Improving CBCT diagnosis with pathological descriptors in radiology training

This study assessed (1) CBCT interpretation improvement in Oral Radiology residents, (2) whether accurate pathological descriptor selection improved differential diagnosis, and (3) which descriptor categories best predicted diagnostic accuracy.

Three residents interpreted 200 CBCTs with known pathologies during a 24-month residency (2013-2015) at the University of Washington. Sixty-seven cases were preselected based on confirmed pathology and diversity. Residents selected descriptors from 101 options and provided up to 5 differential diagnoses, compared to a senior radiologist's gold standard. Spearman's Rho and Kruskal-Wallis tests evaluated correlations (P < .05).

Residents significantly improved in selecting accurate descriptors and diagnoses (P < .05). Key predictors of correct diagnoses-location, size, and dentoalveolar involvement-correlated consistently across all residents (Rho > 0.6), while shape and border showed weaker associations.

Training with pathological descriptors enhances CBCT diagnostic accuracy. Identifying strengths and weaknesses in descriptor use can guide targeted training, improving diagnostic skills. This approach offers a scalable method for radiology education and clinical application.