MultiCubeNet: Multitask deep learning for molecular subtyping and prognostic prediction in gliomas.

BACKGROUND

Gliomas, the most prevalent type of primary brain tumors, require precise molecular characterization for effective diagnosis and treatment. Despite advancements in radiomics, simultaneous prediction of key molecular markers, such as isocitrate dehydrogenase () mutation, co-deletion, and telomerase reverse transcriptase () promoter mutation, along with prognosis, remains challenging. We aimed to develop and validate a deep learning (DL) model capable of simultaneously predicting key genetic molecular markers and prognosis in gliomas.

METHODS

We conducted a retrospective analysis of 457 adult-type diffuse gliomas (193 training cohorts; 162 and 102 cases in SZS and The Cancer Genome Atlas (TCGA) validation cohorts, respectively). We developed MultiCubeNet, a multisequence, multiscale, multitask DL framework designed to predict mutation, co-deletion, promoter mutation, and prognosis. Model performance was benchmarked against conventional radiomics pipelines and neuroradiologist annotations. Classification accuracy was evaluated by the area under the receiver operating characteristic curve (AUC), with prognostic performance quantified using Harrell's concordance index (C-index).

RESULTS

The median age of the patients was 49 years, and 266 were men (58.2%). The model demonstrated high efficiency in the training set, achieving AUCs of 0.966 for mutation, 0.961 for co-deletion, and 0.851 for promoter mutation. In the external test set (SZS), the model maintained strong performance with AUCs of 0.877, 0.730, and 0.705 for mutation, co-deletion, and promoter mutation, respectively. The performance in TCGA cohort was less optimal, with AUCs below 0.8. The framework consistently matched or exceeded both radiomics pipelines and neuroradiologists in molecular marker identification. Survival analysis revealed significant prognostic stratification across all cohorts (C-index: 0.706-0.866).

CONCLUSIONS

MultiCubeNet, a multitask DL model leveraging multisequence and multiscale magnetic resonance imaging, demonstrated strong performance in predicting key molecular markers and prognosis in gliomas, thereby supporting personalized treatment approaches.