Title:Exploring Adult Glioma through MRI: A Review of Publicly Available Datasets to Guide Efficient Image Analysis

Abstract:Publicly available data is essential for the progress of medical image analysis, in particular for crafting machine learning models. Glioma is the most common group of primary brain tumors, and magnetic resonance imaging (MRI) is a widely used modality in their diagnosis and treatment. However, the availability and quality of public datasets for glioma MRI are not well known. In this review, we searched for public datasets for glioma MRI using Google Dataset Search, The Cancer Imaging Archive (TCIA), and Synapse. A total of 28 datasets published between 2005 and May 2024 were found, containing 62019 images from 5515 patients. We analyzed the characteristics of these datasets, such as the origin, size, format, annotation, and accessibility. Additionally, we examined the distribution of tumor types, grades, and stages among the datasets. The implications of the evolution of the WHO classification on tumors of the brain are discussed, in particular the 2021 update that significantly changed the definition of glioblastoma. Additionally, potential research questions that could be explored using these datasets were highlighted, such as tumor evolution through malignant transformation, MRI normalization, and tumor segmentation. Interestingly, only two datasets among the 28 studied reflect the current WHO classification. This review provides a comprehensive overview of the publicly available datasets for glioma MRI currently at our disposal, providing aid to medical image analysis researchers in their decision-making on efficient dataset choice.