Rethinking Diagnostic Performance Metrics in Imaging Artificial Intelligence: Lessons from Caries Detection on Radiographs
Caries Research · 26.03.2026 · 1-14
DOI: 10.1159/000551765 ↗Kurz gefasst
Übersichtsarbeit in der Zeitschrift Caries Research, online erschienen am 26.03.2026. Verfasst von Carolina Ganss und 1 weiteren Autor:innen. Beteiligte Einrichtung: Marburg University, University Dental Medicine, Clinic for Operative Dentistry, Endodontics, and Pediatric Dentistry, Section for Cariology, Marburg,…. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 48 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Background: Artificial intelligence (AI) systems for radiographic caries detection are commonly evaluated using a small set of performance metrics, yet these measures are frequently reported and interpreted without the contextual information required for clinically meaningful conclusions. Summary: This narrative review focuses on performance metrics – confusion-matrix measures (sensitivity, specificity, accuracy, predictive values, F1 score and related summaries) and discrimination metrics (area under the receiver operating characteristic curve and area under the precision recall curve) – and discusses how their interpretation depends on explicit reporting of the disease definition (grading cut-offs), unit of analysis (tooth surface/patient), decision threshold(s), and prevalence/case-mix. We additionally cover complementary metric-based assessments, including calibration (calibration curves, expected calibration error, Brier score), decision-analytic metrics (decision curve analysis and cost-sensitive clinical loss), and robustness summaries (uncertainty and subgroup/worst-group performance). A structured summary is provided of formulae, interpretive limitations, and the minimum information that should accompany each metric to support transparent reporting, comparability, and transportability. Key Messages: AI-assisted radiographic caries detection cannot be judged by headline metrics alone; traditional performance measures should be reported with clear disease definitions, operating points, prevalence, and the clinical impact of false results. We provide a practical guide to improving the transparency, comparability and clinically meaningful interpretation of future studies.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Marburg University, University Dental Medicine, Clinic for Operative Dentistry, Endodontics, and Pediatric Dentistry, Section for Cariology, Marburg, Germany.
- Department of Conservative Dentistry, Periodontology and Preventive Dentistry, Hannover Medical School (MHH), Hannover, Germany, vach.kirstin@mh-hannover.de.
- Center for Data Science, Eberswalde University for Sustainable Development, Eberswalde, Germany, vach.kirstin@mh-hannover.de.
Zitieren
APA
Ganss, C., & Vach, K. (2026). Rethinking Diagnostic Performance Metrics in Imaging Artificial Intelligence: Lessons from Caries Detection on Radiographs. Caries Research, 1-14. https://doi.org/10.1159/000551765
Vancouver
Ganss C, Vach K. Rethinking Diagnostic Performance Metrics in Imaging Artificial Intelligence: Lessons from Caries Detection on Radiographs. Caries Research. 2026:1-14. doi:10.1159/000551765
BibTeX
@article{Ganss2026,
author = {Ganss, Carolina and Vach, Kirstin},
title = {Rethinking Diagnostic Performance Metrics in Imaging Artificial Intelligence: Lessons from Caries Detection on Radiographs},
journal = {Caries Research},
year = {2026},
pages = {1--14},
doi = {10.1159/000551765}
}
EUVIMED