Feasibility of Localizing Epileptogenic Tissue with Naturalistic Stimulation in fMRI
Poster No:
2023
Submission Type:
Abstract Submission
Authors:
Ali Tafakkor1,2, Hana Abbas1,2, Caroline Chadwick1,2, Ali Khan1,3,4, Ingrid Johnsrude1,2
Institutions:
1Brain and Mind, University of Western Ontario, London, Ontario, Canada, 2Department of Psychology, University of Western Ontario, London, Ontario, Canada, 3Robarts Research Institute, The University of Western Ontario, London, Ontario, Canada, 4Schulich School of Medicine & Dentistry, The University of Western Ontario, London, Ontario, Canada
First Author:
Ali Tafakkor
Brain and Mind, University of Western Ontario|Department of Psychology, University of Western Ontario
London, Ontario, Canada|London, Ontario, Canada
Brain and Mind, University of Western Ontario|Department of Psychology, University of Western Ontario
London, Ontario, Canada|London, Ontario, Canada
Co-Author(s):
Hana Abbas
Brain and Mind, University of Western Ontario|Department of Psychology, University of Western Ontario
London, Ontario, Canada|London, Ontario, Canada
Brain and Mind, University of Western Ontario|Department of Psychology, University of Western Ontario
London, Ontario, Canada|London, Ontario, Canada
Caroline Chadwick
Brain and Mind, University of Western Ontario|Department of Psychology, University of Western Ontario
London, Ontario, Canada|London, Ontario, Canada
Brain and Mind, University of Western Ontario|Department of Psychology, University of Western Ontario
London, Ontario, Canada|London, Ontario, Canada
Ali Khan
Brain and Mind, University of Western Ontario|Robarts Research Institute, The University of Western Ontario|Schulich School of Medicine & Dentistry, The University of Western Ontario
London, Ontario, Canada|London, Ontario, Canada|London, Ontario, Canada
Brain and Mind, University of Western Ontario|Robarts Research Institute, The University of Western Ontario|Schulich School of Medicine & Dentistry, The University of Western Ontario
London, Ontario, Canada|London, Ontario, Canada|London, Ontario, Canada
Ingrid Johnsrude
Brain and Mind, University of Western Ontario|Department of Psychology, University of Western Ontario
London, Ontario, Canada|London, Ontario, Canada
Brain and Mind, University of Western Ontario|Department of Psychology, University of Western Ontario
London, Ontario, Canada|London, Ontario, Canada
Introduction:
One in 300 people suffers from medication-resistant epilepsy, a condition with profound cognitive and socio-economic consequences [1]. Surgery can be an effective treatment, hinging on accurate identification of lesions responsible for seizure generation. This can involve invasive electrode implantation, which is expensive and carries risk.
When people undergo fMRI while watching an engaging movie, synchronized blood oxygen level-dependent (BOLD) signal is observed broadly across the cortex, quantifiable via inter-subject correlation (ISC) [2]. This common neural synchronization may afford sensitivity to abnormality in people with epilepsy, offering safe and cost-effective localization. Specifically, epileptogenic tissue may show asynchronous BOLD patterns compared to controls. We provide proof-of-concept by demonstrating that people with epilepsy exhibit abnormalities of regional synchronization relative to controls.
When people undergo fMRI while watching an engaging movie, synchronized blood oxygen level-dependent (BOLD) signal is observed broadly across the cortex, quantifiable via inter-subject correlation (ISC) [2]. This common neural synchronization may afford sensitivity to abnormality in people with epilepsy, offering safe and cost-effective localization. Specifically, epileptogenic tissue may show asynchronous BOLD patterns compared to controls. We provide proof-of-concept by demonstrating that people with epilepsy exhibit abnormalities of regional synchronization relative to controls.
Methods:
Structural and functional MRI data were acquired using a 3 Tesla SIEMENS Prisma with a 32-channel head coil. T1-weighted MPR sequences had a slice thickness of 0.8 mm and 192 slices. fMRI acquisition used multi-band, accelerated EPI, acquisition (TR=1250 ms, slice thickness=2.5 mm, and 60 slices). In 47 participants, 384 volumes were acquired during an 8-minute film ('Bang! You're Dead', 1961). After excluding 10 datasets due to hearing loss, stimulus issues or missing data, 19 focal epilepsy patients under evaluation for surgery (20-60 years, 9 females) and 18 controls (19-58 years) remained in the study. Preprocessing was conducted using fMRIprep (v. 20.2.6) and FreeSurfer (v. 7.2), followed by confound removal, mapping to the fsLR 32k surface, and parcellation utilizing the multimodal Glasser atlas [3].
The subset of regions of interest (ROIs) that exhibit significant ISC within healthy subjects, and thus sensitivity to abnormality, were identified using a non-parametric test. ISCs were calculated for each ROI using a leave-one-out method, averaged across subjects (Fig.1A). To define a null distribution for absence of synchrony, subjects' time series were randomly circularly shifted 1000 times before leave-one-out averaging [4] (Fig.1A). Original ISCs were compared against null, with false discovery rate (FDR) correction at q<0.01 for all ROIs [5].
We identified regions with weaker BOLD synchrony in patients compared to within controls by subtracting patient-to-control ISC from within-control ISC. The within-control ISCs were calculated as explained, while patient-to-control ISCs were determined as the average over correlations of patients with the control group's average (Fig.1B). Statistical validation used a randomization test, comprising 10000 iterations with randomized subject labels for half of the controls and patients in null groups, thereby generating a null distribution for difference of within-control and patient-to-control ISCs (Fig.1B). FDR correction was applied at q<0.05 over ROIs with significant within-control ISC [5].
The subset of regions of interest (ROIs) that exhibit significant ISC within healthy subjects, and thus sensitivity to abnormality, were identified using a non-parametric test. ISCs were calculated for each ROI using a leave-one-out method, averaged across subjects (Fig.1A). To define a null distribution for absence of synchrony, subjects' time series were randomly circularly shifted 1000 times before leave-one-out averaging [4] (Fig.1A). Original ISCs were compared against null, with false discovery rate (FDR) correction at q<0.01 for all ROIs [5].
We identified regions with weaker BOLD synchrony in patients compared to within controls by subtracting patient-to-control ISC from within-control ISC. The within-control ISCs were calculated as explained, while patient-to-control ISCs were determined as the average over correlations of patients with the control group's average (Fig.1B). Statistical validation used a randomization test, comprising 10000 iterations with randomized subject labels for half of the controls and patients in null groups, thereby generating a null distribution for difference of within-control and patient-to-control ISCs (Fig.1B). FDR correction was applied at q<0.05 over ROIs with significant within-control ISC [5].
Results:
Significant synchrony was noted in early auditory and visual ROIs among control subjects during the movie-watching paradigm, extending to certain parietal and frontal ROIs. This indicates the method's broad coverage. Elevated within-control ISC in early sensory ROIs, compared to other regions with significant ISC, validated the synchrony induced by the paradigm (Fig.2A). Importantly, ROIs L_FOP2, R_STSva, R_A1, R_MBelt, R_LBelt, R_A4, R_TA2, R_PHA3 in left inferior frontal and right temporal lobe showed significantly higher ISC within controls compared to patient-to-control comparisons, despite the variability in patients' epileptogenic regions (Fig.2B).
Conclusions:
Our results suggest ISC analysis of naturalistic fMRI could identify abnormality in epileptic patients, although patient heterogeneity calls for individualized and vertex-level analysis. However, patient specific presumed seizure localization and surgical outcomes were not available, limiting us to a broader population analysis.
Disorders of the Nervous System:
Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) 2
Modeling and Analysis Methods:
Task-Independent and Resting-State Analysis 1
Keywords:
Epilepsy
Neurological
Other - Naturalistic Stimuli, Inter-Subject Correlation, Biomarker
1|2Indicates the priority used for review
Provide references using author date format
2. Hasson, Uri, Yuval Nir, Ifat Levy, Galit Fuhrmann, and Rafael Malach. 2004. “Intersubject Synchronization of Cortical Activity during Natural Vision.” Science 303 (5664): 1634–40.
3. Glasser, Matthew F., Timothy S. Coalson, Emma C. Robinson, Carl D. Hacker, John Harwell, Essa Yacoub, Kamil Ugurbil, et al. 2016. “A Multi-Modal Parcellation of Human Cerebral Cortex.” Nature 536 (7615): 171–78.
4. Tohka, Jussi. n.d. “Non-Parametric Test for Inter-Subject Correlations.” https://www.nitrc.org/docman/view.php/947/2017/parametric-test-inter.pdf.
5. Benjamini, Yoav, and Daniel Yekutieli. 2001. “The Control of the False Discovery Rate in Multiple Testing under Dependency.” Annals of Statistics 29 (4): 1165–88.