A year ago, we reported on a study the Foundation had funded called “Mapping the Brain.” This was a study aimed at identifying all the brain centers that are activated when a patient is experiencing TN pain so that we could track the pathway of the pain signal. This study has now been enhanced by using some new analysis tools: two different artificial intelligence (AI) inspired deep-learning methods were used to analyze the data. Overall, the study identified more new areas of TN pain and more consistency in the data around the targeted areas. We hope that researchers can use these findings, including the new AI methods, along with the findings of our genetics study, to create new clinical interventions to stop the pain. 

Update Details:

The first analysis performed on the data was a normal human correlation study aimed at identifying centers that correlate with pain level fluctuations. Other researchers have run the same sort of study in the past on other pain conditions, but the test and correlations are difficult to read due to a number of factors. Table 1 shows the results of the five most recent  studies to measure TN activity using resting state data and stimulated pain rather than naturally occurring TN pain.

Table #1 Regions Showing Abnormal Activities in TN per Published Literature

Brain Areas

Yuan et al.

Wang et al.
Zhu et al.
Yan et al.

Moisset et al.


Cingulate cortex


Middle temporal gyrus


Medial frontal gyrus

Superior frontal gyrus


Parietal lobule

Postcentral gyrus


Para hippocampal

Inferior temporal gyrus

Lingual gyrus

Paracentral lobule

Fusiform gyrus
Middle occipital gyrus
Precentral gyrus

Secondary somatosensory cortex

Supplementary motor area

Superior temporal gyrus


As you can see, except for the cerebellum, there is little consistency across the studies, rendering the data of limited use.

The team led by Mingzhou Ding has just published the first of three papers on their research study into this issue, entitled “Imaging the Neural Substrate of Trigeminal Neuralgia Pain Using Deep Learning.” The team measured real TN pain that occurred while the patients were being scanned in the fMRI, and then applied conventional correlation analysis and two different artificial intelligence (AI) inspired deep-learning methods to the data. The two deep-learning approaches used were convolution neural networks (CNN) and graph convolution neural networks (GCNN).

You can see the results of these analyses on the Mingzhou Ding study in Table #2. Overall, these approaches found more new areas identified with TN pain and there is more consistency in the data. Six regions were identified in all three methods including the superior temporal, insula, fusiform, precentral gyrus, superior frontal gyrus, and the supramarginal gyrus. This is important because, as studies build convergence around the pain pathway, the  the data becomes more valuable in identifying sites for clinical interventions to stop the pain. The data is also important as we begin to integrate the effect of genetic mutations on TN. Scientists can compare if certain mutations are related to significant signal areas in the pain processing pathway and better understand the mechanisms involved.

Table #2 Signature Centers of TN Pain Identified by Correlation and AI-inspired Analysis




Superior frontal


Superior temporal




Lateral orbitofrontal


Rostral middle frontal

Lateral occipital


Inferior temporal

Pars opercularis




Rostral anterior cingulate




Inferior parietal

Dorsal ACC


These results utilizing a functional paradigm in which TN patients tracked their real pain level in real-time shed new light on the pain pathway issue. In particular, it was found that the neural activities in these six areas not only closely tracked the pain level fluctuations, but they also mediated network-level communications among different brain regions. Combining these AI-inspired methods with conventional methods to seek converging evidence may become a promising method in future neuroimaging studies.