Glioblastoma Overall Survival Prediction With Vision Transformers

TL;DR

The study tackles glioblastoma overall survival prediction using MRI by introducing OS_ViT, a segmentation-free Vision Transformer that processes downsampled 3D MRI volumes into 3D patches, incorporates patient age, and classifies survival into three categories. The model avoids tumor segmentation, reducing preprocessing and computational demands while maintaining competitive accuracy (~62.5% on BRATS test data) and balanced precision/recall across classes. Key contributions include a 3D patch embedding strategy, a lightweight transformer with a two-layer encoder, and integration of clinical age as auxiliary information. While results are promising and clinically appealing due to its simplicity, the work is limited to BRATS data and requires external validation to establish robustness across institutions.

Abstract

Glioblastoma is one of the most aggressive and common brain tumors, with a median survival of 10-15 months. Predicting Overall Survival (OS) is critical for personalizing treatment strategies and aligning clinical decisions with patient outcomes. In this study, we propose a novel Artificial Intelligence (AI) approach for OS prediction using Magnetic Resonance Imaging (MRI) images, exploiting Vision Transformers (ViTs) to extract hidden features directly from MRI images, eliminating the need of tumor segmentation. Unlike traditional approaches, our method simplifies the workflow and reduces computational resource requirements. The proposed model was evaluated on the BRATS dataset, reaching an accuracy of 62.5% on the test set, comparable to the top-performing methods. Additionally, it demonstrated balanced performance across precision, recall, and F1 score, overcoming the best model in these metrics. The dataset size limits the generalization of the ViT which typically requires larger datasets compared to convolutional neural networks. This limitation in generalization is observed across all the cited studies. This work highlights the applicability of ViTs for downsampled medical imaging tasks and establishes a foundation for OS prediction models that are computationally efficient and do not rely on segmentation.

Figures (3)

• Figure 1: The structure of the OS ViT model is shown. It starts with raw MRI images and follows a structured pipeline to ensure effective feature extraction and classification.
• Figure 2: Overview of OS prediction accuracy, including results from 14 existing studies
• Figure 3: Confusion Matrix: 0 for Long-Term Survivors, 1 for Medium-Term Survivors, and 2 for Short-Term Survival