The transforaminal lumbar interbody fusion (TLIF) has become one of the most commonly used procedures for lumbar fusion in patients experiencing lumbar instability. While TLIF provides benefits, such as direct decompression and single-position surgery, it also has notable disadvantages related to the use of smaller cage sizes. These disadvantages include reduced fusion rates due to the smaller cage footprint, compromised biomechanics, and higher rates of subsidence. In contrast, anterior or lateral interbody fusion procedures generally demonstrate lower subsidence rates because they allow placement of larger cages that can support the apophyseal ring, thereby providing better stability and load distribution [1–4].

Numerous studies have investigated potential solutions to these challenges, including the use of larger cages or oblique lumbar interbody fusion cages as alternatives to traditional TLIF cages or expandable cages. However, the inappropriate use of larger cages can increase the risk of nerve root injury, especially in patients with a smaller Kambin’s triangle. Therefore, thorough preoperative planning, including magnetic resonance imaging (MRI) assessments, is essential for interbody cage selection and minimizing complications. Measuring the Kambin’s triangle is crucial before using a large cage. Previous studies have proposed measurement methods using axial MRI to assess the distance between the lateral border of the thecal sac and the medial border of the exiting nerve root [5,6]. However, these methods have limitations in certain cases, such as when the thecal sac is displaced or difficult to identify, as occurs in severe central canal stenosis.

Currently, several studies have examined Kambin’s width using axial MRI, magnetic resonance neurography, and cadaveric measurements [7–9]. However, research focusing on MRI accuracy for preoperatively predicting the corridor width in TLIF procedures remains limited.

This prospective study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Ethical approval was obtained from the Human Ethics Committee of Thammasat University (approval no., MTU-EC-OT-0-015/68). Informed consent was obtained from all participants.

This study included 60 consecutive patients aged 18 years and older who underwent a TLIF procedure between December 1, 2024, and March 31, 2025. Demographic and surgical data were collected, including age, sex, affected level, side of TLIF, type of approach (open, minimally invasive-TLIF [MIS-TLIF], or endoscopic approach [unilateral biportal endoscopic lumbar interbody fusion, UBE-LIF]), and the presence of scoliosis or moderate-to-severe stenosis. Patients lacking axial or coronal MRI data were excluded from the study. Scoliosis was assessed using plain radiographs, with patients classified as having scoliosis if the Cobb’s angle exceeded 10°. Central canal stenosis was evaluated on axial MRI using the Lee classification system. Moderate-to-severe stenosis was defined as Lee grade II (some aggregation of cauda equina fibers) or grade III (no separation of the cauda equina) [10].

The width of the Kambin’s triangle was measured at the upper and lower endplates using two different MRI methods: (1) axial MRI and (2) coronal MRI, following a protocol with 2 mm slice thickness for both axial and coronal planes.

For the axial MRI method, the distance between the lateral border of the thecal sac and the medial border of the exiting nerve root was measured at each level, using both the upper and lower endplate images (Fig. 1).

The coronal MRI method involved establishing a vertical reference line for the thecal sac by superimposing the sections that best represented its lateral borders at both the upper and lower vertebral body levels. (1) Identify the lateral border of the thecal sac at the lower vertebral body, typically at the mid-pedicle level, and mark its position using a cursor tool on the MRI slice that best visualizes this border (Fig. 2A). (2) Move to the MRI slice that best visualizes the exiting nerve root and mark the reference point corresponding to the lateral border of the thecal sac. (3) Identify the lateral border of the thecal sac at the upper vertebral body in the same manner, marking the reference point on the slice that best visualizes the exiting nerve root. In most cases, the slice showing the lateral border of the thecal sac at the upper vertebral body coincides with the slice where the exiting nerve root is most clearly visualized. (4) Draw a vertical reference line connecting the thecal sac reference points at both the upper and lower vertebral body levels to represent the decompressed thecal sac boundary. (5) Measure the distance between this vertical reference line and the medial border of the exiting nerve root (Fig. 2B).

Measurements were conducted by two evaluators, with the first evaluator performing the assessments twice, separated by a 2-week interval. The TLIF procedure was then performed, and intraoperative measurements of the upper and lower endplates of the Kambin’s triangle were obtained by a single spine surgeon (Fig. 3), who was blinded to the MRI measurements. All intraoperative measurements were performed after completing the facetectomy, once the exiting nerve root and the thecal sac were clearly identified. The corresponding upper and lower endplate levels were confirmed under fluoroscopic guidance to ensure anatomical accuracy before intraoperative measurements were obtained.

No previous studies have reported variance parameters for coronal MRI-based estimation of Kambin’s triangle width compared with intraoperative measurements. Therefore, a formal a priori effect-size calculation was not feasible. A precision-based equivalence framework was adopted using data obtained using data from our pilot study (n=30). The standard deviation (SD) of the difference between intraoperative and coronal MRI corridor widths (σ=SD [test-reference]) was 0.73 mm for the upper endplate and 1.40 mm for the lower endplate. Using α=0.05, β=0.20 (power=0.8), and an equivalence margin (δ) of 1.0 mm, the required sample sizes were calculated as follows:

The calculated n values were five for the upper and 17 for the lower endplate. The final study included 60 consecutive cases, exceeding these requirements and ensuring adequate power and precision.

Statistical analysis was conducted using R ver. 4.4.2 (The R Foundation for Statistical Computing, Vienna, Austria; 2024). A p-value of less than 0.05 was considered statistically significant. The analysis assessed the mean absolute error (MAE) and Pearson’s correlation coefficients (r) for each measurement method relative to the intraoperative measurements. Furthermore, both inter-observer and intra-observer reliability were evaluated and reported as the intraclass correlation coefficient (ICC).

Of the 60 patients included in our study, 63% (38 out of 60) were female, 37% (22 out of 60) were male, with a mean age of 68.81±7.16 years. Regarding the procedures performed, 52% (31 out of 60) underwent open TLIF, 36.67% (22 out of 60) underwent MIS-TLIF, and 11.67% (seven out of 60) underwent endoscopic TLIF. Additionally, 41.67% (25 out of 60) of patients had scoliosis (Table 1).

The mean intraoperative Kambin’s triangle width was 8.33±1.49 mm at the upper endplate and 12.82±2.03 mm at the lower endplate. The mean upper endplate Kambin’s triangle width measured from axial views did not significantly differ from the intraoperative measurements (9.27±3.85 mm, p=0.08). In contrast, the mean lower endplate Kambin’s triangle width from axial views showed a significant difference compared to intraoperative measurements (18.03±4.85 mm, p<0.001). The mean upper endplate Kambin’s triangle width measured from coronal views did not differ significantly from intraoperative measurements (8.04±1.41 mm, p=0.27), whereas the mean lower endplate width from coronal views was significantly different compared to intraoperative measurements (13.56±1.94 mm, p=0.04) (Table 2).

Correlation analysis revealed a high positive correlation for the upper endplate Kambin’s triangle width measured on coronal MRI (r=0.81, p<0.001) and for the lower endplate width measured on coronal MRI (r=0.82, p<0.001). In contrast, the correlation between Kambin’s triangle width at upper endplate measurements on axial MRI was weakly positive (r=0.18, p=0.15), while the correlation at the lower endplate measurements using axial MRI was moderately positive (r=0.49, p<0.001) (Table 2).

The MAE for widths measured on coronal MRI was significantly lower than that on axial MRI. At the upper endplate, the MAE was 0.70±0.61 mm for coronal MRI, compared with 2.69±2.90 mm for axial MRI (p<0.001). Similarly, at the lower endplate, the MAE was 1.00±0.98 mm for coronal MRI versus 5.53±4.58 mm for axial MRI (p<0.001) (Table 3).

There was no statistically significant difference in MAE between scoliosis and non-scoliosis cases for either measurement method (Table 3). Similarly, MAE did not differ significantly between patients with mild-to-moderate stenosis and those with severe stenosis when measured using either method (Table 3).

Both intra- and inter-observer reliability were generally higher for measurements obtained from the coronal view. Intra-observer agreement was excellent for lower endplate measurements on coronal views (ICC, 0.874; 95% CI, 0.795–0.924) and substantial for upper endplate measurements (ICC, 0.721; 95% CI, 0.567–0.825). In contrast, axial view measurements showed moderate intra-observer agreement for the lower endplate (ICC, 0.578; 95% CI, 0.376–0.727) and fair agreement for the upper endplate (ICC, 0.297; 95% CI, 0.042–0.516) (Table 4).

A similar pattern was observed for inter-observer agreement: excellent agreement for lower endplate measurements on coronal views (ICC, 0.850; 95% CI, 0.757–0.909) and substantial agreement for upper endplate measurements (ICC, 0.723; 95% CI, 0.571–0.828), but only moderate agreement for both upper and lower endplate measurements on axial views (ICC, 0.557 [95% CI, 0.346–0.713] and ICC, 0.467 [95% CI, 0.235–0.649], respectively) (Table 4).

This study aimed to evaluate the correlation between MRI measurements of the Kambin’s triangle width in coronal and axial sections with intraoperative distances. Our findings indicate that coronal MRI measurements of both upper and lower endplates strongly correlate with intraoperative distances, whereas axial MRI measurements show a weak correlation for the upper endplate and a moderate correlation for the lower endplate. Furthermore, subgroup analysis showed that the accuracy of these measurements was not significantly influenced by the presence of severe spinal stenosis or scoliosis. These findings suggest that coronal MRI is a more reliable method than axial MRI for preoperative assessment of the Kambin’s triangle width.

The significantly lower MAE observed with coronal MRI compared to axial MRI further supports the superiority of coronal imaging for accurate preoperative planning. This finding is particularly relevant given the increasing use of larger cages in TLIF procedures, where precise measurement of the Kambin’s triangle is essential for the appropriate selection of the TLIF cage size.

Previous studies have examined the utility of a large cage in TLIF or extreme TLIF procedures to increase cage footprint, improve fusion rate, and decrease subsidence rate. You et al. [3] demonstrated a significantly lower subsidence rate in the large-size titanium cage group (16 mm width), compared to the regular-size titanium cage group (11 mm width), with a comparable fusion rate. Although several studies have attempted to define the safe zone, morphology [11] or average size of Kambin’s triangle [7,8,12,13] to guide safe TLIF procedures, no gold standard for measurement has been established, as none have directly compared imaging-based methods with intraoperative findings or cadaveric measurements.

Prior studies by Heo et al. [5] and Xu et al. [6] investigated the use of larger cages in UBE-LIF and proposed measurement methods using axial MRI, followed by re-evaluation with intraoperative measurement. Xu et al. [6] reported a significant increase in disc space height, while neither study provided data on complications. Similarly, Eum et al. [14] evaluated the outcomes of extreme TLIF and demonstrated an increase in disc space height, segmental lordosis, and lumbar lordosis compared to preoperative radiographs. Kambin’s triangle was also measured on axial MRI and re-evaluated with intraoperative measurement. No dural tears were reported in this study.

Unlike previous studies, our study highlights the limitations of axial MRI in accurately assessing Kambin’s triangle. In cases with displaced thecal sacs, such as scoliosis or severe central canal stenosis, axial MRI measurements demonstrated reduced accuracy and reliability, as indicated by slightly higher MAE in these cases. Across all cases, coronal MRI measurements demonstrated significantly better MAE and higher intra- and inter-observer agreement. This improved accuracy and reliability is likely attributable to the clearer visualization of the exiting nerve root. Furthermore, using a vertical reference line to define the lateral border of the thecal sac on coronal MRI can help accurately define the thecal sac, even in the presence of pathological changes or anatomical distortions. This approach provides a stable anatomical landmark that remains relatively unaffected by local disc herniation, facet hypertrophy, or foraminal narrowing. It is particularly useful in representing the true thecal sac anatomy after a TLIF procedure, once the compressive pathology has been removed.

The findings of our study have important implications for clinical practice. In our experience, preoperative MRI assessment of the Kambin’s triangle is valuable for planning cage size and approach selection. Specifically, a conventional TLIF cage or expandable cage was used in patients with a smaller Kambin’s width (<12 mm), whereas a larger cage was chosen for those with wider corridors. This strategy allowed appropriate cage size selection without any cases of iatrogenic durotomy or nerve root injury. These findings emphasize the potential value of MRI-based assessment in preoperative planning and underscore the need for future studies translating imaging metrics to surgical and clinical outcomes. Beyond cage size selection, we have identified another application of coronal MRI measurements: determining the optimal side for TLIF in patients experiencing bilateral symptoms. In cases where TLIF is feasible on both sides, coronal MRI can help identify the side with the larger corridor, potentially allowing a safer surgical procedure.

Our study has several limitations. First, the relatively small sample size; a larger cohort might have strengthened our results. Second, limitations related to blinding should be acknowledged. In our setting, it was not feasible to blind the operating surgeon from the preoperative MRI, and thus, complete MRI blinding could not be achieved for the intraoperative measurements. However, we attempted to maximize blinding where possible. The surgeon performing the procedure did not have access to any MRI measurement results, and the two MRI evaluators were independent. Future studies should evaluate clinical outcomes and complications, especially nerve root injury and dura tears, as well as the actual cage sizes used when coronal MRI is applied for preoperative evaluation. Additionally, evaluating the applicability of this measurement method to other similar procedures could be valuable. Finally, our study focuses on measuring the specific expanded corridor [15], a common procedure to open TLIF, MIS-TLIF, and UBE-LIF. However, for percutaneous lumbar interbody fusion, where the superior articular process (SAP) of the facet joint is preserved, this measurement may not be appropriate for preoperative planning.

This study demonstrates that coronal MRI measurements of the Kambin’s triangle width strongly correlate with actual intraoperative distances and exhibit a significantly lower MAE compared to axial MRI, indicating superior accuracy and reliability for preoperative planning. These findings may improve preoperative surgical planning, facilitate appropriate cage size selection, and reduce the risk of complications.

However, the study’s small sample size limits the generalizability of these results, and further research is necessary to evaluate clinical outcomes, including nerve injury and dural tears. Additionally, while our focus was on the expanded corridor applicable to open TLIF, MIS-TLIF, and UBE-LIF, these measurements may not be suitable for percutaneous lumbar interbody fusion, where the SAP is preserved.