Brain Tumor Classification on MRI in Light of Molecular Markers

TL;DR

This work addresses predicting $1p/19q$ co-deletion status in brain gliomas from MRI to support treatment planning. It builds a from-scratch, 14-layer CNN with convolution stacking, dropout, and Gaussian noise, evaluated on a Kaggle MRI dataset with 12-fold cross-validation and compared against fine-tuned pre-trained models. The proposed model achieves a high F1 score of about 0.964 (precision ≈ 0.975, recall ≈ 0.963) and outperforms transfer-learning baselines, demonstrating robust performance on a small, imbalanced dataset. The study highlights the value of scratch-trained models and noise-regularization for radiogenomic tasks and discusses potential extensions to diffusion-based data generation and embedded deployment for clinical use.

Abstract

In research findings, co-deletion of the 1p/19q gene is associated with clinical outcomes in low-grade gliomas. The ability to predict 1p19q status is critical for treatment planning and patient follow-up. This study aims to utilize a specially MRI-based convolutional neural network for brain cancer detection. Although public networks such as RestNet and AlexNet can effectively diagnose brain cancers using transfer learning, the model includes quite a few weights that have nothing to do with medical images. As a result, the diagnostic results are unreliable by the transfer learning model. To deal with the problem of trustworthiness, we create the model from the ground up, rather than depending on a pre-trained model. To enable flexibility, we combined convolution stacking with a dropout and full connect operation, it improved performance by reducing overfitting. During model training, we also supplement the given dataset and inject Gaussian noise. We use three--fold cross-validation to train the best selection model. Comparing InceptionV3, VGG16, and MobileNetV2 fine-tuned with pre-trained models, our model produces better results. On an validation set of 125 codeletion vs. 31 not codeletion images, the proposed network achieves 96.37\% percent F1-score, 97.46\% percent precision, and 96.34\% percent recall when classifying 1p/19q codeletion and not codeletion images.

Figures (5)

• Figure 1: Brain Magnetic Resonance Image
• Figure 2: Network Architectures
• Figure 3: Proposed training procedure
• Figure 4: Evaluation matrix
• Figure 5: Confusion matrix