Preoperative imaging evaluation of primary trigeminal neuralgia using 3D TOF-MRA and 3D FIESTA-c: A retrospective study of 412 cases Authors: Yu Huang, Ying Huang, Chaoyong Xiao, Qingling Huang, Xue Chai Published: 2026 ABSTRACT AbstractObjectiveTo evaluate the clinical value of three-dimensional time-of-flight magnetic resonance angiography (3D TOF-MRA), three-dimensional fast imaging employing steady-state acquisition with phase cycling (3D FIESTA-c), and their combination for preoperative assessment of responsible vessels in primary trigeminal neuralgia (PTN) before microvascular decompression (MVD).MethodsWe retrospectively reviewed 412 patients with PTN confirmed by MVD who underwent preoperative 3D TOF-MRA and 3D FIESTA-c. Intraoperative findings served as the reference standard. Diagnostic performance was assessed using 2 criteria: binary identification of the presence or absence of a responsible vessel and complete concordance of vessel type. Agreement with surgical findings was evaluated using Cohen’s kappa. Subgroup analyses were performed according to responsible vessel type.ResultsResponsible vessels were identified intraoperatively in 407 of 412 patients (98.79%), whereas no responsible vessel was found in 5 patients (1.21%). The superior cerebellar artery (SCA) was the most common responsible vessel (250/412, 60.68%), followed by involvement of the SCA and petrosal vein (54/412, 13.11%). Under binary criteria, the accuracy of 3D TOF-MRA, 3D FIESTA-c, and the combined approach was 90.53%, 91.99%, and 96.36%, respectively, with a significant overall difference (Cochran’s Q = 23.400, P < 0.001). Under complete concordance criteria, the corresponding matching rates were 74.27%, 82.77%, and 87.86%, respectively, with kappa values of 0.531, 0.702, and 0.786 (Cochran’s Q = 64.892, P < 0.001). Pairwise comparisons showed that the combined approach outperformed either sequence alone, and 3D FIESTA-c also outperformed 3D TOF-MRA under complete concordance. The combined approach also performed better in the SCA and anterior inferior cerebellar artery (AICA) subgroups.ConclusionThe combination of 3D TOF-MRA and 3D FIESTA-c improves preoperative assessment of responsible vessels in PTN, with higher diagnostic accuracy and better agreement with intraoperative findings than either modality alone.Keywords: Primary trigeminal neuralgia, 3D time-of-flight magnetic resonance angiography, 3D FIESTA-c, Microvascular decompression, Magnetic resonance imaging Introduction PTN is a common cranial nerve disorder in neurosurgery and pain medicine. It is characterized by recurrent, brief episodes of severe, electric shock-like or stabbing pain in the trigeminal distribution and can markedly impair quality of life [1, 2]. The main pathophysiological mechanism is generally considered to be chronic pulsatile compression of the trigeminal nerve at the root entry zone (REZ) by adjacent vessels, resulting in focal demyelination and abnormal neural excitability [3–5]. MVD, which relieves the vascular compression of the trigeminal nerve, is widely regarded as one of the most effective surgical treatments for PTN [2–4]. Accurate preoperative identification of the responsible vessel, including its presence, type, and relationship to the trigeminal nerve, is therefore important for surgical planning and intraoperative orientation. With advances in magnetic resonance imaging (MRI), high-resolution neurovascular imaging has become an important component of preoperative evaluation in PTN [2, 6–17]. Among the available techniques, 3D TOF-MRA is useful for depicting responsible arteries [6–8, 10, 16], whereas 3D FIESTA-c provides better visualization of the spatial relationship between the trigeminal nerve and adjacent vessels [9, 12, 14]. However, each sequence has limitations, particularly in the presence of low-flow vessels, venous compression, multiple-vessel involvement, or crowded local anatomy. Combined imaging may therefore improve diagnostic accuracy in preoperative responsible-vessel assessment. Previous studies evaluating 3D TOF-MRA and 3D FIESTA-c in trigeminal neuralgia have mostly been based on small to moderate sample sizes and have primarily focused on the binary presence or absence of neurovascular contact or compression, with less attention given to whether the responsible vessel type identified on imaging is fully concordant with intraoperative findings [6, 8, 13, 16, 18]. In this retrospective study, we analyzed 412 patients with PTN confirmed by MVD and systematically assessed the value of 3D TOF-MRA, 3D FIESTA-c, and their combination from 2 perspectives: binary classification and complete concordance. Our aim was to provide further evidence for preoperative imaging evaluation and surgical decision-making in PTN. Materials and methods Patient data This retrospective study included 412 consecutive patients with PTN who underwent MVD in the Department of Neurosurgery at our institution between January 2017 and October 2025. There were 164 men (39.81%) and 248 women (60.19%), with a male-to-female ratio of 1:1.51. Age ranged from 22 to 88 years, with a mean age of 62.75 ± 9.75 years. Symptom duration ranged from 1 month to 30 years, with a mean duration of 4.67 ± 4.81 years. Pain involved the right side in 238 patients (57.77%) and the left side in 174 (42.23%). Inclusion criteria were as follows: (1) diagnosis of PTN according to the International Classification of Headache Disorders, 3rd edition (ICHD-3) [19]; (2) complete clinical and imaging data; and (3) treatment with MVD with complete intraoperative documentation of the responsible vessel. Exclusion criteria were: (1) secondary trigeminal neuralgia due to tumour, inflammation, multiple sclerosis, or other identifiable causes; (2) incomplete imaging data or inadequate image quality; and (3) a history of posterior fossa surgery resulting in substantial anatomical distortion. Given the retrospective nature of the study and the absence of direct patient involvement, the requirement for informed consent was waived. The study protocol was reviewed and approved by the institutional ethics review committee. MRI protocol All MRI examinations were performed on a 3.0-T scanner (GE Discovery 750; GE Healthcare, Milwaukee, WI, USA) using a standard head coil, with patients in the supine position. For 3D FIESTA-c, imaging parameters were as follows: repetition time (TR), 6.4 ms; echo time (TE), 3.1 ms; field of view (FOV), 240 × 226 mm; matrix, 512 × 512; number of excitations (NEX), 2; and slice thickness, 1 mm. For 3D TOF-MRA, parameters were: TR, 12 ms; TE, 2.9 ms; FOV, 180 × 208 mm; matrix, 512 × 512; flip angle, 18°; NEX, 2; and slice thickness, 1 mm. Raw image data were transferred to the picture archiving and communication system (PACS) workstation for post-processing. Multiplanar reformation (MPR) was applied to 3D FIESTA-c images, and maximum intensity projection (MIP) to 3D TOF-MRA images. The anatomical relationship between the trigeminal nerve and adjacent vessels was assessed on axial, coronal, and sagittal views. Image analysis All images were independently reviewed by two experienced neuroradiologists who were blinded to the intraoperative MVD findings. Any disagreement was resolved by consensus. According to the criteria described by Arbab et al. [16], the neurovascular relationship was classified into three categories: type I, no contact, defined as absence of a visible vessel adjacent to the nerve or the presence of a clear gap between the vessel and nerve; type II, contact, defined as loss of the intervening gap without visible nerve distortion at the contact point; and type III, compression, defined as visible deformation or displacement of the nerve at the site of neurovascular contact. Type I was considered negative for a responsible vessel, whereas types II and III were considered positive.Query Using intraoperative MVD findings as the reference standard, imaging performance was evaluated according to two criteria. First, under the binary criterion, only the presence or absence of a responsible vessel was assessed; an imaging result was considered correct when it agreed with the operative finding at this level. Second, under the complete concordance criterion, the responsible vessel identified on imaging had to match the intraoperative finding exactly in terms of vessel type. Cases in which imaging indicated the presence of a responsible vessel but failed to identify the correct vessel type were classified as discordant. The binary criterion was intended to reflect the ability of each modality to detect responsible vessels, whereas the complete concordance criterion was used to assess the accuracy of vessel-type identification. Statistical analysis Statistical analyses were performed using SPSS version 26.0. Continuous variables are presented as mean ± standard deviation, and categorical variables as counts and percentages. Overall comparisons among the three imaging modalities in the same patients were performed using Cochran’s Q test. When the overall comparison was significant, post hoc pairwise comparisons were conducted using the continuity-corrected McNemar test with Bonferroni adjustment; the adjusted significance threshold was set at α = 0.017. Agreement between each imaging modality and intraoperative findings was assessed using Cohen’s kappa coefficient, with corresponding 95% confidence intervals. Interpretation of kappa values followed the criteria proposed by McHugh [20]. Analyses of responsible-vessel subgroups with small sample sizes were considered exploratory. A two-sided P value < 0.05 was considered statistically significant. Results Distribution of responsible vessel types at MVD Among the 412 patients, intraoperative findings confirmed the presence of responsible vessels in 407 cases (98.79%), whereas no definite responsible vessel was identified in 5 cases (1.21%). Arterial compression predominated, with the SCA being the most common responsible vessel, followed by combined SCA and PV involvement. Other arterial, venous, and mixed vascular compression patterns were relatively uncommon. The complete distribution of responsible vessel types is shown in Table 1. Table 1Distribution of responsible vessel types at MVDResponsible Vessel TypeCasesPercentage %SCA25060.68SCA + PV5413.11PV245.83AICA245.83SCA + AICA235.58VA/BA153.64SCA + VA/BA92.18AICA + VA/BA51.21AICA + PV30.73No responsible vessel51.21Total412100.00 Distribution of responsible vessel types at MVD Comparison of diagnostic accuracy for determining the presence or absence of a responsible vessel among the three methods Under the binary criterion, the diagnostic accuracy of the three imaging methods in determining the presence or absence of a responsible vessel differed significantly. The combined method showed the best diagnostic performance and was significantly superior to either sequence alone, whereas no significant difference was observed between 3D FIESTA-c and 3D TOF-MRA. Detailed accuracy values and pairwise statistical comparisons are provided in Table 2. Table 2Comparison of accuracy for detecting the presence or absence of a responsible vessel and complete matching rate of vessel type among the three methodsMethodAccuracyMatching rateTest statistic (accuracy/matching rate)P value(accuracy/matching rate)3D TOF-MRA373 (90.53)306 (74.27)--3D FIESTA-c379 (91.99)341 (82.77)--Combined397 (96.36)362 (87.86)--Cochran’s Q--Q = 23.400/64.892 < 0.001/< 0.001Combined vs 3D FIESTA-c--χ2 = 16.056/19.048 < 0.001/< 0.001Combined vs 3D TOF-MRA--χ2 = 22.042/54.018 < 0.001/< 0.0013D FIESTA-c vs 3D TOF-MRA--χ2 = 0.658/16.2820.417/< 0.001 Comparison of accuracy for detecting the presence or absence of a responsible vessel and complete matching rate of vessel type among the three methods Complete matching rates and agreement analysis for responsible vessel identification among the three methods Under the complete concordance criterion, the combined method achieved the highest complete matching rate for responsible vessel type, followed by 3D FIESTA-c and 3D TOF-MRA, and 3D FIESTA-c also showed a significant advantage over 3D TOF-MRA. Both the overall and pairwise comparisons supported this trend. Kappa agreement analysis showed that the combined method had the strongest agreement with intraoperative findings, followed by 3D FIESTA-c, whereas 3D TOF-MRA showed relatively lower agreement. Detailed results are shown in Table 2 and Fig. 1. Agreement of the three methods with intraoperative findings (Kappa values and 95% Confidence intervals) Comparison of complete matching rates across responsible vessel type subgroups Subgroup analysis showed that the complete matching rates of the three methods varied according to responsible vessel type. The combined method showed an advantage in the SCA subgroup and was superior to 3D FIESTA-c in the AICA subgroup. In venous-involved subgroups, particularly the SCA + PV and PV subgroups, both 3D FIESTA-c and the combined method outperformed 3D TOF-MRA. No significant differences were observed in the SCA + AICA or AICA + VA/BA subgroups. Representative cases are shown in Figs. 2,3,4, and the comparison results are shown in Table 3 and Fig. 5. A 54-year-old woman with a 2-year history of left-sided trigeminal neuralgia. Axial (A) and sagittal (B) 3D TOF-MRA images show the superior cerebellar artery as a high-signal structure (red arrows) compressing the left trigeminal nerve (yellow arrows) from above. Axial (C) and sagittal (D) 3D FIESTA-c images show the superior cerebellar artery as a low-signal structure (red arrows) compressing the left trigeminal nerve (yellow arrows) from above A 57-year-old woman with a history of left-sided trigeminal neuralgia for more than 5 years. Axial (A) and sagittal (B) 3D TOF-MRA images show only the superior cerebellar artery (red arrows) compressing the left trigeminal nerve (yellow arrows) from above, with no other vessels identified. Axial (C) and sagittal (D) 3D FIESTA-c images show the superior cerebellar artery (red arrows) and the petrosal vein (blue arrows) jointly compressing the left trigeminal nerve (yellow arrows) from above and below, respectively A 68-year-old woman with a 1-year history of left-sided trigeminal neuralgia. Axial (A) and sagittal (B) 3D TOF-MRA images show the petrosal vein (blue arrows) as an isointense structure compressing the left trigeminal nerve (yellow arrows) from below. Axial (C) and sagittal (D) 3D FIESTA-c images show the petrosal vein (blue arrows) as a low-signal structure compressing the left trigeminal nerve (yellow arrows) from below Table 3Pairwise comparison of complete matching rates by responsible vessel type subgroupResponsible Vessel Type3D TOF-MRA vs 3D FIESTA-c χ2/PCombined vs 3D TOF-MRA χ2/PCombined vs 3D FIESTA-c χ2/PSCA-8.100/0.0048.100/0.004AICA1.778/0.1821.333/0.2486.125/0.013SCA + PV23.040/< 0.00124.038/< 0.0010.000/1.000PV9.091/0.0039.091/0.003-SCA + AICA0.167/0.6832.250/0.1340.500/0.480AICA + VA/BA0.500/0.4800.500/0.480-AICA + PV---SCA + VA/BA---VA/BA---“—” indicates that the paired results of the two methods were completely identical in that subgroup; therefore, no McNemar test was performed Pairwise comparison of complete matching rates by responsible vessel type subgroup “—” indicates that the paired results of the two methods were completely identical in that subgroup; therefore, no McNemar test was performed Forest plot of complete matching rates by responsible vessel type Discussion This study systematically evaluated the value of single-sequence 3D TOF-MRA, single-sequence 3D FIESTA-c, and their combination for preoperative responsible-vessel assessment in 412 patients with PTN confirmed by MVD. Unlike previous studies that used only the detection of a responsible vessel as the primary endpoint, the present study applied both a binary criterion and a complete concordance criterion, thereby assessing imaging performance from 2 clinically relevant perspectives: identification of the presence or absence of a responsible vessel and precise determination of responsible-vessel type. This design more closely reflects the practical demands of surgical planning. The trigeminal REZ is anatomically complex, with numerous small surrounding vessels and substantial vascular variation, making it the site most vulnerable to neurovascular compression. By tracing the vascular origin of compression, we found that arterial compression was the predominant mechanism in PTN, in keeping with previous reports[17, 21, 22]. Earlier studies have shown that the superior cerebellar artery and anterior inferior cerebellar artery are the most common responsible vessels, accounting for approximately 46.6%–88% and 5%–39% of cases, respectively[7, 14, 23, 24]. Venous compression is also a relevant cause, with a reported incidence of approximately 5.5%–18% [7, 14, 22, 25]. In our cohort, intraoperative MVD confirmed the presence of a responsible vessel in 98.79% of patients. The SCA accounted for 60.68% of cases, the AICA for 5.83%, and the VA/BA for 3.64%, while combined arterial compression accounted for 9.22%. The petrosal vein, either alone or in combination with other vessels, was involved in 19.67% of cases (5.83% + 13.11% + 0.73%). Overall, these findings are broadly consistent with the published literature. Under the binary criterion, the accuracy rates of 3D TOF-MRA, 3D FIESTA-c, and the combined approach for determining the presence or absence of a responsible vessel were 90.53%, 91.99%, and 96.36%, respectively. The combined approach performed significantly better than either single sequence alone, whereas no significant difference was observed between the 2 single-sequence techniques. Single-sequence imaging remained susceptible to missed or incorrect identification, particularly in cases with atypical contact angles, crowded local anatomy, small-caliber vessels, or venous involvement[8, 11, 26]. Because 3D TOF-MRA is based on flow-related enhancement, it selectively visualizes rapidly flowing blood and typically depicts arteries as high-signal structures, thereby facilitating recognition of the neurovascular relationship. However, its ability to depict slowly flowing veins and small-caliber arteries is limited[8, 10, 12, 16]. By contrast, 3D FIESTA-c enhances the contrast between cerebrospinal fluid and surrounding tissues, allowing low-signal vessels and isointense nerves to be visualized more clearly against the high-signal cerebrospinal fluid background. Nevertheless, when the responsible vessel lies close to the pons surface or closely overlaps with the trigeminal nerve in the absence of an intervening cerebrospinal fluid space, assessment with 3D FIESTA-c becomes difficult [7, 9, 12]. These two techniques are therefore complementary, and their combined use can substantially improve the accuracy of responsible-vessel detection while reducing both missed and incorrect diagnoses. Compared with simply determining whether vascular compression is present, the criterion of complete concordance with respect to responsible vessel is more stringent and more clinically relevant, as it better reflects the practical value of preoperative imaging for surgical decision-making. In the present study, the combined approach achieved the highest complete matching rate for responsible vessel type (87.86%), significantly outperforming 3D FIESTA-c (82.77%) and 3D TOF-MRA (74.27%). Kappa analysis further showed that the combined approach yielded the best agreement with intraoperative findings (0.786), followed by 3D FIESTA-c (0.702) and 3D TOF-MRA (0.531). These findings indicate that the combined approach provides greater accuracy in identifying responsible vessel type and more faithfully reflects operative findings. For preoperative evaluation before MVD, the combined approach is therefore valuable not only for determining the presence of vascular compression, but also for clarifying the type of responsible vessel, thereby offering more reliable imaging guidance for surgical planning and intraoperative decompression strategies. Subgroup analysis further demonstrated that the diagnostic performance of the three imaging methods was strongly dependent on responsible vessel type. This likely reflects differences in vessel caliber, flow characteristics, anatomic variation, and patterns of contact with the trigeminal nerve. In the SCA subgroup, all 3 methods achieved relatively high matching rates; however, the combined approach yielded the highest rate, with a statistically significant advantage over either single sequence. This finding suggests clear complementarity between 3D TOF-MRA and 3D FIESTA-c in terms of flow visualization and spatial localization, indicating that their combined use can further improve the accuracy of identifying SCA as the responsible vessel. In the AICA subgroup, although the combined approach showed the highest matching rate, the difference compared with 3D TOF-MRA was not statistically significant. One possible explanation is that the AICA, as an arterial responsible is already relatively well depicted on 3D TOF-MRA because of its flow-related signal characteristics. In addition, the AICA showed substantial anatomic variation, and the anatomy of the cerebellopontine angle is complex. Thus, although combined imaging provides additional spatial information, this advantage may not be sufficient to translate into a significantly higher matching rate. In the VA/BA subgroup, no significant differences were observed among the three methods. This may be attributable to the relatively large caliber, stable anatomic location, and easier recognition of the morphology and course of the vertebrobasilar artery. By contrast, in subgroups with responsible venous vessels, the complete matching rate of 3D TOF-MRA was clearly lower than that of 3D FIESTA-c and the combined approach. This is most likely because 3D TOF-MRA has limited ability to depict venous structures with slow flow, small caliber, or complex courses, whereas 3D FIESTA-c can more clearly demonstrate the relationship between veins and the trigeminal nerve, including apposition, crossing, and compression. Accordingly, 3D FIESTA-c appears to have a distinct advantage in the identification of venous responsible vessels. In addition, no significant differences were found among the three methods in subgroups involving multiple responsible vessels. This may be because small tortuous vessels accompanying a major vessel, such as the vertebrobasilar artery, may be located within the narrow space between the major vessel and adjacent brain tissue and are therefore difficult to identify preoperatively. Alternatively, release of cerebrospinal fluid and vascular shift after craniotomy may alter the local neurovascular relationship, leading to discrepancies between imaging and intraoperative findings. The lack of statistical significance may also reflect the limited sample size of these rare subgroups, and these results should therefore be interpreted with caution. The present study suggests that the combination of 3D FIESTA-c and 3D TOF-MRA provides clearer visualization of responsible vessels in patients with trigeminal neuralgia and improves identification of most arterial and venous responsible vessels, in line with previous reports[14, 23, 24]. Compared with simply determining the presence or absence of vascular compression, accurate identification of responsible vessel type is of greater clinical relevance for preoperative MVD assessment, particularly in distinguishing isolated arterial, isolated venous, and combined vascular compression. Different responsible vessel types are associated with distinct neurovascular relationships and intraoperative management strategies. Accordingly, more accurate preoperative imaging may enable surgeons to develop an anatomical understanding that more closely reflects the operative field, thereby improving surgical planning and increasing the precision of decompression. This study has several limitations. First, it was a single-center retrospective study, and all included patients underwent MVD, which may have introduced selection bias. Second, imaging assessment was based on consensus between two experienced readers, and interobserver agreement was not evaluated. Third, the present study focused on responsible vessel detection and type matching, but did not examine the relationship between nerve morphological changes, compression severity, and postoperative pain relief. Fourth, some rare subgroups involving multiple responsible vessels had small sample sizes, limiting statistical power; therefore, these findings should be regarded as exploratory. Future prospective multicenter studies with larger sample sizes and inclusion of clinical outcome measures are warranted to further validate these results. Conclusion Both 3D TOF-MRA and 3D FIESTA-c provide valuable information for preoperative assessment of responsible vessels in patients with primary trigeminal neuralgia undergoing microvascular decompression. However, their diagnostic performance varies according to responsible vessel type. The combined approach significantly improves the accuracy of identifying the presence or absence of a responsible vessel, increases the complete matching rate for responsible vessel type, and shows better agreement with intraoperative findings. For venous-related responsible vessels, 3D FIESTA-c and the combined approach appear to be superior to 3D TOF-MRA alone, whereas for common arterial responsible vessels, the combined approach can further refine preoperative assessment. In clinical practice, complementary interpretation of multiple MRI sequences should therefore be emphasized according to the suspected responsible vessel type to improve the quality of preoperative evaluation and support surgical decision-making. Acknowledgements The authors are grateful to everyone who offered their time, insight, and assistance to this study. Funding The authors did not receive support from any organization for the submitted work. Data availability The data are available from the corresponding author on reasonable request. Declarations Conflicts of interest The authors report no conflicts of interest for this work. Ethical approval This retrospective study was reviewed and approved by the Institutional Ethics Committee of Nanjing Brain Hospital. Given the retrospective nature of the study and the use of de-identified clinical data, the requirement for written informed consent was waived by the ethics committee. All procedures were conducted in accordance with the Declaration of Helsinki and relevant institutional and national guidelines. Footnotes Yu Huang and Ying Huang are equally contributed to this work. References Associated Data Data Availability Statement The data are available from the corresponding author on reasonable request.