Segmenting pediatric optic pathway glioma from MRI using deep learning

Challenge

The challenge was to verify if an ML model based on brain MRI that automatically segments glioblastoma (GBM) in adults can be effectively used in automated segmentation of pediatric optic pathway gliomas (OPG), although OPGs have vastly different locations and characteristics than GBMs. The goal of the project was to improve the specificity of diagnostic process and therapy planning.

What we did

Our role was to conduct the whole experimental study, performed over two clinical datasets and involving quantitative, qualitative and statistical analysis. We have also set up and coordinated the cooperation between medical facilities. We leveraged our medical imaging analysis technology and delivered results in 3 months.

Results

The experiment indicated high agreement between automatically calculated and ground-truth volumetric measurements of the tumors and very fast operation of the proposed approach, both of which can increase the clinical utility of the suggested software tool. This rigorous experimental study was published in ‘Computers in Biology and Medicine’.

Image comes from: J. Nalepa, S. Adamski, K. Kotowski, S. Chelstowska, M. Machnikowska-Sokolowska, O. Bozek, A. Wisz, E. Jurkiewicz: Segmenting pediatric optic pathway gliomas from MRI using deep learning. Computers in Biology and Medicine, Volume 142, 2022, 105237, ISSN 0010-4825, https://doi.org/10.1016/j.compbiomed.2022.105237.