Page 91 - MEGIN Book Of Abstracts - 2023

sifier has never seen. Using LOOCV, we found the best METHODS We used source-imaged magnetoencepha-
average classification accuracy was reduced to 83.80% lography (MEG) and an established visuospatial pro-
using the same set of neuromarkers obtained from the cessing task to elicit multi-spectral neuronal responses
ROI group with functional hyper-connections. This per- in 35 biomarker-confirmed patients on the AD spec-
formance surpassed the results reported using wavelet- trum and 20 biomarker-negative older adults. Neuronal
based features by approximately 15%. Overall, our work oscillatory responses were imaged to the level of the
suggests that (1) certain ROIs are particularly effective cortex, and group classifications and neurocognitive
for MCI detection, especially when multi-resolution relationships were modeled using logistic and linear
wavelet biomarkers are employed for such diagnosis; regression, respectively.
(2) there exists a significant performance difference in
system evaluation between research-based experimen- RESULTS Visuospatial neuronal oscillations in the
tal design and clinically accepted evaluation standards. theta, alpha, and gamma ranges significantly predicted
the classification of patients on the AD spectrum.
Keywords: MCI detection, MEG, connectivity coherence, Importantly, the direction of these effects differed by
hyperconnectivity, hypoconnectivity, wavelet-based response frequency, such that patients on the AD spec-
neuromarker trum exhibited weaker alpha-frequency responses in
lateral occipital regions, and stronger gamma-frequen-
Sensors (Basel, Switzerland) (2021), Vol. 21, No. 18 cy responses in the primary visual cortex, as compared
(34577423) (1 citation) to biomarker-negative older adults. In addition, alpha
and gamma, but not theta, oscillations robustly predict-
ed cognitive status (i.e., MoCA and MMSE scores), such
Visuospatial alpha and gamma oscillations scale that patients with neural responses that deviated more
with the severity of cognitive dysfunction in from those of healthy older adults exhibited poorer
patients on the Alzheimer's disease spectrum (2021) cognitive performance.

Wiesman, Alex I; Murman, Daniel L; May, Pamela E; CONCLUSIONS We find that the multi-spectral neural
Schantell, Mikki; Wolfson, Sara L; Johnson, Craig M; dynamics supporting visuospatial processing differ-
Wilson, Tony W entiate patients on the AD spectrum from cognitively
normal, biomarker-negative older adults. Oscillations
Department of Neurological Sciences, University of Nebraska in the alpha and gamma bands also relate to cognitive
Medical Center, Omaha, NE, USA. alexander.wiesman@ status in ways that are informative for emerging clinical
mail.mcgill.ca; Memory Disorders & Behavioral Neurology interventions.
Program, UNMC, Omaha, NE, USA; Institute for Human Neu-
roscience, Boys Town National Research Hospital, Omaha, Keywords: Alzheimer’s disease, Magnetoencephalog-
NE, USA; Geriatrics Medicine Clinic, UNMC, Omaha, NE, USA; raphy, Neural oscillations, Source imaging, Visuospatial
Department of Radiology, UNMC, Omaha, NE, USA processing

BACKGROUND Entrainment of neural oscillations in Alzheimer's research & therapy (2021), Vol. 13, No. 1
occipital cortices by external rhythmic visual stimuli (34404472) (7 citations)
has been proposed as a novel therapy for patients with
Alzheimer's disease (AD). Despite this increased interest
in visual neural oscillations in AD, little is known regard-
ing their role in AD-related cognitive impairment and in
particular during visuospatial processing.

 ontents  Index 70
C