Analysis: Neuro
Traditional Poster
Analysis Methods
Tuesday, 12 May 2026
This session will highlight recent advances in the analysis of neuro MRI, with a focus on quantitative and AI-based methods. Contributions will span methodological developments and neuroimaging applications, reflecting current challenges and opportunities in the field.
|
470-08-132.
Gender-Dependent Structural Brain Changes in Adults: A Multi-Metric Structural MRI Analysis
Impact: By characterizing gender-dependent
brain differences using multiple structural metrics within the same group of
adults, this study provides a unified framework for comparison and highlights
the complex, distinct, and multi-dimensional organization of male and female
brains beyond single-metric structural analysis.
|
||
|
470-08-133.
Unsupervised k-means clustering reliably identifies BOLD Activations in Fast fMRI Data
Impact: It is possible to detect BOLD responses in an unsupervised manner in fMRI data with high temporal resolution. The k-means classification with 5 clusters was very easy to implement and provided reliable results in spatially specific locations.
|
||
|
470-08-134.
Toward the evaluation of Aβ plaque load in human brain with MRI at 3T
Impact: This research
develops an MRI method to image the beta-amyloid plaque load in the human brain
in vivo, which allows us to detect and evaluate AD pathological changes and
associated functional/structural neurodegeneration using a single imaging
modality.
|
||
|
470-08-135.
Beyond the Hand Knob: Automated Gyrus-Based Mapping of the Hand Motor Cortex for Precise VIM Targeting
Impact: Our automated gyrus‑based pipeline enables rapid, accurate, and personalized, functionally informed neuromodulation targeting using only structural MRI.
|
||
|
470-08-136.
Reducing Misdiagnosis in Disorders of Consciousness Via a Resting-State fMRI-based Hierarchical Brain Dynamics Network (HBDN)
Impact: HBDN surpasses state-of-the-art machine learning frameworks by modeling multi-scale
neural dynamics, yielding higher accuracy and stability. HBDN’s semi-supervised
strategy effectively leverages weakly labeled data—minimizing label noise while
maximizing data utilization.
|
||
|
470-08-137.
Comparing Deep Learning Denoising Models for Preserving Diagnostic Value in Neuromelanin-Sensitive MRI of the Locus Coeruleus
Impact: This work advances accessible neuromelanin MRI for neurodegenerative diseases through deep learning denoising, which reduces acquisition time and preserves diagnostic locus coeruleus contrast. It establishes a benchmark for assessing biomarker reliability, highlighting the importance of diagnostic fidelity in AI-driven neuroimaging.
|
||
|
470-08-138.
Cortex MAE: Learning Generalizable Morphological Representations of Cortical Surface via Deformation-Aware Masked Autoencoder
Impact: Our Cortex MAE pretraining learns fine-grained cortical features via deformation-based masked autoencoding, yielding more accurate and generalizable representations for downstream tasks, and may facilitate surface-based analyses in neurodevelopmental assessment and clinical neuroscience.
|