Imaging the Neural Substrate of Trigeminal Neuralgia Pain Using Deep Learning

Abstract

Trigeminal neuralgia (TN) is a severe and disabling facial pain condition and is characterized by intermittent, severe, electric shock-like pain in one (or more) trigeminal subdivisions. This pain can be triggered by an innocuous stimulus or can be spontaneous. Presently available therapies for TN include both surgical and pharmacological management; however, the lack of a known etiology for TN contributes to the unpredictable response to treatment and the variability in long-term clinical outcomes. Given this, a range of peripheral and central mechanisms underlying TN pain remain to be understood. We acquired functional magnetic resonance imaging (fMRI) data from TN patients who (1) rested comfortably in the scanner during a resting state session and (2) rated their pain levels in real time using a calibrated tracking ball-controlled scale in a pain tracking session. Following data acquisition, the data was analyzed using the conventional correlation analysis and two artificial intelligence (AI)-inspired deep learning methods: convolutional neural network (CNN) and graph convolutional neural network (GCNN). Each of the three methods yielded a set of brain regions related to the generation and perception of pain in TN. There were six regions that were identified by all three methods, including the superior temporal cortex, the insula, the fusiform, the precentral gyrus, the superior frontal gyrus, and the supramarginal gyrus. Additionally, 17 regions, including dorsal anterior cingulate cortex(dACC) and the thalamus, were identified by at least two of the three methods. Collectively, these 23 regions represent signature centers of TN pain and provide target areas for future studies relating to central mechanisms of TN.

Causal Interactions in Brain Networks Predict Pain Levels in Trigeminal Neuralgia

“…This study is a reanalysis of previously published data (Liang et al, 2022). The questions addressed here are, however, entirely different from the previously published work.…”

“…The tracking ball controlled the movement of a cursor along a straight line with 0 and 100 indicated at the two ends of the line on the monitor. It is worth noting that for 7 of the 39 patients, the pain tracking session was 10 minutes in duration (Liang et al, 2022).…”

“…In our previous study, applying correlation analysis, convolutional neural network (CNN) analysis, and graph convolutional neural network (GCNN) analysis (Liang et al, 2022), and requiring consistent results across multiple methods, we have identified 23 brain regions that are related to the generation and perception of TN pain; see Table 1. In the third analysis, we examined the causal interactions in this TN pain network and how these causal interactions related to TN pain experienced by the patient.…”

“…Causal interactions among these ROIs were evaluated and correlated with pain levels in both sessions. In the third analysis, we further expanded the list of ROIs to include 23 regions that have been identified in a recent study as being specific to TN pain (Liang et al, 2022). In all three analyses consistent findings between the resting and the pain tracking sessions were sought to enhance the statistical rigor of the analysis.…”

Decoding fMRI Data: A Comparison Between Support Vector Machines and Deep Neural Networks

“…Both decoding accuracy and heatmaps can be derived from the trained CNN models. Recently, we applied a CNN model to fMRI data recorded from patients suffering from trigeminal neuralgia to reveal novel insights into the generation and perception of TN pain, which were not possible with other methods (Liang et al, 2022). Given the prevalence of SVM and the emerging significance of CNN in MVPA analysis of neuroimaging data, we consider it timely to compare CNNs and SVMs by applying them to the same datasets.…”