Icon Legend

This session is not in your schedule.

This session is in your schedule. Click again to remove it.

Biomedical Imaging and Instrumentation

Biomedical Imaging and Instrumentation - Poster Session D

Poster G6 - Refinement of Functional Connectivity Evaluation in Glioblastoma Using Healthy Control Subjects

Friday, October 25, 2024

3:30 PM – 4:30 PM EST

Location: Exhibit Hall E, F & G

Co-Author(s)

DM

Derek Madden

Graduate Research Assistant
The University of Oklahoma, United States

Presenting Author(s)

Abigail Kauten (she/her/hers)

Undergraduate Research Assistant
The University of Oklahoma
Norman, Oklahoma, United States

Co-Author(s)

August Limvaree

Graduate Research Assistant
The University of Oklahoma, United States
MH

Mingwei Huang

Graduate Research Assistant
The University of Oklahoma, United States
TS

Tressie M. Stephens, PT, DPT

Affiliate
The University of Oklahoma Health Sciences Center, United States
AB

Andrew M. Bauer, MD, MBA

Assistant Professor
University of Oklahoma Health Sciences Center
Oklahoma City, Oklahoma, United States
ID

Ian F. Dunn, MD

Dean
The University of Oklahoma Health Sciences Center, United States

Primary Investigator(s)

HY

Han Yuan

Associate Professor
The University of Oklahoma, United States

Introduction: A deadly form of cancer, glioblastoma of grades III and IV results in five-year survival rates of only 5% and median overall survival between 12 and 16 months (Delgado-López & Corrales-García, 2016). With resection of cancerous tissue as the primary form of treatment, neurological morbidity or cognitive impairment can occur due to loss of functional brain tissue (Rahman et al., 2017; Sanai & Berger, 2018). Though a multitude of techniques have been utilized to reveal tissue eloquence and preserve postsurgical function, functional connectivity measurements have been explored to further reveal tissue function (Fox & King, 2018; Tarapore et al., 2012). Specifically, functional connectivity in a glioma region of interest (ROI) is compared to functional connectivity in a mirrored, contralateral region in the opposite hemisphere to determine the level of abnormality of tissue connectivity (Tarapore et al., 2012). However, the brain is not perfectly symmetrical, and contralateral tissue does not serve as a perfect control for normal regional function. This study addresses this limitation by exploring hemispheric connectivity asymmetry in healthy control subjects of the Human Connectome Project Aging (HCP-A) dataset, with connectivity evaluated in ROIs defined by glioma regions in glioblastoma patients. Hemispheric differences in ROIs can be related to patient data, revealing levels of abnormality of connectivity differences in patient values.

Materials and

Methods: Healthy subject data were obtained from the HCP-A (Bookheimer et al., 2019). HCP-A subjects are neurologically healthy adults between 36 and 79 years. Resting-state functional magnetic resonance images (fMRI) and high-resolution MRI images were acquired for each subject with a 3T Siemens Prisma scanner.

The HCP data was condensed spatially and temporally to match the dimensions of brain glioma patient data acquired at the University of Oklahoma Health Sciences Center (Madden et al., Under Review). Then hemispheric connectivity differences were determined for each subject by first randomly selecting 2,000 voxels in each quadrant of the left hemisphere, totaling 8,000 seeds for evaluation. Each seed was centered at the randomly selected voxel and assigned a spherical mask of radius 5. The average time series of the mask was calculated and correlated with the time series of every voxel. This connectivity process was repeated in a mirrored contralateral seed region producing pairs of whole-brain connectivity maps. The average connectivity difference with the whole-brain were then calculated and associated with the original seed voxel. Finally, the average connectivity difference for each voxel were visualized in AFNI as a heat map.

Based on glioma ROIs defined from patient data, ROI masks were applied to the whole-brain connectivity difference heat maps in all healthy subjects to reveal connectivity difference. The mean and standard deviation of the control data were used to calibrate the connectivity values in patients.

Results, Conclusions, and Discussions:

Results: In a preliminary investigation, 27 healthy subjects were included to examine 2 glioblastoma patients (namely A and B). Using patient A’s glioma ROI, the mean and standard deviation connectivity difference were 0.0006 and 0.0102, respectively, based on the 27 healthy subjects. This patient exhibited an average connectivity difference of -0.0082, indicating a patient z-score of -0.8627 (p = 0.4). Thus, the patient exhibited a connectivity difference value within an expected range, predicting a favorable outcome of surgical resection. This patient exhibited minimal cognitive morbidity measured by a change of 0 in the KPS evaluation, indicating a favorable outcome.

For another patient B’s ROI, the mean and standard deviation connectivity difference were 0.0003 and 0.0073, respectively, based on the 27 healthy subjects. The patient was revealed to have an average connectivity difference of -0.0491, indicating a patient z-score of -6.7671 (p < 0.001). Indeed, this patient exhibited neurological morbidity measured by a 90 point decrease in the KPS evaluation, which was suggested by the extreme connectivity difference value.

Discussions/

Conclusions: Functional connectivity is a promising method to evaluate and predict outcomes of surgical resection in glioblastoma, but such measurements require control values to reveal expected connectivity values in a given region. Though hemispheric contrast is a good start, evaluation of healthy control subjects is necessary to understand the abnormality of functional connectivity differences between hemispheres in glioma patients. In this preliminary study, we have presented both the possibility and promise of using HCP-A healthy control subjects to determine abnormality and predict surgical outcomes. Going forward, this analysis will be greatly expanded, incorporating 500 healthy subjects for patient evaluation.

Acknowledgements (Optional): This work was supported in part by NIH NIGMS P20GM135009, P20GM103447-24S1, NSF RII Track-4 213182, and Undergraduate Research Opportunities Program at The University of Oklahoma.