Radiomics Analysis for Predicting Medication‐Related Osteonecrosis of the Jaw Using Computed Tomography
Oral Diseases · 2026-10 · 32 (10) · 3256-3264
DOI: 10.1111/odi.70386 ↗At a glance
Journal article in Oral Diseases, published online on 2026-06-14. Written by Masaru Konishi. Institution: Department of Oral and Maxillofacial Radiology, Hiroshima University Hospital, Hiroshima, Japan. Major MEDLINE topics: Radiomics, Bisphosphonate-Associated Osteonecrosis of the Jaw, Tomography, X-Ray Computed. Cited 1 times so far (Crossref); 34 references. The full text is freely available (open access, CC BY-NC 4.0).
Abstract (original language)
Objectives
To determine whether pretreatment radiomics can predict medication‐related osteonecrosis of the jaw (MRONJ).
Methods
Patients with mandibular MRONJ at Hiroshima University Hospital (2009–2024) who underwent MDCT before initiating bone resorption inhibitors were retrospectively identified. Records from 2009 to 2020 were also reviewed to identify patients who received these agents, had pretreatment MDCT, and remained free of MRONJ for at least 2 years. Two types of three‐dimensional segmentations were performed: one including cancellous bone alone and the other including cancellous bone plus the surrounding cortical bone. Radiomic features were extracted and quantified. Least absolute shrinkage and selection operator (LASSO) regression was employed to select effective informative features. Machine learning models — Random Forest (RF), Support Vector Machine (SVM), and Multilayer Perceptron (MLP) — were trained and evaluated using sensitivity, specificity, and area under the curve (AUC).
Results
This study included 94 patients who underwent MDCT imaging at our institution before receiving bone resorption inhibitors. The cohort included 44 patients with mandibular MRONJ and 50 without. In testing for the cancellous bone‐only group, RF, SVM, and MLP showed sensitivities of 0.568, 0.568, and 0.659; specificities of 0.720, 0.680, and 0.760; and AUCs of 0.741, 0.706, and 0.802, respectively. Using testing datasets from the group including both cancellous and mandibular border cortical bones, the results were: sensitivity: 0.818, 0.796, and 0.886; specificity: 0.843, 0.824, and 0.886; AUC: 0.931, 0.917, and 0.921, respectively.
Conclusions
Radiomics analysis of pretreatment MDCT images may accurately predict MRONJ risk before exposure to bone resorption inhibitors.
Subjects
MeSH terms (MEDLINE)
Author keywords
Authors and institutions
- Department of Oral and Maxillofacial Radiology, Hiroshima University Hospital, Hiroshima, Japan.
ORCID profiles: Masaru Konishi
Cite
APA
Konishi, M. (2026). Radiomics Analysis for Predicting Medication‐Related Osteonecrosis of the Jaw Using Computed Tomography. Oral Diseases, 32(10), 3256-3264. https://doi.org/10.1111/odi.70386
Vancouver
Konishi M. Radiomics Analysis for Predicting Medication‐Related Osteonecrosis of the Jaw Using Computed Tomography. Oral Diseases. 2026;32(10):3256-3264. doi:10.1111/odi.70386
BibTeX
@article{Konishi2026,
author = {Konishi, Masaru},
title = {Radiomics Analysis for Predicting Medication‐Related Osteonecrosis of the Jaw Using Computed Tomography},
journal = {Oral Diseases},
year = {2026},
volume = {32},
number = {10},
pages = {3256--3264},
doi = {10.1111/odi.70386}
}
EUVIMED