Classification of migraine with aura based on grey matter structure within the visual networks

1439 

Abstract Submission 

David Niddam¹, Kuan-Lin Lai², Shuu-Jiun Wang²

¹National Yang Ming Chiao Tung University, Taipei, Taiwan, ²Taipei Veterans General Hospital, Taipei, Taiwan

David Niddam, Prof  
National Yang Ming Chiao Tung University
Taipei, Taiwan

Kuan-Lin Lai  
Taipei Veterans General Hospital
Taipei, Taiwan

Shuu-Jiun Wang  
Taipei Veterans General Hospital
Taipei, Taiwan

Migraine auras are transient neurological symptoms that typically occur before a headache attack. The occipital cortex is thought to have a pivotal role in the initiation of visual auras, as cortical spreading depression, the mechanism underlying visual auras, arises within this region (Hadjikhani et al., 2001). Interictal changes in resting-state functional connectivity and grey matter structure within the occipital cortex has also been observed in migraine with visual auras (Karsan et al., 2023; Niddam et al., 2016).

Previous neuroimaging studies on migraine auras have primarily used mass-univariate analyses in which each voxel is considered as a spatially independent unit. Multivariate pattern analysis is a complementary approach in which information contained jointly among multiple voxels is taken into account at once. Multivariate pattern classification can identify distributed patterns of voxels which can be used to differentiate groups of participants and to make predictions about diagnoses at the individual level.

Since the occipital cortex is involved in the generation of visual aura and may exhibit altered interictal grey matter structure and functional connectivity, we hypothesized that grey matter within the functionally defined resting-state visual networks contains discriminative information that can be used to differentiate migraine patients with visual auras (MA) from migraine patients without visual auras (MO) and healthy controls (HC).

Structural and functional resting-state MRI images were obtained from 50 MA patients, 50 MO patients and 50 HCs. All patients were in the interictal state and had low-frequency episodic migraine. Independent component analysis with 40 components was first used in each of the three groups to identify the functional visual networks (see Figure). The resulting images were thresholded and binarized. These binarized images were then used to constrain the structural images. The masked images entered a multivariate analysis in which Gaussian process classification was used to generate pair-wise models (see Figure). The performance of the models was indexed by the balanced accuracy (BA) and the area under the receiver operating characteristic curve (AUC). Generalizability was assessed by 5-fold cross-validation and non-parametric permutation tests were used to estimate significance levels. Only results passing a false discovery rate corrected threshold for the multiple models set up were considered significant.

Five functional networks were identified within the occipital cortex. Of these, one corresponded to the occipital visual network and one corresponded to the lateral visual network (Laird et al., 2011). The remaining 3 networks covered the anterior and the posterior dorsal and ventral medial visual networks (Laird et al., 2011). The multivariate pattern of grey matter voxels within the ventral medial visual network contained significant information related to the MA diagnosis (MA vs. HC: BA=78% [P<0.001]; AUC=0.84 [P<0.001]; MA vs. MO: BA=71% [P<0.001]; AUC=0.73 [P=0.003]). Grey matter voxels with the anterior medial visual network also contained significant information related to the MA diagnosis, albeit less significant (MA vs. HC: BA=67% [P=0.004]; AUC=0.70 [P=0.005]; MA vs. MO: BA=68% [P=0.001]; AUC=0.79 [P<0.001]).

Migraine with visual aura is characterized by multivariate patterns of grey matter changes within the medial visual cortex that have discriminative power and may reflect pathological mechanisms. The medial visual cortex is known to processes simple visual stimuli, both static and moving (Laird et al., 2011). This is congruent with common symptoms associated with visual aura.

Classification and Predictive Modeling ¹

Perception: Pain and Visceral

Perception: Visual ²

Headache

Machine Learning

Multivariate

Neurological

Pain

STRUCTURAL MRI

Other - Migraine aura