Multiscale Neuroimaging Features for the Identification of Medication Class and Non-Responders in Mood Disorder Treatment

TL;DR

This study targets predicting medication class (antidepressant vs mood stabilizer) and non-responders in mood disorders using resting-state fMRI. It introduces a multi-scale Neuromark template with 105 intrinsic connectivity networks extracted via group multi-scale ICA, combined with spatially constrained ICA to compute functional network connectivity, and a kernel SVM classifier with a Riemannian PABS kernel. A soft sequential forward selection (SSFS) process enhances feature selection across multiple domains, improving AUC and F1 metrics and revealing novel multi-scale ICNs as potential biomarkers. The approach demonstrates that multi-scale neuroimaging features, especially when paired with SSFS and FNC features, provide robust biomarkers for medication-response prediction, enabling faster, data-driven guidance on mood disorder pharmacotherapy.

Abstract

In the clinical treatment of mood disorders, the complex behavioral symptoms presented by patients and variability of patient response to particular medication classes can create difficulties in providing fast and reliable treatment when standard diagnostic and prescription methods are used. Increasingly, the incorporation of physiological information such as neuroimaging scans and derivatives into the clinical process promises to alleviate some of the uncertainty surrounding this process. Particularly, if neural features can help to identify patients who may not respond to standard courses of anti-depressants or mood stabilizers, clinicians may elect to avoid lengthy and side-effect-laden treatments and seek out a different, more effective course that might otherwise not have been under consideration. Previously, approaches for the derivation of relevant neuroimaging features work at only one scale in the data, potentially limiting the depth of information available for clinical decision support. In this work, we show that the utilization of multi spatial scale neuroimaging features - particularly resting state functional networks and functional network connectivity measures - provide a rich and robust basis for the identification of relevant medication class and non-responders in the treatment of mood disorders. We demonstrate that the generated features, along with a novel approach for fast and automated feature selection, can support high accuracy rates in the identification of medication class and non-responders as well as the identification of novel, multi-scale biomarkers.

Figures (3)

• Figure 1: Performance metrics of the Kernel SVM for different feature sets. We evaluated the original NeuroMark template (n53-Blue) and the best-matching components from the multiscale template (n105-Green). We then evaluated the performance of each template using SSFS (NeuroMark: n53*-Orange, MultiScale: n105*-Red). For each of these feature sets we evaluated the performance of utilizing spatial maps (SM) alone and spatial maps along with static FNC matrices (SM+sFNC). Performance metrics were aggregated over 1000 randomly initialized models and 5-fold cross validation.
• Figure 2: ICNs computed using sequential feature selection in both templates. We visualize each ICN in different colors corresponding to Neuromark domains.
• Figure 3: ICNs selected for each template using soft forward selection. Each of the ICNs are visualized in different colors corresponding to the domains for the corresponding template.