Patrick H. Luckett, PhD¹, Ki Yun Park, BS¹˒⁸, John J. Lee, MD, PhD², Eric J. Lenze, MD⁴, Julie L. Wetherell, PhD⁵˒⁶, Lisa Eyler, PhD⁶, Abraham Z. Snyder, MD, PhD²˒⁷, Beau M. Ances, MD, PhD, MSc⁷, Joshua S. Shimony, MD, PhD², Eric C. Leuthardt, MD¹˒⁸˒⁹˒¹⁰˒¹¹˒¹²˒¹³

Abstract

Objective: Resting state functional MRI (RS-fMRI) enables mapping of function within the brain and is emerging as an efficient tool for pre-surgical evaluation of eloquent cortex. Models capable of reliable and precise mapping of resting state networks (RSN) with reduced scanning time would lead to improved patient comfort while reducing cost per scan. The aims of this study were to develop a deep 3D convolutional neural network (3DCNN) capable of voxelwise mapping of language (LAN) and motor (MOT) resting state networks (RSN) with minimal quantities of RS-fMRI data.

Methods: Imaging data was gathered from multiple ongoing studies at Washington University School of Medicine and other thoroughly characterized, publicly available data sets. All study participants (n = 2252 healthy adults) were cognitively screened and completed structural neuroimaging and RS-fMRI. Random permutations of RS-fMRI regions of interest were used to train a 3DCNN. After training, model inferences were compared using varying amounts of RS-fMRI data from the control data as well as five patients with glioblastoma multiforme.

Results: The trained model achieved 96% out-of-sample validation accuracy on data encompassing a large age range collected on multiple scanner types and varying sequence parameters. Testing on out-of-sample control data showed 97.9% similarity between results generated using either 50 or 200 RS-fMRI time points, corresponding to approximately 2.5 and 10 minutes respectively (96.9% LAN, 96.3% MOT true positive rate). In evaluating data from patients with brain tumors, the 3DCNN was able to accurately map LAN and MOT networks in spite of structural and functional alterations.

Conclusion: Functional maps produced by the 3DCNN can inform surgical planning in patients with brain tumors in a time-efficient manner. We present a highly efficient method for pre-surgical functional mapping, hence, improved functional preservation in patients with brain tumors.