Predictive and discriminative localization of IDH genotype in high grade gliomas using deep convolutional neural nets

Citations Over TimeTop 19% of 2019 papers

Abstract

Radiomics and state-of-art convolutional neural networks (CNNs) have demonstrated their usefulness for predicting genotype in gliomas from brain MRI images. However, these techniques rely heavily on accurate tumor segmentation and do not facilitate any insights into the working of CNN to understand what areas distinguish these classes. To mitigate this, we employ a novel technique called Convolutional Neural Nets with discriminative localization (DL-CNN) on a clinical T2 weighted MRI dataset of IDH1 mutant and wild-type tumor patients. The technique not only is free of tumor segmentation with high classification accuracy of 86.7% but also locates the most discriminative regions. We demonstrate that in majority IDH1 mutants only the tumoral area is significant while in majority IDH1 wildtype the peri-tumoral edema is also involved. Overall, our method besides prediction provides information that is particularly important for clinical interpretability and can be used in targeted therapy.

Related Papers

• → ML interpretability: Simple isn't easy(2024)29 cited
• → When consumers need more interpretability of artificial intelligence (AI) recommendations? The effect of decision-making domains(2023)7 cited
• → Measures of Model Interpretability for Model Selection(2018)10 cited
• → Measuring Interpretability for Different Types of Machine Learning Models(2018)15 cited
• → ML Interpretability: Simple Isn't Easy(2022)2 cited