Reproducibility of Radiomic Features in Glial Brain Tumors.

In our recent research, we have effectively demonstrated the feasibility of classifying magnetic resonance images (MRI) of glial tumors into four histological types utilizing standardized volume of interest (VOI), radiomics and machine learning. This research aims to determine the reproducibility of our approach when the locations of VOI are changed. We were able to demonstrate high reproducibility of ML results when the same feature selection methodology was employed across different VOIs. However, the reproducibility of radiomic features and their sets among various VOIs was not ensured for the sample size (n = 85) studied. The limited reproducibility of radiomic features should be taken into account when evaluating radiomics studies in glial tumors.

Radiomics Enhances Diagnostic and Prognostic Value of Diffusion Kurtosis Imaging in Diffuse Axonal Injury.

The aim of our study was to investigate the potential of advanced radiomics in analyzing diffusion kurtosis MRI (DKI) to increase the informativeness of DKI in diffuse axonal injury (DAI). We hypothesized that DKI radiomic features could be used to detect microstructural brain injury and predict outcomes in DAI. The study enrolled 31 patients with DAI (mean age 31.48 ± 11.10 years, 8 (25.8%) female) and 12 healthy volunteers (mean age 33.67 ± 11.06 years, 4 (33.3%) female). A total of 342,300 radiomic features were calculated (2282 features per each combination of 10 parametric DKI maps with 15 ROIs). Our results showed that several radiomic features were capable of distinguishing between healthy and injured brain tissue and accurately predicting outcomes with an accuracy of over 0.9. Advanced DKI radiomic features show high diagnostic and prognostic potential in DAI and may outperform average ROI values in DKI maps.

Noninvasive Glioma Grading with Deep Learning: A Pilot Study.

Gliomas are the most common neuroepithelial brain tumors, different by various biological tissue types and prognosis. They could be graded with four levels according to the 2007 WHO classification. The emergence of non-invasive histological and molecular diagnostics for nervous system neoplasms can revolutionize the efficacy and safety of medical care and radically reduce healthcare costs. Our pilot study aimed to evaluate the diagnostic accuracy of deep learning (DL) in subtyping gliomas by WHO grades (I-IV) based on preoperative magnetic resonance imaging (MRI) from Burdenko Neurosurgery Center's database. A total of 707 MRI studies was included. A "3D classification" approach predicting tumor type for the entire patient's MRI data showed the best result (accuracy = 83%, ROC AUC = 0.95), consistent with that of other authors who used different methodologies. Our preliminary results proved the separability of MR T1 axial images with contrast enhancement by WHO grade using DL.