Enhanced Quantitative Assessment of Lumbar Facet Joint Degeneration: Refining the Joint Degeneration Grading System With Grayscale Curve Fitting and Imaging Parameter Authors: Xiaowen Liu, Mingjian Sun, Shouyu He, Hao Zhang, Runze Li, Jiangang Shi, Xuexiao Ma Published: 2025 ABSTRACT ABSTRACTBackgroundLow back pain (LBP) is a leading cause of disability, often progressing to chronic low back pain (CLBP), which often involves facet joint degeneration. The Weishaupt grading system is widely used to evaluate facet joint degeneration, but precise, quantitative methods for early diagnosis are lacking. This study aims to develop a predictive model for facet joint degeneration using magnetic resonance imaging (MRI)‐derived curve fitting and some radiographic parameters, enhancing early, quantitative assessment and providing more accurate analysis for effective treatment planning.MethodsThis study included 48 patients (161 joints), aged 20–80 years. Imaging parameters such as intervertebral space height (ISH), lumbar range of motion (ROM‐L), intervertebral disc degeneration, facet joint degeneration, facet joint cross‐sectional area (FJCSA), vertebral bone quality (VBQ) scores, subcutaneous fat tissue thickness (SFTT) and grayscale values of lumbar facet joint were analyzed. Statistical correlations were evaluated using spearman rank correlation and ordinal logistic regression. Curve fitting was performed to analyze and model the grayscale value changes in lumbar facet joints.ResultsCorrelation analysis revealed positive correlations between facet joint degeneration and intervertebral disc degeneration, FJCSA and SFTT, while negative correlations were observed with ISH and ROM‐L. Ordinal logistic regression identified ROM‐L and FJCSA as significant factors influencing degeneration. The locally estimated scatterplot smoothing (LOESS) curve fitting method of grayscale value for facet joint degeneration provided a more accurate predictive model, refining the joint degeneration grading system for improved quantitative assessment.ConclusionThe study identified some radiographic factors, including ISH, FJCSA, ROM‐L, SFTT and intervertebral disc degeneration, associated with facet joint degeneration. This study refined the joint degeneration grading system using LOESS curve fitting, providing a more quantitative approach for evaluating facet joint degeneration. This grading system offered greater precision and reliability in clinical assessments, supporting better diagnostic accuracy for patients with facet joint‐related CLBP. Introduction Low back pain (LBP) is the most common cause of disability worldwide, with a lifetime prevalence of up to 70%–85% [1]. For many individuals, LBP lasting more than 12 weeks will progress to chronic low back pain (CLBP) [2, 3]. Due to long‐term functional loss, CLBP leads to decreased work efficiency and increased healthcare costs [4, 5]. Persistent back pain can be caused by various factors, including disc, facet joint and ligament dysfunction; facet joint dysfunction is considered a key factor in the development of CLBP [6]. Facet joints, also known as zygapophyseal joints, are synovial joints located between the superior and inferior articular processes of adjacent vertebrae, playing a crucial role in the stability and flexibility of the lumbar spine [7]. Degenerative changes in these joints typically involve cartilage erosion, subchondral bone remodeling, osteophyte formation and synovitis, all of which can lead to mechanical instability and chronic pain [8]. Literature suggests that over 30% of patients experience LBP originating from facet joint degeneration [9]. The facet joints are very important components of the posterior column in spinal motion units and have attracted extensive attention in academic research [10]. However, research on the early, quantitative and precise diagnosis of facet joint degeneration remains insufficient. In addition, the Weishaupt grading system has become the gold standard for evaluating lumbar facet joint degeneration [11]. One of the key strengths of the Weishaupt system is its clear correlation between imaging findings and clinical symptoms, which provides surgeons with a practical tool for diagnosing the severity of lumbar facet joint pathology and tailoring treatment plans accordingly. Given its ability to evaluate facet joint degeneration, the aim of this study is to predict facet joint degeneration based on some radiographic parameters, such as intervertebral space height (ISH), lumbar range of motion (ROM‐L), intervertebral disc degeneration (based on the Pfirrmann grade), facet joint degeneration (based on the Weishaupt classification), facet joint cross‐sectional area (FJCSA), vertebral bone quality (VBQ) scores, subcutaneous fat tissue thickness (SFTT) and grayscale values of lumbar facet joint, which will allow for a more precise and quantitative analysis of joint degeneration. Magnetic resonance signals are time‐varying currents or voltages. These signals can be described and defined by amplitude, frequency and phase [12, 13]. Amplitude, also known as signal intensity, reflects the final brightness of pixels in magnetic resonance imaging (MRI). This signal intensity is often referred to as the grayscale value. In previous literature, studies on MRI grayscale values have mainly focused on tasks such as brain image segmentation [14], diagnosing focal cortical dysplasia (FCD) lesions [15] and quantifying muscle volume and fat in the hand [16]. However, exploration of the application of grayscale values in spinal imaging is limited, particularly regarding facet joint degeneration. Currently, curve fitting based on large datasets has become an essential part of predictive model development [17, 18]. This study aims to construct a predictive model for facet joint degeneration using curve fitting based on changes in MRI facet joint grayscale values. Additionally, it aims to provide a quantitative description of the facet joint degeneration grading system, offering more accurate analysis. Materials and Methods Inclusion and Exclusion Criteria All patients met the following inclusion criteria: (1) Age range from 20 to 80 years old; (2) Persisted LBP more than 12 weeks [19]; (3) Complete preoperative x‐ray and 1.5T MRI examination. Exclusion criteria included: (1) scoliosis, Cobb angle > 20° [20]; (2) lack of preoperative x‐ray or MRI examination; (3) 3.0T MRI; (4) the interval between preoperative x‐ray and MRI examinations is more than 1 month; (5) previous lumbar spine internal fixation surgery; (6) Lumbar spondylolisthesis > 5 mm, with severe dislocation of facet joints; (7) intravertebral bone cement implantation; (8) severe spinal joint disease (e.g., mandatory spondylitis); (9) spinal tumors; (10) the maximum extension depth of modic changes (MC) relative to vertebral body height was greater than 25% [21]; (11) vertebral axial rotation: The coronal imaging demonstrates asymmetry between the spinous process midpoint and the medial borders of the pedicles on both sides, or Nash–Moe rotation being equal or greater than grade I on the standing anteroposterior (AP) radiograph [22, 23]. Study Population Patients aged ≥ 20 years who required imaging for the evaluation of lumbar facet joint degeneration between September 2020 and July 2024 were included. A total of 161 joints (from 48 patients) were initially included. Of these, 10 joints were excluded for the patients had only 3.0T MRI; 1 patient had Modic changes; 6 joints from three patients were excluded for lacking MRI examinations; 1 patient had scoliosis; 1 patient had lumbar spondylolisthesis; two patients had an interval of more than 1 month between preoperative x‐ray and MRI examinations and 2 joints from the same patient were excluded for the severe spinal joint disease (mandatory spondylitis) (total number of joints = 23) (Figure 1). The remaining 138 joints from 39 patients (mean age, 58.79 ± 10.23 years; men: women, 19:20 [mean ages, 57.16 ± 11.00 and 60.35 ± 9.44 years, respectively]) were eventually included in our analysis. The case collection flowchart. Imaging Equipment and Position Each subject underwent x‐ray imaging of the lumbar spine (L1‐S1, SIEMENS, Fluorospot Compact FD, Germany, field of view (FOV): approximately 200 mm, image matrix: 2600 × 1600, image zoom: 1.00) in a standing position. MR imaging was performed on the lumbar region of patients using a clinical 1.5 T MR system (GE MEDICAL SYSTEMS, SIGNA EXCITE, USA). Imaging parameters were as follows: The MRI sequences included axial T1‐weighted (Time of Echo (TE): 9 ms, Time of Repetition (TR): 420 ms, Echo train: 3; image zoom: 1.00, slice thickness: 4 mm, flip angle: 90°), axial T2‐weighted (TE: 100 ms, TR: 2500 ms, Echo train: 26; image zoom: 1.00, slice thickness: 4 mm, flip angle: 90°) and transverse T2‐weighted MRI sequences (TE: 110 ms, TR: 3200 ms, Echo train: 24, image matrix: 256 × 256, image zoom: 1.00, slice thickness: 3 mm, flip angle: 90°). All patients underwent lumbar spine x‐rays in the standing anteroposterior and lateral position. Following standard radiographic procedures, patients were positioned upright facing the radiation source, with the midsagittal plane aligned with the midline of the table and Bucky [24]. The vertical central ray was directed to the midline of L3 at the level of the inferior costal margin. To maintain consistency in positioning, patients were assisted by research staff throughout the process. In addition, patients were instructed to perform standardized lumbar flexion and extension movements under the supervision of an experienced radiology technician to ensure reproducibility. For the flexion view, patients were asked to stand upright with knees fully extended and then bend forward as far as comfortably possible while keeping their pelvis stable and their arms naturally hanging down. For the extension view, patients returned to the upright position and then extended the trunk backward to the maximal extent tolerated without bending the knees or shifting the pelvis. The technician carefully monitored body posture to minimize compensatory hip motion and ensure consistent lumbar movement across all subjects. All imaging procedures were standardized. The radiographs and MRI scans were performed by the same experienced radiologist and stored in the picture archiving and communication system (PACS). Measurements and evaluations of the facet joints were subsequently carried out using this system. Imaging Parameters Imaging parameters analyzed included ISH, ROM‐L, intervertebral disc degeneration (based on the Pfirrmann grade), facet joint degeneration (based on the Weishaupt classification), FJCSA, VBQ scores, SFTT and grayscale values of lumbar facet joint. (1) Intervertebral space height: The mean of anterior and posterior ISH (Figure 2a) [25]; (2) Lumbar range of motion: The difference between the angle of intervertebral space in flexion and extension position (Figure 2b,c) [26]; (3) Facet joint cross‐sectional area: Based on MRI cross‐sectional images, when the reference line is positioned at the mid‐intervertebral disc level, the corresponding cross‐sectional area of the lumbar facet joint is defined as the FJCSA (Figure 3) [27]; (4) Vertebral bone quality: The VBQ score is a recently developed method for assessing trabecular bone using MRI. Following a previously described approach, the VBQ score was derived from the signal intensity (SI) obtained from T1‐weighted, non‐contrast MRI images of the lumbar spine. Mid‐sagittal slices were utilized for the analysis. Initially, regions of interest (ROI) were placed within the trabecular bone of the L1–L4 vertebral bodies and in the cerebrospinal fluid (CSF) space at the L2 or L3 levels. The VBQ score was then calculated using the following formula: VBQ score = SIL1−L4/SICSF (Figure 4a) [28]; (5) Subcutaneous fat tissue thickness: The SFTT at L1 level in the study represents the distance between the anterior aspect of dorsal subcutaneous fat and the back skin (Figure 4b) [29, 30]; (6) Grayscale values of lumbar facet joint: Based on T2‐weighted MRI cross‐sectional images, when the reference line is at the mid‐level of the intervertebral disc, a perpendicular bisector is drawn across the facet joint space, ensuring that the midpoint of this line coincides with the midpoint of the facet joint space. The grayscale values corresponding to 20 points along this line are recorded using the PACS system. A line graph is then generated using the Excel system (Figure 5a,b). Measurement standard of ISH and ROM‐L in radiography. (a) ISH; (b) ROM‐L in extension position; (c) ROM‐L in flexion position. Measurement standard of FJCSA. Measurement standard of VBQ score and SFTT. (a) VBQ score; (b) SFTT. Measurement standard of grayscale values of lumbar facet joint. (a) Grayscale values of lumbar facet joint in PACS system; (b) Line graphs of grayscale values for lumbar facet joint in Excel system. Reference Standards To reduce errors, a senior radiologist and an orthopedic specialist collaborated to define the criteria to compute the aforementioned imaging parameters. These criteria were explained to two attending physicians with 1–5 years of clinical experience. Then these two independent attending physicians, blinded to each other's evaluations, were tasked with assessing and grading the included cases. Any discrepancies were resolved through consensus. All image datasets were anonymized, coded and randomized to maintain blinding during the evaluation process. Statistical Analysis The correlations between the measured radiographic parameters, such as the ISH, ROM‐L, FJCSA, intervertebral disc degeneration, VBQ and SFTT and the facet joint degeneration were assessed using the Spearman rank correlation test if the normality test (i.e., the Kolmogorov–Smirnov test) was satisfied. The Mann–Whitney U‐test (two sided) was used if the normality test failed. A p value less than 0.05 was considered statistically significant. Correlation coefficients (r) and coefficients of determination (r 2) were recorded and used to quantify the correlation. Additionally, the inter‐rater reliability of radiographic measurements, including the ISH, ROM‐L, FJCSA, intervertebral disc degeneration, VBQ, SFTT and the facet joint degeneration was assessed using the intraclass correlation coefficient (ICC). The inter‐rater reliability for these parameters was evaluated as follows: 0–0.20, poor; 0.21–0.40, fair; 0.41–0.60, moderate; 0.61–0.80, good and 0.81–1.00, excellent agreement. Test of parallel lines was performed before the ordinal logistic regression. All the aforementioned imaging parameters were used in an ordinal logistic regression to identify the factors affecting facet joint degeneration (ordinal outcome). Data were expressed as regression coefficient (β) [95% confidence interval (CI)]. The collected data is compiled in Excel 2021 (Microsoft Excel, Microsoft Corp., Redmond, WA, USA) and line graphs were generated to visualize the grayscale value changes of the facet joints. All the above statistical analyses were performed using IBM SPSS Statistics version 26.0 (IBM Corp., Armonk, NY, USA). The line graph of the grayscale values of the zygapophyseal joints was first normalized to mitigate the influence of the absolute grayscale values on subsequent model construction, as this study focuses more on the trend of grayscale value changes due to degeneration of the zygapophyseal joints. Curve fitting was performed for the facet joints with different degeneration grades, using various models, including locally estimated scatterplot smoothing (LOESS), generalized additive model (GAM), mean curve fitting, spline fitting, polynomial regression fitting, mixed‐effects model and principal component analysis (PCA), to fit the grayscale value curves of the zygapophyseal joints. Due to the inability of polynomial regression fitting, mixed‐effects models, PCA to handle complex nonlinear curves, these methods were excluded from further analysis. The span for LOESS was set to 0.2 to ensure that the curve fits the data as closely as possible while maintaining smoothness. The accuracy of the different fitting methods was evaluated by comparing the residual sum of squares (RSS) and mean squared error (MSE), both of which reflect the deviation between the fitted curve and the actual data and are used to assess the model's accuracy. Smaller values of RSS and MSE indicate better performance of the curve fitting model. All statistical analyses were performed using R version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria). Ethical Review The study protocol was approved by the Medical Institutional Review Board of the Hospital. All the methods were performed in accordance with the relevant guidelines and regulations. The trial was registered at ClinicalTrials. gov (NCT06786611). Result We collected data continuously from September 2020 to July 2024, retrospectively reviewing information from 48 patients (161 joints). Ultimately, 39 patients (138 joints) were included in the study. Patient characteristics are presented in Table 1. The mean age of the patients was 58.79 ± 10.23 years (range: 20–80 years), and women accounted for 51.3% of all patients. No significant differences were found between patients in terms of age or sex (p > 0.05). Among the 138 facet joints, 2 were at the L1/2 level, 12 at L2/3, 51 at L3/4, 69 at L4/5 and 4 at L5/S1. The inter‐rater reliability was high for the following parameters: ISH (ICC: 0.87), ROM‐L (ICC: 0.82), intervertebral disc degeneration (ICC: 0.81), facet joint degeneration (ICC: 0.83), FJCSA (ICC: 0.83), VBQ scores (ICC: 0.86) and SFTT (ICC: 0.85). These results suggest that the two attending physicians involved in grading facet joint degeneration have a strong theoretical and practical foundation, supporting the reliability of this study. TABLE 1Summary of characteristics of patients in the study.Patient characteristicsNumber (N)Total (joints)138Total (patients)39L1/22L2/312L3/451L4/569L5/S14Age (years)58.79 ± 10.23Sex19 males, 20 femalesISH (mm)12.00 ± 2.69ROM‐L (°)4.23 ± 2.56FJCSA (mm2)224.93 ± 58.01VBQ scores3.23 ± 0.70SFTT (mm)19.37 ± 7.10Facet joint degeneration (based on the Weishaupt classification) (n)Grade 123Grade 241Grade 344Grade 430Intervertebral disc degeneration (based on the Pfirrmann grade) (n)Grade 17Grade 232Grade 354Grade 430Grade 515 Note: Data were presented as mean ± standard deviation (SD).Abbreviations: FJCSA: Facet joint cross‐sectional area; ISH: intervertebral space height; ROM‐L: Lumbar range of motion; SFTT: subcutaneous fat tissue thickness; VBQ: vertebral bone quality scores. Summary of characteristics of patients in the study. Note: Data were presented as mean ± standard deviation (SD). Abbreviations: FJCSA: Facet joint cross‐sectional area; ISH: intervertebral space height; ROM‐L: Lumbar range of motion; SFTT: subcutaneous fat tissue thickness; VBQ: vertebral bone quality scores. In this study, the univariate correlation analysis revealed that intervertebral disc degeneration, FJCSA and SFTT were positively correlated with facet joint degeneration, with correlation coefficients (r) of 0.4511, 0.4714 and 0.2633, respectively. In contrast, intervertebral space height and lumbar range of motion were negatively correlated with facet joint degeneration, with correlation coefficients of −0.4333 and −0.7541, respectively. The scatter plots were presented in Figure 6. However, no statistically significant correlation was found between facet joint degeneration and VBQ score (p = 0.16) (Table 2). Therefore, according to Cohen's classification, the correlations between ROM‐L and facet joint degeneration were strong, whereas those between ISH, FJCSA and intervertebral disc degeneration were moderate, and SFTT showed a weak correlation. Scatter plots showing the correlations between Weishaupt grades and key radiographic parameters: (a) FJCSA; (b) SFTT; (c) ROM‐L; (d) ISH. TABLE 2Correlation coefficients, coefficients of determination and statistical differences for degeneration of facet joint.ParametersCorrelation coefficient (r)Coefficients of determination (r 2) p Intervertebral disc degeneration0.45110.2035 p < 0.001* FJCSA0.47140.2222 p < 0.001* SFTT0.26330.0693 p = 0.0018* ISH−0.43330.1877 p < 0.001* ROM‐L−0.75410.5687 p < 0.001* VBQ score−0.12040.0145 p = 0.1594Abbreviations: FJCSA: Facet joint cross‐sectional area; ISH: intervertebral space height; ROM‐L: Lumbar range of motion; SFTT: subcutaneous fat tissue thickness; VBQ: vertebral bone quality.*Represents statistically significant. Correlation coefficients, coefficients of determination and statistical differences for degeneration of facet joint. Abbreviations: FJCSA: Facet joint cross‐sectional area; ISH: intervertebral space height; ROM‐L: Lumbar range of motion; SFTT: subcutaneous fat tissue thickness; VBQ: vertebral bone quality. Represents statistically significant. Before performing ordinal logistic regression, a test of parallel lines was conducted, and the p value was 0.06, which is greater than 0.05, thereby satisfying the test of parallel lines. The ordinal logistic regression analysis showed that some variables, such as the ROM‐L (β = −0.669, p < 0.001) and FJCSA (β = 0.011, p = 0.004) were statistically significantly associated with the facet joint degeneration. These variables appeared to be factors influencing the degeneration of facet joints. However, we must emphasize that although the FJCSA showed a statistically significant association with facet joint degeneration in the ordinal logistic regression, its regression coefficient was relatively small, indicating a weak effect size. Additionally, the ISH (p = 0.307), intervertebral disc degeneration (p > 0.05), VBQ scores (p = 0.150) and SFTT (p = 0.719) showed no statistically significant correlation with the degeneration of facet joints (Table 3). TABLE 3The ordinal logistic regression analysis between the imaging parameters and facet joint degeneration.Independent variableRegression coefficient (β)95% CI p FJCSA0.011(0.003, 0.019) p = 0.004* ROM‐L−0.669(−0.904, −0.435) p < 0.001* ISH−0.097(−0.283, 0.089) p = 0.307VBQ scores−0.405(−0.957, 0.147) p = 0.150SFTT−0.009(−0.061, 0.402) p = 0.719Intervertebral disc degenerationNN p > 0.05Abbreviations: CI: Confidence interval; FJCSA: Facet joint cross‐sectional area; ISH: intervertebral space height; N: not applicable; ROM‐L: Lumbar range of motion; SFTT: subcutaneous fat tissue thickness; VBQ: vertebral bone quality.*Represents statistically significant. The ordinal logistic regression analysis between the imaging parameters and facet joint degeneration. Abbreviations: CI: Confidence interval; FJCSA: Facet joint cross‐sectional area; ISH: intervertebral space height; N: not applicable; ROM‐L: Lumbar range of motion; SFTT: subcutaneous fat tissue thickness; VBQ: vertebral bone quality. Represents statistically significant. To achieve a more quantitative and precise analysis of the facet joint degeneration grading system, we performed curve fitting analysis on the grayscale value curves of 138 facet joints included in the study. The accuracy of four different fitting models—LOESS, GAM, mean curve fitting and spline fitting—was evaluated by calculating their RSS and MSE. The results demonstrated that the LOESS model provided the most accurate fit for describing grayscale value curves of facet joints (Tables 4 and 5). Ultimately, this study selected the LOESS fitting model to simulate the grayscale value curves of facet joint degeneration across different Weishaupt grades, generating an optimal standard fit curve for subsequent interpretation and exploration of its clinical significance. TABLE 4RSS of four curve fitting methods.Weishaupt grading systemLOESSGAMMean curve fittingSpline fittingGrade 11.960597e‐30 a 0.055397860.304966.529735e‐17Grade 21.685574e‐30 a 0.007747768101.47461.615867e‐17Grade 32.022226e‐30 a 0.002211251159.33037.213687e‐18Grade 41.588507e‐30 a 0.0032793388.398540.0006491545Abbreviations: GAM: Generalized additive model; LOESS: Locally estimated scatterplot smoothing; RSS: Residual sum of squares. a Represents the smallest RSS value, indicating that the corresponding fitting model is the most accurate. RSS of four curve fitting methods. Abbreviations: GAM: Generalized additive model; LOESS: Locally estimated scatterplot smoothing; RSS: Residual sum of squares. Represents the smallest RSS value, indicating that the corresponding fitting model is the most accurate. TABLE 5MSE of four curve fitting methods.Weishaupt grading systemLOESSGAMMean curve fittingSpline fittingGrade 19.802983e‐32 a 0.002769890.13109773.264868e‐18Grade 28.427869e‐32 a 0.00038738840.12374958.079336e‐19Grade 31.011113e‐31 a 0.00011056250.18105713.606843e‐19Grade 47.942535e‐32 a 0.00016396650.14733093.245773e‐05Abbreviations: GAM: Generalized additive model; LOESS: Locally estimated scatterplot smoothing; MSE: Mean squared error. a Represents the smallest MSE value, indicating that the corresponding fitting model is the most accurate. MSE of four curve fitting methods. Abbreviations: GAM: Generalized additive model; LOESS: Locally estimated scatterplot smoothing; MSE: Mean squared error. Represents the smallest MSE value, indicating that the corresponding fitting model is the most accurate. Based on the results of LOESS curve fitting analysis, we made the following facet joint degeneration grading system: Grade 1:The average signal intensity within the facet joint space is slightly higher than or equal to the average signal intensity of the adjacent facet joint bones on both sides. The boundaries between the facet bone, subchondral cartilage and synovial fluid are clearly defined. This grade typically represents normal facet joints.Grade 2:The signal intensity within the facet joint space is reduced, approximately 50% of the average signal intensity of the adjacent facet joint bones. Additionally, the signal intensity of the bones on both sides becomes heterogeneous, with mixed signals appearing. At this stage, the joint space shows slight narrowing compared to Grade 1.Grade 3:The high signal within the facet joint space nearly disappears, with the facet joint cartilage becoming significantly thinner or even absent, and a wide low‐signal region emerging. The average signal intensity of the superior facet bone shows a decreasing trend compared to the inferior facet bone. At this stage, the joint space further narrows.Grade 4:Severe degeneration of the facet joint is accompanied by pronounced osteophyte formation and significant bone proliferation. The signal intensity of the superior facet bone is significantly reduced compared to the inferior facet bone, often accompanied by marked bone destruction. The boundaries of the facet bone, subchondral cartilage and other structures are indistinct, and the joint space becomes unrecognizable. Grade 1:The average signal intensity within the facet joint space is slightly higher than or equal to the average signal intensity of the adjacent facet joint bones on both sides. The boundaries between the facet bone, subchondral cartilage and synovial fluid are clearly defined. This grade typically represents normal facet joints. Grade 1: The average signal intensity within the facet joint space is slightly higher than or equal to the average signal intensity of the adjacent facet joint bones on both sides. The boundaries between the facet bone, subchondral cartilage and synovial fluid are clearly defined. This grade typically represents normal facet joints. Grade 2:The signal intensity within the facet joint space is reduced, approximately 50% of the average signal intensity of the adjacent facet joint bones. Additionally, the signal intensity of the bones on both sides becomes heterogeneous, with mixed signals appearing. At this stage, the joint space shows slight narrowing compared to Grade 1. Grade 2: The signal intensity within the facet joint space is reduced, approximately 50% of the average signal intensity of the adjacent facet joint bones. Additionally, the signal intensity of the bones on both sides becomes heterogeneous, with mixed signals appearing. At this stage, the joint space shows slight narrowing compared to Grade 1. Grade 3:The high signal within the facet joint space nearly disappears, with the facet joint cartilage becoming significantly thinner or even absent, and a wide low‐signal region emerging. The average signal intensity of the superior facet bone shows a decreasing trend compared to the inferior facet bone. At this stage, the joint space further narrows. Grade 3: The high signal within the facet joint space nearly disappears, with the facet joint cartilage becoming significantly thinner or even absent, and a wide low‐signal region emerging. The average signal intensity of the superior facet bone shows a decreasing trend compared to the inferior facet bone. At this stage, the joint space further narrows. Grade 4:Severe degeneration of the facet joint is accompanied by pronounced osteophyte formation and significant bone proliferation. The signal intensity of the superior facet bone is significantly reduced compared to the inferior facet bone, often accompanied by marked bone destruction. The boundaries of the facet bone, subchondral cartilage and other structures are indistinct, and the joint space becomes unrecognizable. Grade 4: Severe degeneration of the facet joint is accompanied by pronounced osteophyte formation and significant bone proliferation. The signal intensity of the superior facet bone is significantly reduced compared to the inferior facet bone, often accompanied by marked bone destruction. The boundaries of the facet bone, subchondral cartilage and other structures are indistinct, and the joint space becomes unrecognizable. We presented in Figures 7 and 8 the optimal grayscale fitting curves and representative MRI images for each grade of facet joint degeneration to enhance interpretability and visual clarity. The optimal fit curve of facet joint degeneration grayscale value with different Weishaupt grades (The red line represents the best fit curve for different grades, and the red area represents the 95% confidence interval). (a) Grade 1; (b) Grade 2; (c) Grade 3; (d) Grade 4. Representative MRI of facet joint degeneration with different Weishaupt grades. (a) Grade 1; (b) Grade 2; (c) Grade 3; (d) Grade 4. Discussion Low back pain has been a leading health issue for decades and remains a major cause of disability worldwide. The prevalence of LBP is increasing, with lumbar facet joint degeneration being one of the most frequent causes [31, 32]. Facet joint disease, often linked to chronic arthritis or segmental degeneration, continues to be a common source of CLBP in individuals of all ages, affecting both younger and older populations [31]. Facet joints, as a crucial component of the posterior column in the spinal motion unit, have long been the subject of considerable attention [33]. In recent years, the development and application of new concepts in the comprehensive evaluation of the lumbar spine, such as VBQ scores and SFTT, have provided a more holistic understanding of lumbar degenerative diseases. However, we found that these emerging concepts lack deeper exploration and analysis, particularly in studying their relationship with facet joint degeneration [34]. Our study demonstrated that the FJCSA and ROM‐L were significant predictors of facet joint degeneration. The positive correlation between FJCSA and degeneration supported the notion that the enlargement of the facet joint, likely resulting from hypertrophy and osteophyte formation, plays a role in the degenerative process [35]. This finding was consistent with previous research that has identified joint hypertrophy as a hallmark of facet degeneration [36]. Additionally, the inverse relationship between ROM‐L and facet joint degeneration reflected the correlation between reduced spinal mobility (possibly due to joint stiffness and cartilage wear) and small joint degeneration [37, 38]. This observation aligned with studies emphasizing restricted lumbar mobility in the presence of advanced degenerative changes. However, we need to clarify that FJCSA demonstrated a moderate correlation in the univariate analysis. Compared with ROM‐L, it showed a smaller regression coefficient in the ordinal logistic regression analysis, indicating a limited independent effect size (i.e., a lower potential causal influence). Therefore, although FJCSA can be regarded as a statistically significant associated factor, its independent impact on facet joint degeneration may be limited relative to ROM‐L. Disc degeneration and ISH had been classically linked with facet joint pathology, which was consistent with the conclusions from the correlation analysis in our study [39]. However, the ordinal logistic regression analysis revealed that ISH and intervertebral disc degeneration (Pfirrmann grade) were not significantly correlated with facet joint degeneration. The regression analysis results suggest that these relationships might be more complex, possibly mediated by other factors such as vertebral alignment or mechanical load distribution [40, 41]. The lack of a correlation between ISH and facet degeneration could be attributed to the compensatory mechanisms that maintain spinal stability despite reduced disc height, delaying the onset of facet degeneration. Previous literature has reported an association between lumbar disc degeneration and the subcutaneous fat tissue thickness at the L1 disc level measured by MRI [30, 42]. However, there has been no consensus on the correlation between SFTT and facet joint pathology. A novel finding of our study was the positive correlation between subcutaneous fat tissue thickness and facet joint degeneration. This result highlights the potential role of metabolic factors in spinal degeneration. Obesity and increased body fat have been associated with higher mechanical loads on the spine, accelerating degenerative changes in both intervertebral discs and facet joints [43, 44]. However, the results of the logistic regression analysis indicate that SFTT is not an independent risk factor for facet joint degeneration. This suggests that further research is needed to explore the relationship between the two. Since Professor Ehresman J. introduced the concept of VBQ score, it has gained widespread attention [45]. Previous studies have demonstrated a significant correlation between the VBQ score and both the vertebral T‐score and absolute bone mineral density (BMD) [46]. In this study, we incorporated the VBQ scoring standard for assessing vertebral osteoporosis into the analysis of its correlation with facet joint degeneration. The results showed that there was no statistically significant correlation between the VBQ score and facet joint degeneration. We thought that the VBQ score is more effective in reflecting the degree of vertebral osteoporosis but is less reliable in assessing the degree of facet joint osteoporosis. As evidenced by the correlation and regression analyses in this study, there is no statistically significant relationship between the VBQ score and facet joint degeneration. Therefore, we concluded that the VBQ score was not currently suitable for evaluating facet joint degeneration. Our findings underscore the importance of assessing ROM‐L when evaluating patients with chronic low back pain [47]. This parameter can serve as a key indicator of facet joint degeneration, aiding in the development of tailored treatment plans. For instance, patients with significant facet hypertrophy and reduced ROM‐L may benefit from interventions targeting spinal mobility and facet unloading, such as physical therapy or minimally invasive procedures. Conversely, metabolic factors, as indicated by SFTT, should not be overlooked in managing spinal degeneration, with weight management potentially playing a supportive role in mitigating degenerative progression [48]. The Weishaupt grading system, which has been widely adopted as the standard for evaluating facet joint degeneration, offers numerous advantages, including broad clinical applicability and convenience [49]. However, the system also has limitations, such as vague grading criteria and a lack of quantitative measures for assessing the degree of degeneration. These shortcomings restrict the potential of the grading system in clinical practice. In recent years, advancements in the analysis of MRI signal intensity have expanded its applications across various fields, including hand surgery, neurosurgery and obstetrics and gynecology [50, 51, 52]. However, the use of signal intensity changes, particularly grayscale value variations, in the study of facet joint degeneration has not been previously reported. Therefore, in this study, we aim to leverage the changes in grayscale values observed in MRI scans to establish a more quantitative and precise grading system for facet joint degeneration. To achieve this, we incorporated grayscale value curves from different Weishaupt grades and applied the LOESS curve fitting method to generate standard curves based on a relatively large sample size. The results of these standard curves were then interpreted to provide a more precise and quantitative complement to the facet joint degeneration grading system. Grade 1: At this stage, the facet joint maintains a good anatomical structure, with the normal bone at both ends showing high signal intensity and the synovial fluid in the middle joint space. The low signal regions represent the facet joint cartilage, and the fitted standard curve exhibits a “W” shape (Figure 7a). Grade 2: As shown in Figure 7b, the reduction in signal intensity within the joint space is due to decreased synovial fluid secretion as facet joint degeneration progresses, leading to a reduction in synovial fluid volume and signal intensity [53]. Furthermore, the signal intensity in the bone at both ends of the joint is no longer uniform, with regions of lower grayscale values appearing in the facet bone, particularly in the superior facet bone. Grade 3: As shown in Figure 7c, the synovial fluid (high signal) within the facet joint is almost entirely absent, resulting in a distinct low‐grayscale value region. In addition, the grayscale value of the superior facet bone (left side) shows a decreasing trend relative to the inferior facet bone (right side). This is attributed to the distinct anatomical and biomechanical properties of the superior and inferior facet joints. Previous studies have reported that the superior facet joint experiences greater shear forces compared to the inferior facet joint [54, 55]. This is primarily due to the anatomical position of the superior facet joint relative to the inferior facet joint during normal spinal movement, particularly during flexion, extension, rotation and lateral bending. The superior facet joint is positioned more externally and typically has a larger range of motion, which results in it bearing greater shear forces [56]. A cadaveric study also reported that the proportion of osteophyte formation in the superior lumbar facet joints was significantly higher than that in the inferior facet joints (45.8% vs. 18.6%) [57]. Additionally, the superior facet joint often has a more complex joint morphology and surface structure, with joint surfaces that may be more convex or irregular. This complexity increases the likelihood of poor joint surface contact during movement, leading to cartilage damage. In contrast, the inferior facet joint typically has a flatter joint surface, resulting in more evenly distributed loading, which may make it less prone to localized joint wear compared to the superior facet joint. These findings are consistent with the results observed in the standard curves generated from our grayscale value fitting. Grade 4: The progression of degeneration leads to a significant disparity in the degenerative processes of the superior and inferior facet joints. At this stage, the boundaries of structures such as the facet bone and subchondral cartilage become indistinguishable (Figure 7d). In conclusion, by incorporating standard curve generated from LOESS curve fitting into the grading process, we have refined the grading system for facet joint degeneration, providing a more robust and quantitative method for evaluating the severity of degeneration. This approach enhances the precision of the grading system, allowing for a more accurate and consistent assessment of facet joint degeneration in clinical practice. Despite its strengths, this study has several limitations. First, the sample size was relatively small, and future studies with larger cohorts are necessary to validate these findings. Second, while we used standardized imaging protocols, the retrospective nature of our analysis may have introduced selection bias. Additionally, the use of only 1.5T MRI may limit the generalizability of our results to clinical settings utilizing higher‐resolution imaging modalities. In addition, we need to clarify that the grayscale trends provided by the curve fitting in this study represent a descriptive rather than a definitive model; therefore, the presence of substantial heterogeneity in clinical data may affect its generalizability. Conclusion Our study identified some radiographic factors, particularly ISH, FJCSA, ROM‐L and SFTT that are significantly associated with facet joint degeneration. In addition, the study underscores the importance of ROM‐L and FJCSA as significant predictors of facet joint degeneration. Additionally, our analysis revealed that SFTT correlates with facet joint degeneration, although it is not an independent risk factor. By incorporating LOESS curve fitting to generate standard grayscale value curves for different grades of facet joint degeneration, we introduced a more quantitative approach to grading. This enhances the precision and consistency of degeneration assessments, addressing the limitations of the original system. The modified grading system provides a more accurate and reliable tool for clinicians to assess facet joint degeneration. Funding The authors have nothing to report. Conflicts of Interest The authors declare no conflicts of interest.