Predicting TMS Tinnitus Relief via Brain Biomarkers

A new study has identified a specific brain structure that can predict which patients with subjective tinnitus are most likely to benefit from repetitive transcranial magnetic stimulation (rTMS). The research, led by Zhongling Ding, Bo Peng, and Mengfang Gong, found that a larger volume of gray matter in a region called the right pars triangularis of the inferior frontal gyrus was the top predictor of a positive treatment outcome. This structural feature distinguished responders from non-responders with 85% accuracy in their predictive model.

Pinpointing a Predictive Brain Feature

The research team enrolled 64 patients with subjective tinnitus and 18 healthy controls. All patients underwent a standard two-week course of rTMS treatment. Before treatment began, each participant received a high-resolution structural MRI (sMRI) brain scan. From these scans, the researchers extracted 242 distinct measurements of brain morphometry, covering the volume, thickness, and surface area of cortical and subcortical regions across the entire brain.

The goal was to see if the baseline structure of a patient’s brain held clues about their future response to therapy. Patients were classified as responders if they showed meaningful clinical improvement after the rTMS course. In this cohort, 36 patients (56.25%) met the criteria for being a responder.

How a Machine Learning Model Identified the Key Predictor

Using univariate analysis, the scientists first identified 10 regional brain features that statistically differed between the responder and non-responder groups. These features were located within networks involved in executive function (prefrontal), emotion (limbic), sensory processing (sensorimotor), and integration (parietal).

They then fed these 10 features into a machine learning algorithm to build a predictive model. The model’s performance was rigorously tested using 5-fold cross-validation, a method that prevents overfitting and gives a reliable estimate of real-world accuracy. The best-performing model was an ExtraTreesGini_BAG_L1 classifier. It achieved an area under the curve (AUC) of 0.85, indicating very good predictive ability. The model correctly classified 77% of patients overall (accuracy), with a particularly strong ability to correctly identify those who would respond (97% recall).

To understand which brain feature contributed most to the model’s decisions, the researchers used SHapley Additive exPlanations (SHAP) analysis. This technique clearly identified the gray matter volume of the right pars triangularis of the inferior frontal gyrus (IFGtriang-R) as the top predictor. A larger volume in this region had a positive influence on the prediction of treatment success.

A Structural Signature Unique to Responders

The team took a further step to interpret this finding by comparing the IFGtriang-R volume across three groups: responders, non-responders, and healthy controls. The results were telling. Responders had a significantly larger volume in this region (0.90 ± 0.08) compared to both healthy controls (0.86 ± 0.06) and non-responders (0.86 ± 0.07). There was no significant difference between healthy controls and non-responders.

This pattern suggests that having a larger IFGtriang-R is not simply a marker of having tinnitus, but a specific signature associated with a brain that is primed to respond positively to rTMS. The authors propose this may represent a threshold of “neuroplastic reserve” needed for the treatment to induce beneficial changes in the brain networks underlying tinnitus. For more on the role of brain structure in treatment prediction, see our article on Predicting Tinnitus Treatment Response with Brain Biomarkers.

Interestingly, correlation analyses found that the size of the IFGtriang-R did not directly correlate with the degree of symptom improvement (ΔVAS or ΔTHI scores), nor did any structural feature strongly correlate with baseline tinnitus severity. This supports the idea of a threshold effect rather than a linear relationship.

Practical Implications for Personalizing Tinnitus Treatment

The variable efficacy of rTMS for tinnitus has been a significant clinical challenge. This study moves the field toward a more precision-based approach. The findings indicate that a routine pre-treatment sMRI scan could be analyzed to assess the volume of the right IFG pars triangularis, providing clinicians with a data-driven biomarker to aid in patient selection.

Identifying patients with a higher likelihood of response could make rTMS a more efficient and cost-effective treatment option. It could prevent individuals with a low probability of benefit from undergoing an intensive, and sometimes costly, treatment course that may not work for them. Instead, they could be directed toward other evidence-based interventions sooner. This aligns with a broader movement in hearing health toward Integrated Auditory Health Advances that consider the whole patient.

For patients predicted to be non-responders to rTMS, other neuromodulation or multidisciplinary approaches remain important. For instance, research on Manual Neck Jaw Exercises Reduce Tinnitus Severity highlights a different, non-invasive pathway for managing symptoms that may be effective for a different subset of individuals.

Conclusion: Toward Precision Neuromodulation

The work by Ding, Peng, and Gong provides a concrete, biologically grounded biomarker that explains some of the variability in rTMS outcomes for tinnitus. The enlargement of the right inferior frontal gyrus in responders points to the critical role of fronto-temporal cognitive and emotional control networks in successful treatment. By validating this biomarker in future, larger studies, clinicians may soon be able to use simple brain scans to stratify patients, making neuromodulation for tinnitus more targeted and effective.

Source: Ding Z, Peng B, Gong M. Pre-treatment brain structural biomarkers for predicting repetitive transcranial magnetic stimulation efficacy in subjective tinnitus. Front Neurol. 2026;17:1808769.