Trigeminal neuralgia (TN) is a severe neurological disorder characterized by paroxysmal pain (1, 2). The primary surgical treatment modalities for TN include: microvascular decompression(MVD), percutaneous radiofrequency thermocoagulation (RFT), and gamma knife radiosurgery (GKS) (1, 3, 4). The coexistence of multiple surgical approaches for TN stems from the distinct advantages and disadvantages associated with each method. RFT offers a higher safety profile, with no reported mortality or severe disability, making it suitable for patients without confirmed vascular compression on CNV, elderly patients, or those with comorbidities precluding craniotomy. Nevertheless, the relatively high postoperative recurrence rate significantly impacts the efficacy of RFT (5–9).

There are some reports about magnetic resonance imaging (MRI) and postoperative recurrence rate of RFT. 5%-10% of patients with severe vascular compression (Grade III) fund by MRI experience recurrence within 6 months postoperatively (1, 10–12). Forevermore, our previous work revealed that when classifying TN patients using MR Diffusion Tensor Imaging (MR-DTI) findings, the probability of recurrence within 6 months after RFT was 10% for the L-FA type but as high as 60% for the N-FA type (13, 14). While these studies have identified some factors contributing to short-term recurrence after RFT, the complete set of risk factors predicting short-term recurrence for individual TN patients remains incompletely defined. Predicting and preventing short-term recurrence is therefore a critical concern.

Functional magnetic resonance imaging (fMRI) is another MR technique and can analyze changes in brain activity induced by pain stimuli (15).Abnormal functional brain regions in TN are critical findings on fMRI. Key questions warranting attention are whether the abnormal activation in specific brain regions can serve as novel predictive indicators for short-term recurrence following RFT (16–18).

In this study, while applying traditional fMRI methodology, we also first established a novel rs-fMRI approach: individualized comparison of each TN patient against a normal control group. This new individualized method – individualized rs-fMRI (irs-fMRI) – holds promise for providing clinically meaningful predictive indicators of surgical treatment outcomes for individual TN patients.

Using conventional rs-fMRI methods, this study identified four novel abnormal brain regions. Furthermore, applying the irs-fMRI method to 15 consecutive patients undergoing percutaneous stereotactic radiofrequency ablation(PSR)with DTI guiding surgery, we discovered that abnormal activation in six brain regions, including Insula_L, Fusiform_L, Vermis_3, Cerebellum_3_R, Temporal_Sup_L, and Fusiform_R, can serve as predictive indicators for short-term recurrence after RFT in TN patients.

Beyond group-level effects, TN likely reflects meaningful inter-individual heterogeneity in pain-related brain circuits. Precision neuropsychiatry emphasizes linking biological heterogeneity to individualized biomarkers that can support patient stratification and targeted clinical decision-making. In chronic pain conditions such as TN, brain networks involved in sensory processing, salience, affect, and memory may jointly shape symptom severity and treatment response. Therefore, an individualized rs-fMRI strategy that evaluates each patient against a normative reference could provide a practical imaging biomarker for identifying patients at higher risk of early relapse and for informing precision-oriented follow-up and targeted intervention planning.

Recurrence in TN patients after RFT can be categorized into two types based on the timing of onset: short-term recurrence (STR) which occurring within 1 year postoperatively, and long-term recurrence (LTR) which occurring more than 1 year postoperatively. Reported rates of STR and LTR vary considerably in the literature. STR rates fall into three ranges: low range (<9.8%) (22–26), medium range (12%—17%) (27–30), and high range (21%—56%) (31–34). Two-year LTR rates are distributed into two ranges: low range (7%—17%) (25, 26, 35) and high range (20%—50%) (5, 28, 29, 33, 36, 37).

The currently prevailing theoretical explanation for LTR is the regeneration of nerve fibers within the TGT occurring 1-3 years postoperatively, leading to the reappearance of pain (5), termed neuroregeneration-mediated recurrence. Regarding STR, we propose that it represents non-neuroregeneration-mediated recurrence under the premise of effective RFT, i.e., recurrence due to patient-specific factors.

Indeed, not only STR research had been received considerable attention (31–34), but STR occurring even earlier (within 1-6 months) has also been noted (23, 24, 31, 38). J. Piquer et al. (39) reported a 1-year recurrence rate of 13% (13/98), but patients recurring within one year accounted for 46% (13/30) of all recurrences observed over more than three years. In our cohort, STR patients constituted 80% (4/5) of all recurrences within 3 years postoperatively. We speculate that the unusually high STR rate in our group may be related to a relatively higher proportion of patients with specific constitutional factors.

To date, 23 brain regions have been identified by various researchers as abnormally functional in TN patients (2, 40–42). We identified 19 abnormal functional brain regions in TN patients. The majority of these abnormal regions (68.4%, 13/19) were fully consistent with literature reports, a smaller portion (21.0%, 4/19) partially consistent, while abnormalities in the Rolandic_Oper_L and Cerebellum_9_L regions have not been previously reported in the literature.

Furthermore, in 15 PSR group patients with VAS scores of 10, we identified 16 abnormal brain regions. Among these, abnormalities in 9 regions related to pain perception, pain processing, emotional processing, vision, and memory differed from those observed in the broader TN group. This finding suggests that TN patients with severe pain possess unique characteristics in their abnormal brain region distribution. Among these 16 abnormal regions, abnormalities in Cerebellum_9_L, Lingual_R, and Calcarine_L have not been previously reported.

Overall, our study identified four novel abnormal brain regions in TN patients: Rolandic_Oper_L, Cerebellum_9_L, Lingual_R, and Calcarine_L. We attribute these findings to differences in the clinical characteristics of enrolled TN patients and variations in fMRI methodologies across studies. In fact, Liang Y et al. (43) also noted a lack of consistency in the locations of abnormal brain regions reported by different centers conducting fMRI research on TN patients.

Traditionally associated with motor control, the cerebellum is increasingly recognized for its critical role in non-motor functions, including pain modulation and emotional processing. The “cognitive-affective cerebellum” theory suggests that the posterior lobe and vermis serve as integral nodes within the pain matrix and salience network, specifically influencing the affective-motivational dimension of pain. Persistent abnormalities in these regions (e.g., Vermis_3 and Cerebellum_3_R) observed in our high-risk cohort may reflect entrenched central sensitization or maladaptive emotional learning associated with chronic neuropathic pain. Consequently, these central circuit disturbances could drive early symptom recurrence, independent of the technical success of the peripheral nerve intervention.

Departing from previous group-level comparison methods, this study also performed individual case-control comparisons for each of the 15 PSR group patients against a normal control group. Based on this cohort, all patients exhibiting abnormalities in one or more of the following regions (contralateral Insula_L, Fusiform_L, Vermis_3, Temporal_Sup_L, and ipsilateral Cerebellum_3_R, Fusiform_R with V1 involvement) experienced short-term recurrence after PSR-DTI. We therefore refer to these regions as candidate high-risk fMRI regions for recurrence risk stratification, pending external validation. Consequently, we define these 6 brain regions as high-risk recurrence regions on fMRI (H-fMRI) for individual TN patients undergoing PSR. Abnormalities in the Vermis_3 and Cerebellum_3_R regions in TN patients have not been previously reported.

The distinct predictive value of V1 division involvement combined with Fusiform_R abnormalities may be attributed to both anatomical and functional factors. Clinically, radiofrequency ablation of the ophthalmic (V1) division often requires more conservative temperature parameters to preserve the corneal reflex, which can potentially lead to less complete denervation compared to procedures targeting the V2 or V3 divisions. Functionally, the fusiform gyrus is integral to higher-order visual and object processing. The specific coupling of V1 pain—which involves the ophthalmic nerve—with fusiform abnormalities suggests a specific maladaptive neuroplasticity within visual-pain associative pathways. This unique central network alteration may be more resistant to standard thermal coagulation, thereby predisposing patients to higher rates of early recurrence.

In this study, PSR surgery resulted in the elimination or reduction of recurrence-related differential brain regions in each TN patient. This outcome aligns with Dou Z et al. (44) reporting decreased ReHo values in the Middle Temporal Gyrus, Postcentral Gyrus, and left Insula after RFT; matches Moisset X et al. (2)reporting the disappearance of Insula activation 1-2 months post-RFT; and is largely consistent with Wen-Ching Liu et al. (45) reporting significant regression of Insula and Cerebellar activation 1-2 weeks post-RRT. These findings demonstrate that PSR-DTI surgery effectively modulates abnormal brain region activity in TN patients.

Research on individualized rs-fMRI (irs-fMRI) is clinically essential for using abnormal brain regions as an indicator to predict postoperative recurrence risk. Theoretically, irs-fMRI results could be influenced by differences in patient state during pre- and postoperative scans. In this study, irs-fMRI examinations focusing on abnormal regions within the potential recurrence library confirmed that all 15 abnormal regions were either eliminated or attenuated following effective immediate treatment. This result suggests that targeting specific brain regions is an effective approach for individualized fMRI assessment.

There are reports that TN patients whom with increased ReHo values in Insula and/or Vermis experienced moderate short-term recurrence post-RFT (2, 44). Those reports had some extent corroborate our finding that Insula and Vermis might contribute to short-term recurrence after RFT. However, the other four of our fund 6 H-fMRI regions as causes of short-term recurrence not previously reported. For the highest recurrence rates (100%) of fMRI compare with MRI (5%-15%) and DTI (60%). Taken together, these observations suggest that abnormal functional brain regions may represent an important correlate of short-term recurrence risk after RFT/PSR-DTI in TN. However, the predictive utility and generalizability of these candidate regions require prospective evaluation in larger, independent cohorts.

This study has certain limitations. We applied lateralization and branch-specific restrictions to the H-fMRI regions in this timeframe. As research sample sizes increase in the future, these lateralization and branch restrictions for H-fMRI regions may require further refinement.

Rolandic_Oper_L, Cerebellum_9_L, Lingual_R, and Calcarine_L were newly discovered abnormal brain regions in trigeminal neuralgia patients. For the first time, six high-risk recurrence regions on fMRI highly associated with short-term postoperative recurrence were identified: contralateral Insula_L, Fusiform_L, Vermis_3 and Temporal_Sup_L, ipsilateral Cerebellum_3_R, Fusiform_R (specifically associated with V1 division involvement). This study provides an objective imaging basis for individualized prediction of early recurrence risk after PSR-DTI in trigeminal neuralgia.