projection
• CBCT Report Generation