HJAR Jul/Aug 2024

CTE 14 JUL / AUG 2024 I  HEALTHCARE JOURNAL OF ARKANSAS total neural signaling power, while ReHo gauges the coherence of neighboring neu- ral activity. In this cohort study, we applied these advanced neuroimaging techniques to uncover cortical structure and neurophys- iological differences between high school football players and controls participating in noncontact sports. Our hypothesis was that compared with controls, the football group would exhibit (1) lesser cortical thickness and gyrification and greater sulcal depth, (2) lower localized neural activity (ALFF) and coherence of neural signals (ReHo) and (3) altered RS-FC involving the dorsolateral prefrontal cortex (DLPFC). The selection of the DLPFC as a seed is grounded in its role as a neural network hub, supported by pre- vious studies demonstrating altered activa- tion patterns in the DLPFC after a football season 19 and the presence of neurofibrillary tangle in this region among young adults with chronic traumatic encephalopathy. 3 METHODS Participants This cohort study included male high school football players and male high school noncontact control athletes from 5 high schools in the Midwest. The data col- lection was conducted during preseason before the 2021 and 2022 seasons. Inclu- sion criteria were being current members of the high school football team or noncon- tact sports team (swimming, cross country, and tennis) and being between the ages of 13 and 18 years. Control participants were matched to football participants with sex (male), age, and school they attend. Exclu- sion criteria were a history of moderate-to- severe TBI, organized contact sports experi- ence for the control athletes, and any MRI contraindications. A history of mTBI or concussion was permitted if asymptom- atic for the past 6 months. Race and eth- nicity data were based on self-report and were assessed in this study because provid- ing racial data will help interpret the neuro- logical outcomes in relation to head impact burden from football participation. All par- ticipants and their legal guardians provided informed consent, and the Indiana Univer- sity institutional review board approved the study protocol. This study followed the Strengthening the Reporting of Observa- tional Studies in Epidemiology (STROBE) reporting guideline. MRI Data Acquisition The MRI data were acquired on a 3T Prisma MRI scanner (Siemens) equipped with a 64-channel head/neck coil. Corti- cal morphometry measures were based on high-resolution anatomical images (T1-weighted) acquired using 3D MPRAGE pulse sequence with the following parame- ters: repetition time (TR) and echo time (TE), 2400/2.3 ms; inversion time, 1060 ms; flip angle, 8; matrix, 320×320; bandwidth, 210 Hz/pixel; integrated parallel acquisition technique, 2, which resulted in 0.8-mm iso- tropic resolution. Resting-state blood oxy- gen level–dependent (BOLD) signal was collected using a simultaneous multislice, single-shot echo-planar imaging sequence: TR/TE, 800/30 ms; flip angle, 52°; matrix, 90×90; field-of-view, 216 mm; resolution, 2.4 mm isotropic; and multiband accel- eration factor, 6, with 1000 total volumes acquired over 12 minutes while the partici- pant passively viewed a crosshair. Cortical Morphometry Preprocessing and Analyses The detailed preprocessing workflow is described in the eMethods in Supplement 1. The Computational Anatomy Toolbox 12v (CAT12) version 6 (Structural Brain Map- ping Group) was used for the T1-weighted MRI data preprocessing. The preprocess- ing consisted of bias-field correction, skull- stripping, and alignment to the Montreal Neurological Institute structural template to classify gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF). Spa- tial normalization was conducted with the Diffeomorphic Anatomical Registration Through Exponentiated Lie Algebra reg- istration (1.5 mm). 20 A spherical harmonic method was used to reparametrize the cor- tical surface mesh. 20,21 A voxel-based dis- tance method was used to estimate theWM segment by calculating the distance from the inner GM boundary. Values at the outer GMboundary in theWMdistance map were projected back to the inner GM boundary to generate the GM thickness. The cortical thickness data were spatially smoothed with a Gaussian kernel with a 15 mm full-width at half-maximum (FWHM). The gyrification estimates cortical fold complexity based on spherical harmonics and was calculated as absolute mean curvature. 22 The sulcal depth is calculated as the Euclidean distance between the central surface and its convex hull based on the spherical harmonics, then transformed with the sqrt function. 22 Preprocessing for Surface-Based RS-fMRI Analysis The workflow of the preprocessing for the surface-based RS-fMRI analysis is extensive, as described in the eMethods in Supplement 1. The brief summary of steps is as follows: (1) a reference volume and its skull-stripped version were generated using fMRIPrep; (2) the BOLD reference was coregistered to the T1 image; (3) participants with head motions beyond a frame-wise displacement of more than 3.0 mm and 3.0 degrees were excluded from fMRI analyses; and (4) the BOLD time-series were resam- pled onto their original, native space. ALFF and ReHo Analyses ALFF and ReHo analyses were conducted using DPABISurf version 1.2 (R-fMRI Net- work). 17,23 For ALFF, the resampled func- tional images were spatially smoothed with a FWHM of 6 mm. The square root of the power spectrumwas calculated to obtain a rawALFF map. ALFF values for each voxel were divided by the global mean ALFF value for standardization. Surface-based ReHo maps were produced by calculating the concordance of the Kendall coefficient of the time series of a given vertex in the surface space with nearest neighbors. This computational approach was repeated for