Multiparametric magnetic resonance imaging histogram and habitat analysis for predicting 1p/19q codeletion in isocitrate dehydrogenase-mutant gliomas

Abstract Background Quantitative T2-FLAIR analysis and multiparametric MRI habitat imaging may provide noninvasive information regarding 1p/19q status in IDH-mutant gliomas. The primary aim was to determine whether adding habitat-derived features to whole-region histogram features improves prediction of 1p/19q codeletion. We retrospectively analyzed 122 patients with World Health Organization grade 2 or 3 gliomas. Precontrast T1-weighted, postcontrast T1-weighted, T2-weighted, and FLAIR images were normalized using patient-specific normal-appearing white matter. Whole-region histogram, habitat-derived, and combined models were evaluated in whole-tumor and tumor-core-like regions using 20 repetitions of nested fivefold cross-validation. Results In the whole-tumor analysis, the whole-region histogram, habitat-derived, and combined models achieved areas under the receiver operating characteristic curve of 0.747, 0.681, and 0.734, respectively. The corresponding values in the tumor-core-like analysis were 0.777, 0.734, and 0.766. The whole-tumor habitat-derived model had a lower area under the curve than the whole-region histogram model, with a difference of − 0.066 (95% confidence interval, − 0.130 to − 0.003). The combined model did not improve discrimination in either region. Whole-tumor habitat centroids were more stable than tumor-core-like centroids. Conclusions Whole-region histogram features provided moderate discrimination of 1p/19q codeletion, whereas habitat-derived features did not provide incremental predictive value. T2-FLAIR difference and T2-weighted distributional features were selected most consistently. These findings support further evaluation of the simpler histogram approach in independent datasets but do not support replacement of tissue-based molecular testing.