Observing Deep Radiomics for the Classification of Glioma Grades

Abstract

Deep learning is promising for medical image analysis because it can automatically acquire meaningful representations from raw data. However, a technical challenge lies in the difficulty of determining which types of internal representation are associated with a specific task, because feature vectors, which collectively constitute the feature maps, can vary dynamically according to individual inputs. Therefore, based on the magnetic resonance imaging (MRI) of gliomas, we propose a novel technique to extract a shareable set of feature vectors that encode various parts in tumor imaging phenotypes. Because the set of feature vectors is shared across a population, it can be used in other downstream tasks as if these feature vectors are imaging markers. Then, based on the feature vectors, a classifier is established using logistic regression to predict the glioma grade, and an accuracy of 90% is achieved. Besides, we develop an algorithm to visualize the image region encoded by each feature vector, and demonstrate that the classification model preferentially relies on feature vectors associated with the presence or absence of contrast enhancement in tumor regions. The proposed method provides a data-driven approach to enhance the understanding of physicians on the imaging appearance of gliomas.

Developing and validating a deep learning and radiomic model for glioma grading using multiplanar reconstructed magnetic resonance contrast-enhanced T1-weighted imaging: a robust, multi-institutional study

“…In addition, some studies have examined glioma grading using deep learning imaging, and we noted that these studies use the BraTs dataset (42)(43)(44)(45), an open challenge MRI glioma dataset. The dataset contains T1, gd-enhanced T1, T2, and FLAIR sequences of LGG or HGG patients, and these datasets are pre-processed.…”