Utilization of radiomics model derived from lumbar CT images for grading the diagnosis of osteoarthritis in facet joints

Utilization of radiomics model derived from lumbar CT images for grading the diagnosis of osteoarthritis in facet joints | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Utilization of radiomics model derived from lumbar CT images for grading the diagnosis of osteoarthritis in facet joints baisen chen, yuyu sun, jiaming cui, tianqi wu, guanhua xu, zhiming cui This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7540310/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 4 You are reading this latest preprint version Abstract Purpose Develop machine learning models utilizing computed tomography (CT) and the weishaupt grading criteria to assess the degeneration severity of facet joint of osteoarthritis (FJOA). Methods The machine learning model utilizes features extracted from patient Lumbar CT at the First People's Hospital of Nantong. Use 3D Slicer software to perform semi-automatic image segmentation on CT images and extract radiological features from the segmented regions. Preliminary screening of radiomic features extracted by radiomics using t-test and rank sum test with p<0.05 as the standard. Based on the core features selected by Lasso regression, construct random forest (RF), K-nearest neighbor (KNN), and support vector machine (SVM) models. Use receiver operating characteristic (ROC) curves to evaluate the model's performance, considering metrics such as accuracy, recall, precision, F1 score, and area under curve (AUC). Results The radiomics package of 3D Slicer extracted 1037 radiomic features from ROI. The T-test combined with rank sum test preliminarily screened 589 radiomics features with statistical differences. Subsequently, Lasso regression was used to identify 28 core features. Develop machine learning models based on 28 core feature selections of RF, SVM, and KNN. The AUCs of RF model, SVM model and KNN model in the training set were 0.783, 0.803 and 0.693 respectively, and those in the validation set were 0.699, 0.719 and 0.671 respectively. Conclusion The machine learning model utilizing lumbar CT images can effectively assess lumbar facet joint degeneration. Through this model, diseases can be classified and diagnosed, and doctors can develop personalized treatment plans. Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 1 Introduction Lumbar facet joint osteoarthritis (LFJOA), also referred to as lumbar zygapophyseal joint osteoarthritis, is a degenerative condition commonly seen in middle-aged and elderly individuals [ 1 ]. It is a known cause of lower back pain and limited mobility. Studies report that 15–45% of such pain originates from the facet joints [ 2 ]. The pathological hallmarks of LFJOA include cartilage degeneration, subchondral bone sclerosis, and reactive osteophyte formation along joint margins [ 3 , 4 ]. The prevalence of LFJOA has increased in recent years, adversely affecting patients' quality of life and work productivity [ 5 , 6 ]. Consequently, the diagnosis and management of LFJOA have become key focuses in spine-related clinical research. Conventional X-ray remains the first-line modality for lumbar evaluation. However, its two-dimensional projection and the oblique orientation of the facet joints limit sensitivity for detecting facet joint osteoarthritis [ 7 ]. Magnetic resonance imaging (MRI) forms different tissue contrasts to highlight lesions, allowing direct visualization of articular cartilage, synovium, and joint effusion, yet its relatively low spatial resolution at cortical margins often underestimates the severity of facet joint osteoarthritis [ 8 , 9 ]. Multidetector computed tomography (CT) provides sub-millimetre isotropic resolution and clearly depicts osseous and peri-articular changes, making it the most effective technique for assessing facet joint osteoarthritis [ 7 ]. Weishaupt and colleagues created a grading system using CT images, focusing on six morphological characteristics: joint-space narrowing, osteophyte formation, facet hypertrophy, subfacet bone erosions, subchondral cysts, and the vacuum phenomenon, enabling quantitative stratification of disease severity [ 10 , 11 ]. Routine CT reports rarely include this grading information, which can lead to clinically relevant segments being overlooked. This underscores the critical need to develop a practical method for systematic facet joint classification in lumbar spine assessment. Radiomics is a rapidly expanding field that leverages high-throughput, quantitative analysis of standard medical images to generate imaging biomarkers for precision medicine. By converting pixel data into mineable features, radiomics enables accurate predictions of diagnosis, prognosis, and treatment response [ 12 , 13 ]. In one study, researchers extracted thirteen CT-based radiomic features from 92 patients and found statistically significant differences between the temporomandibular joints of osteoarthritis cases and controls. When these features were combined with clinical and molecular variables to create 52 composite biomarkers and modelled with multiple machine-learning algorithms, the leading model attained a receiver operating characteristic (ROC) curve area of 0.87 for diagnosing temporo-mandibular joint osteoarthritis [ 14 ]. Tingrun Cui and colleagues gathered knee MRI scans from 148 participants and created a machine learning model to assess radiomics analysis performance [ 15 ]. The results showed excellent performance in the diagnosis of knee osteoarthritis. These results highlight the strong diagnostic potential of radiomics for osteoarthritic disease. The present study aims to construct a computed-tomography-based machine-learning model for LFJOA. By establishing a robust radiomics pipeline and demonstrating its utility in assisting FJOA classification, we seek to provide a foundation for future automated detection systems and to further validate machine-learning approaches in clinical decision support for spinal degeneration. 2 Materials and methods 2.1 Subject screening This study received approval from the Ethics Review Committee of the First People's Hospital of Nantong City (Approval No.2024KT411). The ethics committee waived the informed consent requirement because the study was retrospective. Inclusion criteria: (1) Participants aged 18 to 85 years who underwent a CT scan upon admission. (2) CT scan was performed upon admission; Exclusion criteria: (1) Patients who have undergone surgery to remove facet joints; (2) Unrecognized region of interest (ROI) by the machine; (3) The image is unclear or of poor quality. (Fig. 1 ) Flowchart for screening subjects 2.2 Image collection and processing The CT examination utilized the Ingenuity core CT system from Philips, Amsterdam, Netherlands. Classifying weishaupt as 0–1 is defined as the non-severe group, and grading it as 2–3 is defined as the severe group [ 16 ]. The 3D Slicer 5.7.0 software ( http://www.slicer.org/ ) is utilized for semi-automatic image segmentation. Experienced spinal surgeons, with more than six years in the field, conduct ROI segmentation using threshold and seed growth modules. We manually excised vertebral attachments, including spinous processes, to eliminate their influence. An example of image processing is illustrated in Fig. 2 . a. Segmentation of image cross-section. b. Segmentation of sagittal slice of image. c. 3D model construction. d. Model construction processing 2.3 Radiomics feature extraction Employ the radiomics module within the 3D slicer to derive the radiomics features from the generated model. Then, 1037 radiomics features were resampled and extracted using radiomics. These features include first-order statistical features, which reflect the symmetry and uniformity of the measured voxels. Morphological characteristics quantify the size and morphological information of ROI. The relationship between gray levels in image voxels is captured by second-order and texture features. Based on the features of the transformation, including Gaussian Laplace operator (LoG) and wavelet transform, it is utilized to analyze the gray patterns in different spaces. 2.4 Feature sorting and screening We processed the radiological features to address the mixed information they contain. Initially, we applied z-scores to standardize the pretreatment features, where the formula is X-score=(x − µ)/σ, with µ representing the mean and σ the standard deviation. Conduct a t-test to detect features with p-values below 0.05 between two normally distributed samples. Afterward, apply rank sum test to select p < 0.05 features between two individual samples again. Ultimately, 589 radiomic features with significant difference were acquired. Subsequently, dimensionality reduction was performed using a least absolute shrinkage and selection operator (lasso) with a value range of (-2,1), resulting in 28 features being kept for model construction. 2.5 Machine learning model construction We constructed the model using three classifiers: random forest (RF), support vector machine (SVM), and k-nearest neighbor (KNN). Using grid search, the optimal value is determined by investigating all points in the search range. Five-fold cross-validation evaluates the accuracy of the model. The training model is constructed using radiomics features, with a training to test set ratio of 7:3. The model's FJOA recognition was assessed in both the training and testing datasets. 2.6 Statistical processing Python is utilized for analyzing data and constructing machine learning models. The t-test and rank sum test help identify differentiated features with variations. Lasso regression was employed for dimensionality reduction and feature selection, with p less than 0.05 indicating a significant difference. 3 Results 3.1 Subjects’ baseline characteristics Out of the 164 patients, there were 73 males and 91 females. The average age of patients was 55.83 years, ranging from 18 to 84 years, and their average BMI was 25.11 kg/m 2 . Table 1 summarizes the baseline characteristics. Table 1 Patient's baseline data sheet Variables Gender Male 73(44.5%) Female 91(55.5%) Age(years) 55.83 ± 15.61 BMI 25.11 ± 3.42 (BMI: body mass index) 3.2 Feature selecting Lumbar CT scans yielded 1037 radiomics features, with 28 being selected through lasso regression. Five-fold cross-validation was conducted to identify the optimal lambda for lasso regression tuning (refer to Fig. 3 and Fig. 4 ). Table 2 outlines the 28 features selected by lasso regression screening. A 10-fold cross-validation was used to select the best parameter of Lambda Table 2 Summary of lasso features Image type Feature class feature name Lasso coefficient (λ) original shape LeastAxisLength 0.000085 original shape Maximum2DDiameterSlice -0.038866 original firstorder Skewness -0.014232 original glcm ClusterShade -0.022285 original gldm DependenceNonUniformity 0.006221 log-sigma-3-0-mm-3D glszm SizeZoneNonUniformity 0.030452 wavelet-LLH firstorde r90Percentile 0.029003 wavelet-LHL firstorder Mean -0.038178 wavelet-LHL firstorder Median -0.000626 wavelet-LHL glcm Imc2 -0.009418 wavelet-LHL glcm JointAverage -0.021084 wavelet-LHL glcm MCC -0.000721 wavelet-LHL glcm SumAverage -0.004339 wavelet-LHH firstorder Kurtosis -0.007212 wavelet-LHH gldm SmallDependenceEmphasis 0.022789 wavelet-LHH glrlm LongRunLowGrayLevelEmphasis -0.006754 wavelet-LHH glszm SmallAreaEmphasis 0.000287 wavelet-HLL firstorder Mean -0.022637 wavelet-HLL glcm MCC -0.006635 wavelet-HLH glszm GrayLevelNonUniformityNormalized -0.025499 wavelet-HLH glszm LargeAreaHighGrayLevelEmphasis 0.015703 wavelet-HHL firstorder 90Percentile -0.028532 wavelet-HHL glcm MCC -0.078903 wavelet-HHL glszm ZoneEntropy -0.024728 wavelet-HHL ngtdm Busyness 0.002297 wavelet-HHH firstorder Median 0.013184 wavelet-HHH glcm DifferenceVariance -0.002476 wavelet-HHH gldm SmallDependenceLowGrayLevelEmphasis 0.002529 3.3 Model Establishment and Evaluation Figures 5 and 6 illustrate the diagnostic performance evaluation results of the three models in terms of key feature recognition, respectively. The RF model's subject working characteristic curve shows area under curve (AUC) of 0.783 for the training set (95% confidence interval (CI): 0.783 to 0.821), and AUC of 0.699 for the testing set, 95% CI: 0.625 to 0.699). SVM also demonstrates fairly strong diagnostic performance. The AUC for the training set is 0.803, with 95% CI ranging from 0.800 to 0.816, whereas the AUC for the test set is 0.719, with 95% CI from 0.614 to 0.719. In the training set, the AUC for KNN is 0.693 (ranging from 0.693 to 0.727), while in the test set, it is 0.671 (ranging from 0.591 to 0.671). Table 3 provides a summary of the prediction model's parameters. Table 3 Model diagnosis performance parameters Figure Training cohort Test cohort Random forest AUC(95%CI) 0.783(0.783,0.821) 0.699(0.625,0.699) Precision 0.761 0.670 Recall 0.783 0.699 Accuracy 0.809 0.756 F1 score 0.770 0.680 Support vector machine AUC(95%CI) 0.803(0.800,0.816) 0.719(0.614,0.719) Precision 0.890 0.738 Recall 0.803 0.719 Accuracy 0.881 0.817 F1 score 0.833 0.728 K-nearest neighbor AUC(95%CI) 0.693(0.693,0.727) 0.671(0.591,0.671) Precision 0.776 0.690 Recall 0.693 0.671 Accuracy 0.805 0.787 F1 score 0.715 0.679 ROC analysis of test set, comparison of results of random forest, support vector machine and K nearest neighbor Discussion This research involved the collection and processing of lumbar CT images, performed ROI segmentation, extracted radiomic features, and constructed three machine-learning models for diagnosing FJOA. All models demonstrated promising diagnostic performance in both training set and validation set, highlighting the substantial potential of CT-based radiomic features in assessing FJOA severity. The developed radiomics-based model may thus serve as a valuable clinical tool for screening and stratifying patients with FJOA. Patients identified by the model as having a high risk of FJOA are recommended for specialist spinal consultation and advanced diagnostic evaluation. Machine-learning approaches excel at identifying subtle imaging features beyond human perception [ 17 ]. In this context, LoG filtering was employed to enhance image characteristics, thereby preserving critical radiomic signatures [ 18 ]. Finally, 28 features were left, including six first-order features, one second-order feature, and twenty-one high-order features. Wavelet domain images, altered using high (H) or low (L) filters in each of the three dimensions of CT images, encapsulate advanced features [ 19 , 20 ]. The filtering process will generate a total of 8 wavelet filtered images: Wavelet LHL, Wavelet LHH, Wavelet HHH, Wavelet LHL, Wavelet HHL, and Wavelet LLL [ 21 ]. In our study, the wavelet family with the most representative differences is Wavelet LHL (retaining 6 differential features). The results from these models underscore the substantial potential of machine learning in clinical imaging, although further improvements in advanced image processing techniques may yield additional predictive value. Clinically, the treatment of FJOA typically follows a hierarchical management strategy comprising conservative pharmacological therapy, physical rehabilitation, image-guided intra-articular injections, radiofrequency ablation, and, in advanced cases, surgical interventions such as spinal fusion [ 22 ]. Recent international studies have validated the efficacy of percutaneous radiofrequency ablation combined with spinal stabilization exercises in reducing pain and improving functional outcomes [ 23 ]. Similarly, domestic multicenter retrospective analyses confirm these findings [ 24 ], supporting the need for early and accurate identification of FJOA severity to guide appropriate interventions. Integrating radiomics-derived predictive tools into current diagnostic workflows could facilitate personalized management strategies and streamline patient care pathways. Several limitations in this study still need to be enhanced. Initially, the dataset was limited to 164 patients from one institution, which might restrict the applicability of our findings; future validation should involve larger, multi-center cohorts. Second, although the validation cohort achieved an AUC of approximately 0.8, the accuracy remains insufficient to replace standard clinical diagnosis fully, emphasizing the need for model refinement and integration with other imaging modalities. Ultimately, we created and confirmed a CT-based radiomics model that can assess the severity of lumbar FJOA. As an adjunct diagnostic tool, radiomics provides clinically relevant insights, facilitating targeted patient management and supporting precision treatment strategies. Declarations Author Contribution BSC, JMC and ZMC contributed to the study design, participated in the review process and prepared the manuscript. YYS and JMC contributed to collecting the relevant literature, generating figures and critical interpretation. TQW and GHX processed and analyzed data. BSC, ZMC and GHX conceived the paper and modified the manuscript. All authors read and approved the final manuscript. References Pang H, Chen S, Klyne DM, Harrich D, Ding W, Yang S, Han FY (2023) Low back pain and osteoarthritis pain: a perspective of estrogen. Bone Res 11(1):42 Perolat R, Kastler A, Nicot B, Pellat JM, Tahon F, Attye A, Heck O, Boubagra K, Grand S, Krainik A (2018) Facet joint syndrome: from diagnosis to interventional management. Insights Imaging 9(5):773–789 Wang J, Yang Z, He X, Wang Y, Luo D, Xu W, Zhang H, Zhou X (2024) DNM3OS/miR-127-5p/CDH11, activates Wnt3a/beta-catenin/LEF-1 pathway to form a positive feedback and aggravate spine facet joint osteoarthritis. Noncoding RNA Res 9(2):294–306 Wang J, Lu Q, Mackay MJ, Liu X, Feng Y, Burton DC, Asher MA (2022) Spontaneous Facet Joint Osteoarthritis in NFAT1-Mutant Mice: Age-Dependent Histopathologic Characteristics and Molecular Mechanisms. J Bone Joint Surg Am 104(10):928–940 Breivik H, Collett B, Ventafridda V, Cohen R, Gallacher D (2006) Survey of chronic pain in Europe: prevalence, impact on daily life, and treatment. Eur J Pain 10(4):287–333 Jiang J, Zhang J, Wu C, Guo X, Chen C, Bao G, Sun Y, Chen J, Xue P, Xu G, Cui Z (2018) Up-regulation of TRAF2 inhibits chondrocytes apoptosis in lumbar facet joint osteoarthritis. Biochem Biophys Res Commun 503(3):1659–1665 Kwee RM, Kwee TC (2021) Imaging of facet joint diseases. Clin Imaging 80:167–179 Chen C, Bao GF, Xu G, Sun Y, Cui ZM (2018) Altered Wnt and NF-kappaB Signaling in Facet Joint Osteoarthritis: Insights from RNA Deep Sequencing. Tohoku J Exp Med 245(1):69–77 Lakadamyali H, Tarhan NC, Ergun T, Cakir B, Agildere AM (2008) STIR sequence for depiction of degenerative changes in posterior stabilizing elements in patients with lower back pain. AJR Am J Roentgenol 191(4):973–979 Chen C, Cui S, Li W, Jin H, Fan J, Sun Y, Cui Z (2020) Ingenuity pathway analysis of human facet joint tissues: Insight into facet joint osteoarthritis. Exp Ther Med 19(4):2997–3008 Wang A, Wang T, Zang L, Yuan S, Fan N, Du P, Wu Q (2022) Quantitative Radiological Characteristics of the Facet Joints in Patients with Lumbar Foraminal Stenosis. J Pain Res 15:2363–2371 Gillies RJ, Kinahan PE, Hricak H (2016) Radiomics: Images Are More than Pictures, They Are Data. Radiology 278(2):563–577 Klontzas ME, Manikis GC, Nikiforaki K, Vassalou EE, Spanakis K, Stathis I, Kakkos GA, Matthaiou N, Zibis AH, Marias K, Karantanas AH (2021) Radiomics and Machine Learning Can Differentiate Transient Osteoporosis from Avascular Necrosis of the Hip. Diagnostics (Basel) 11(9) Bianchi J, Goncalves JR, de Oliveira Ruellas AC, Ashman LM, Vimort JB, Yatabe M, Paniagua B, Hernandez P, Benavides E, Soki FN, Ioshida M, Cevidanes LHS (2021) Quantitative bone imaging biomarkers to diagnose temporomandibular joint osteoarthritis. Int J Oral Maxillofac Surg 50(2):227–235 Cui T, Liu R, Jing Y, Fu J, Chen J (2023) Development of machine learning models aiming at knee osteoarthritis diagnosing: an MRI radiomics analysis. J Orthop Surg Res 18(1):375 Weishaupt D, Zanetti M, Boos N, Hodler J (1999) MR imaging and CT in osteoarthritis of the lumbar facet joints. Skeletal Radiol 28(4):215–219 Hirvasniemi J, Klein S, Bierma-Zeinstra S, Vernooij MW, Schiphof D, Oei EHG (2021) A machine learning approach to distinguish between knees without and with osteoarthritis using MRI-based radiomic features from tibial bone. Eur Radiol 31(11):8513–8521 Tang VH, Duong STM, Nguyen CDT, Huynh TM, Duc VT, Phan C, Le H, Bui T, Truong SQH (2023) Wavelet radiomics features from multiphase CT images for screening hepatocellular carcinoma: analysis and comparison. Sci Rep 13(1):19559 Li MD, Cheng MQ, Chen LD, Hu HT, Zhang JC, Ruan SM, Huang H, Kuang M, Lu MD, Li W, Wang W (2022) Reproducibility of radiomics features from ultrasound images: influence of image acquisition and processing. Eur Radiol 32(9):5843–5851 Saeedi E, Dezhkam A, Beigi J, Rastegar S, Yousefi Z, Mehdipour LA, Abdollahi H, Tanha K (2019) Radiomic Feature Robustness and Reproducibility in Quantitative Bone Radiography: A Study on Radiologic Parameter Changes. J Clin Densitom 22(2):203–213 Zheng M, Zhu G, Chen D, Xiao Q, Lei T, Ye C, Pan C, Miao S, Ye L (2023) T1-weighted images-based radiomics for structural lesions evaluation in patients with suspected axial spondyloarthritis. Radiol Med 128(11):1398–1406 Kwak DG, Kwak SG, Lee AY, Chang MC (2019) Outcome of intra-articular lumbar facet joint corticosteroid injection according to the severity of facet joint arthritis. Exp Ther Med 18(5):4132–4136 Sancar KS, Seref C, Hamit G, Yavuz AE, Saziye S (2025) Does the severity of facet joint osteoarthritis affect facet medial branch radiofrequency thermocoagulation results? Neurosciences (Riyadh) 30(2):144–149 Anaya JEC, Coelho SRN, Taneja AK, Cardoso FN, Skaf AY, Aihara AY (2021) Differential Diagnosis of Facet Joint Disorders. Radiographics 41(2):543–558 Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviewers invited by journal 03 Nov, 2025 Editor assigned by journal 09 Sep, 2025 Submission checks completed at journal 08 Sep, 2025 First submitted to journal 04 Sep, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7540310","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":539008716,"identity":"55284673-a465-4ef6-b8e7-9bbd2e353677","order_by":0,"name":"baisen chen","email":"","orcid":"","institution":"Nantong City No 1 People's Hospital and Second Affiliated Hospital of Nantong University","correspondingAuthor":false,"prefix":"","firstName":"baisen","middleName":"","lastName":"chen","suffix":""},{"id":539008724,"identity":"3a961b7e-a689-4bbe-ae80-1a403bc93170","order_by":1,"name":"yuyu sun","email":"","orcid":"","institution":"Nantong Third People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"yuyu","middleName":"","lastName":"sun","suffix":""},{"id":539008725,"identity":"93b5ac4e-8b3a-4a5c-8c99-edb5856880d6","order_by":2,"name":"jiaming cui","email":"","orcid":"","institution":"Shenzhen University","correspondingAuthor":false,"prefix":"","firstName":"jiaming","middleName":"","lastName":"cui","suffix":""},{"id":539008726,"identity":"1c6296d4-02db-4f02-bfae-87902c024939","order_by":3,"name":"tianqi wu","email":"","orcid":"","institution":"Nantong University School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"tianqi","middleName":"","lastName":"wu","suffix":""},{"id":539008727,"identity":"02ee46d6-45e4-4a5e-8025-86792408d537","order_by":4,"name":"guanhua xu","email":"","orcid":"","institution":"Nantong City No 1 People's Hospital and Second Affiliated Hospital of Nantong University","correspondingAuthor":false,"prefix":"","firstName":"guanhua","middleName":"","lastName":"xu","suffix":""},{"id":539008728,"identity":"185dcad8-50e3-476e-a05b-3ba05600b949","order_by":5,"name":"zhiming cui","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA0ElEQVRIiWNgGAWjYNACAwYGfgiLmQQtkg3MJGkB6TpArBb+/tPJnwsK7sgZ3+4/JsFQYZ3YwH72AF4tEgfObpOeYfDM2OzOYTYJhjPpiQ08eQn43cPYu42Zx+Bw4rYbyWwSjG2HExskeAzwa2Hm3fwZqKV+8wyQln/EaGHj3SAN1JJgIAHS0kCEFokzvCC/HDaccSPZ2CLhWLpxG08Ofi38/Wc3fy74c1ief0biwxsfaqxl+9nP4NcCAoi4SABiNoLqGUiL8VEwCkbBKBiJAAAuAj9DrrI2bwAAAABJRU5ErkJggg==","orcid":"","institution":"Nantong City No 1 People's Hospital and Second Affiliated Hospital of Nantong University","correspondingAuthor":true,"prefix":"","firstName":"zhiming","middleName":"","lastName":"cui","suffix":""}],"badges":[],"createdAt":"2025-09-05 03:23:19","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7540310/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7540310/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":95807594,"identity":"05bb2818-fcef-46ff-ac73-4e4cea430930","added_by":"auto","created_at":"2025-11-13 08:48:59","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":817015,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/34acd649017ac298129d0679.docx"},{"id":95808104,"identity":"d1d12c60-3320-4f33-ae3c-f009eec8832c","added_by":"auto","created_at":"2025-11-13 08:49:19","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7475,"visible":true,"origin":"","legend":"","description":"","filename":"a7ed1fbae4bd4a518fd7cf549a99f7d0.json","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/3d227a1a3027c2c39a53f75f.json"},{"id":95807568,"identity":"f8a14ef5-bd9a-4986-8022-8928dbda85ea","added_by":"auto","created_at":"2025-11-13 08:48:56","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":75808,"visible":true,"origin":"","legend":"","description":"","filename":"a7ed1fbae4bd4a518fd7cf549a99f7d01enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/174f2fd13d93120d7fe771e6.xml"},{"id":95807874,"identity":"d6662507-f764-4e39-b73b-ea86fd74a840","added_by":"auto","created_at":"2025-11-13 08:49:13","extension":"png","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":56711,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/09a960a1d363e78c850bcbe9.png"},{"id":95807608,"identity":"9362f734-4ef6-440b-8c81-6f9ff013be55","added_by":"auto","created_at":"2025-11-13 08:49:01","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":380550,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/40ff2bf009ff409dd43598e5.png"},{"id":95807534,"identity":"33d0a655-357d-4abf-b4bb-9d609c03840b","added_by":"auto","created_at":"2025-11-13 08:48:53","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":92083,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/0b10b3049f1471a47cbf54ee.png"},{"id":95807754,"identity":"5cb2f7c4-1421-4dc1-853e-49cc22099558","added_by":"auto","created_at":"2025-11-13 08:49:07","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":68629,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/5c0a829bdf71d004758ddcb5.png"},{"id":95807956,"identity":"2b1a0fa7-2da7-4c79-b8a5-775449c53de0","added_by":"auto","created_at":"2025-11-13 08:49:16","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":84191,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/8d648f83c9c0c8c181c9d675.png"},{"id":95807656,"identity":"67dbaf21-5e2a-4b77-ba0f-5e3ac446d1ae","added_by":"auto","created_at":"2025-11-13 08:49:03","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":88793,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/f45893eed041cf2a261b2f9b.png"},{"id":95807619,"identity":"0a59c0d9-67ce-4424-81da-0d8e29f6bfe1","added_by":"auto","created_at":"2025-11-13 08:49:02","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":17611,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/7ce568092aead21719439ab9.png"},{"id":95807589,"identity":"a1378877-eb89-4e22-b317-e7d1b621080b","added_by":"auto","created_at":"2025-11-13 08:48:58","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":44058,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/b4f6aa0fe3e674f58228f068.png"},{"id":95807625,"identity":"0c8a83ab-95b3-494e-af72-63c84aa56a50","added_by":"auto","created_at":"2025-11-13 08:49:02","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19331,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/8657900016d3180681c98389.png"},{"id":95807707,"identity":"2f14cb7c-4be4-4eba-909e-9f171ad6536e","added_by":"auto","created_at":"2025-11-13 08:49:06","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":15042,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/0eb17a3d583a62a668d92d23.png"},{"id":95807668,"identity":"45a68a73-16e3-46ab-9d68-d92538740975","added_by":"auto","created_at":"2025-11-13 08:49:03","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":17928,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/aa589b981d8db97dfb9333c0.png"},{"id":95807716,"identity":"ebadbfb3-f978-44a4-ae38-b51b79b876ba","added_by":"auto","created_at":"2025-11-13 08:49:06","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19188,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/2cf86c05752a33bdf82e307f.png"},{"id":95807613,"identity":"da84a4b5-394d-4136-9e4b-625ceedd01d3","added_by":"auto","created_at":"2025-11-13 08:49:02","extension":"xml","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":73380,"visible":true,"origin":"","legend":"","description":"","filename":"a7ed1fbae4bd4a518fd7cf549a99f7d01structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/9f5ba87cb52e73ef348d2307.xml"},{"id":95807661,"identity":"f79c4700-72b9-433b-bfa9-56cf9f8cb8ca","added_by":"auto","created_at":"2025-11-13 08:49:03","extension":"html","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":79523,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/4449718130201e35b187e56f.html"},{"id":95807696,"identity":"f71218cc-5b65-4e01-bcdc-634253493816","added_by":"auto","created_at":"2025-11-13 08:49:05","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":72670,"visible":true,"origin":"","legend":"\u003cp\u003eFlow diagram\u003c/p\u003e\n\u003cp\u003eFlowchart for screening subjects\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/675d79ff77066f4d935c919e.png"},{"id":95808038,"identity":"6ac78162-3bc1-43fa-9f62-3eafee6a4e53","added_by":"auto","created_at":"2025-11-13 08:49:17","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":98622,"visible":true,"origin":"","legend":"\u003cp\u003eImage cutting\u003c/p\u003e\n\u003cp\u003ea. Segmentation of image cross-section. b. Segmentation of sagittal slice of image. c. 3D model construction. d. Model construction processing\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/77a5445602caca9930bd51dd.png"},{"id":95808109,"identity":"17a06695-c394-4c51-a1c5-4a8e6acf1df9","added_by":"auto","created_at":"2025-11-13 08:49:19","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":58472,"visible":true,"origin":"","legend":"\u003cp\u003eTwenty-eight radiomic features coefficient profile versus the Lambda sequence\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/b5252efffbfe427470cb0b12.png"},{"id":95807864,"identity":"83ec02ca-4d26-4cdf-9569-2cebc3eec898","added_by":"auto","created_at":"2025-11-13 08:49:13","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":86729,"visible":true,"origin":"","legend":"\u003cp\u003eA 10-fold cross-validation was used to select the best parameter of Lambda\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/8433cb160716e516d3b2d5a1.png"},{"id":95807586,"identity":"7c21824a-9440-4780-94b2-6c30b8e7b1b3","added_by":"auto","created_at":"2025-11-13 08:48:58","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":86369,"visible":true,"origin":"","legend":"\u003cp\u003eROC analysis of training set, comparison of results of random forest, support vector machine and K nearest neighbor\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/d6c6e2948c14bbd349215226.png"},{"id":95807792,"identity":"ccfe7f94-a8d7-4e38-a076-b2ac6f5dc041","added_by":"auto","created_at":"2025-11-13 08:49:10","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":90100,"visible":true,"origin":"","legend":"\u003cp\u003eROC analysis of test set, comparison of results of random forest, support vector machine and K nearest neighbor\u003c/p\u003e","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/a03cf1f0897f1d9996ec999f.png"},{"id":95810584,"identity":"92b31c6a-d41e-488c-ac4b-8265cce5176d","added_by":"auto","created_at":"2025-11-13 08:53:05","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1141673,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7540310/v1/653d787b-fa2f-432d-a612-b511ea10f3a3.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Utilization of radiomics model derived from lumbar CT images for grading the diagnosis of osteoarthritis in facet joints","fulltext":[{"header":"1 Introduction","content":"\u003cp\u003eLumbar facet joint osteoarthritis (LFJOA), also referred to as lumbar zygapophyseal joint osteoarthritis, is a degenerative condition commonly seen in middle-aged and elderly individuals [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. It is a known cause of lower back pain and limited mobility. Studies report that 15\u0026ndash;45% of such pain originates from the facet joints [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. The pathological hallmarks of LFJOA include cartilage degeneration, subchondral bone sclerosis, and reactive osteophyte formation along joint margins [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. The prevalence of LFJOA has increased in recent years, adversely affecting patients' quality of life and work productivity [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Consequently, the diagnosis and management of LFJOA have become key focuses in spine-related clinical research.\u003c/p\u003e\u003cp\u003eConventional X-ray remains the first-line modality for lumbar evaluation. However, its two-dimensional projection and the oblique orientation of the facet joints limit sensitivity for detecting facet joint osteoarthritis [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Magnetic resonance imaging (MRI) forms different tissue contrasts to highlight lesions, allowing direct visualization of articular cartilage, synovium, and joint effusion, yet its relatively low spatial resolution at cortical margins often underestimates the severity of facet joint osteoarthritis [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Multidetector computed tomography (CT) provides sub-millimetre isotropic resolution and clearly depicts osseous and peri-articular changes, making it the most effective technique for assessing facet joint osteoarthritis [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Weishaupt and colleagues created a grading system using CT images, focusing on six morphological characteristics: joint-space narrowing, osteophyte formation, facet hypertrophy, subfacet bone erosions, subchondral cysts, and the vacuum phenomenon, enabling quantitative stratification of disease severity [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Routine CT reports rarely include this grading information, which can lead to clinically relevant segments being overlooked. This underscores the critical need to develop a practical method for systematic facet joint classification in lumbar spine assessment.\u003c/p\u003e\u003cp\u003eRadiomics is a rapidly expanding field that leverages high-throughput, quantitative analysis of standard medical images to generate imaging biomarkers for precision medicine. By converting pixel data into mineable features, radiomics enables accurate predictions of diagnosis, prognosis, and treatment response [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. In one study, researchers extracted thirteen CT-based radiomic features from 92 patients and found statistically significant differences between the temporomandibular joints of osteoarthritis cases and controls. When these features were combined with clinical and molecular variables to create 52 composite biomarkers and modelled with multiple machine-learning algorithms, the leading model attained a receiver operating characteristic (ROC) curve area of 0.87 for diagnosing temporo-mandibular joint osteoarthritis [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Tingrun Cui and colleagues gathered knee MRI scans from 148 participants and created a machine learning model to assess radiomics analysis performance [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. The results showed excellent performance in the diagnosis of knee osteoarthritis. These results highlight the strong diagnostic potential of radiomics for osteoarthritic disease.\u003c/p\u003e\u003cp\u003eThe present study aims to construct a computed-tomography-based machine-learning model for LFJOA. By establishing a robust radiomics pipeline and demonstrating its utility in assisting FJOA classification, we seek to provide a foundation for future automated detection systems and to further validate machine-learning approaches in clinical decision support for spinal degeneration.\u003c/p\u003e"},{"header":"2 Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1 Subject screening\u003c/h2\u003e\u003cp\u003e This study received approval from the Ethics Review Committee of the First People's Hospital of Nantong City (Approval No.2024KT411). The ethics committee waived the informed consent requirement because the study was retrospective. Inclusion criteria: (1) Participants aged 18 to 85 years who underwent a CT scan upon admission. (2) CT scan was performed upon admission; Exclusion criteria: (1) Patients who have undergone surgery to remove facet joints; (2) Unrecognized region of interest (ROI) by the machine; (3) The image is unclear or of poor quality. (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e)\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFlowchart for screening subjects\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e2.2 Image collection and processing\u003c/h2\u003e\u003cp\u003eThe CT examination utilized the Ingenuity core CT system from Philips, Amsterdam, Netherlands. Classifying weishaupt as 0\u0026ndash;1 is defined as the non-severe group, and grading it as 2\u0026ndash;3 is defined as the severe group [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eThe 3D Slicer 5.7.0 software (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.slicer.org/\u003c/span\u003e\u003cspan address=\"http://www.slicer.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) is utilized for semi-automatic image segmentation. Experienced spinal surgeons, with more than six years in the field, conduct ROI segmentation using threshold and seed growth modules. We manually excised vertebral attachments, including spinous processes, to eliminate their influence. An example of image processing is illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003ea. Segmentation of image cross-section. b. Segmentation of sagittal slice of image. c. 3D model construction. d. Model construction processing\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e2.3 Radiomics feature extraction\u003c/h2\u003e\u003cp\u003eEmploy the radiomics module within the 3D slicer to derive the radiomics features from the generated model. Then, 1037 radiomics features were resampled and extracted using radiomics. These features include first-order statistical features, which reflect the symmetry and uniformity of the measured voxels. Morphological characteristics quantify the size and morphological information of ROI. The relationship between gray levels in image voxels is captured by second-order and texture features. Based on the features of the transformation, including Gaussian Laplace operator (LoG) and wavelet transform, it is utilized to analyze the gray patterns in different spaces.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e2.4 Feature sorting and screening\u003c/h2\u003e\u003cp\u003eWe processed the radiological features to address the mixed information they contain. Initially, we applied z-scores to standardize the pretreatment features, where the formula is X-score=(x\u0026thinsp;\u0026minus;\u0026thinsp;\u0026micro;)/σ, with \u0026micro; representing the mean and σ the standard deviation. Conduct a t-test to detect features with p-values below 0.05 between two normally distributed samples. Afterward, apply rank sum test to select p\u0026thinsp;\u0026lt;\u0026thinsp;0.05 features between two individual samples again. Ultimately, 589 radiomic features with significant difference were acquired. Subsequently, dimensionality reduction was performed using a least absolute shrinkage and selection operator (lasso) with a value range of (-2,1), resulting in 28 features being kept for model construction.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003e2.5 Machine learning model construction\u003c/h2\u003e\u003cp\u003eWe constructed the model using three classifiers: random forest (RF), support vector machine (SVM), and k-nearest neighbor (KNN). Using grid search, the optimal value is determined by investigating all points in the search range. Five-fold cross-validation evaluates the accuracy of the model. The training model is constructed using radiomics features, with a training to test set ratio of 7:3. The model's FJOA recognition was assessed in both the training and testing datasets.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003e2.6 Statistical processing\u003c/h2\u003e\u003cp\u003ePython is utilized for analyzing data and constructing machine learning models. The t-test and rank sum test help identify differentiated features with variations. Lasso regression was employed for dimensionality reduction and feature selection, with p less than 0.05 indicating a significant difference.\u003c/p\u003e\u003c/div\u003e"},{"header":"3 Results","content":"\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\u003ch2\u003e3.1 Subjects\u0026rsquo; baseline characteristics\u003c/h2\u003e\u003cp\u003eOut of the 164 patients, there were 73 males and 91 females. The average age of patients was 55.83 years, ranging from 18 to 84 years, and their average BMI was 25.11 kg/m\u003csup\u003e2\u003c/sup\u003e. Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e summarizes the baseline characteristics.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003ePatient's baseline data sheet\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"2\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVariables\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGender\u003c/p\u003e\u003cp\u003eMale\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e73(44.5%)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFemale\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e91(55.5%)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAge(years)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e55.83\u0026thinsp;\u0026plusmn;\u0026thinsp;15.61\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBMI\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e25.11\u0026thinsp;\u0026plusmn;\u0026thinsp;3.42\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"2\"\u003e(BMI: body mass index)\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003e3.2 Feature selecting\u003c/h2\u003e\u003cp\u003eLumbar CT scans yielded 1037 radiomics features, with 28 being selected through lasso regression. Five-fold cross-validation was conducted to identify the optimal lambda for lasso regression tuning (refer to Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e and Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e outlines the 28 features selected by lasso regression screening.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eA 10-fold cross-validation was used to select the best parameter of Lambda\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eSummary of lasso features\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eImage type\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eFeature class\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003efeature name\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eLasso coefficient (λ)\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eoriginal\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eLeastAxisLength\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.000085\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eoriginal\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMaximum2DDiameterSlice\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.038866\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eoriginal\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003efirstorder\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSkewness\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.014232\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eoriginal\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eglcm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eClusterShade\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.022285\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eoriginal\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003egldm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eDependenceNonUniformity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.006221\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003elog-sigma-3-0-mm-3D\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eglszm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSizeZoneNonUniformity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.030452\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-LLH\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003efirstorde\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003er90Percentile\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.029003\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-LHL\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003efirstorder\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMean\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.038178\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-LHL\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003efirstorder\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMedian\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.000626\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-LHL\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eglcm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eImc2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.009418\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-LHL\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eglcm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eJointAverage\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.021084\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-LHL\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eglcm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMCC\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.000721\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-LHL\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eglcm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSumAverage\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.004339\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-LHH\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003efirstorder\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eKurtosis\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.007212\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-LHH\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003egldm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSmallDependenceEmphasis\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.022789\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-LHH\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eglrlm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eLongRunLowGrayLevelEmphasis\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.006754\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-LHH\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eglszm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSmallAreaEmphasis\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.000287\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-HLL\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003efirstorder\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMean\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.022637\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-HLL\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eglcm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMCC\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.006635\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-HLH\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eglszm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eGrayLevelNonUniformityNormalized\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.025499\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-HLH\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eglszm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eLargeAreaHighGrayLevelEmphasis\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.015703\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-HHL\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003efirstorder\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e90Percentile\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.028532\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-HHL\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eglcm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMCC\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.078903\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-HHL\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eglszm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eZoneEntropy\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.024728\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-HHL\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003engtdm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eBusyness\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.002297\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-HHH\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003efirstorder\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMedian\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.013184\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-HHH\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eglcm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eDifferenceVariance\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.002476\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ewavelet-HHH\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003egldm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSmallDependenceLowGrayLevelEmphasis\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.002529\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003e3.3 Model Establishment and Evaluation\u003c/h2\u003e\u003cp\u003eFigures \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e and \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e illustrate the diagnostic performance evaluation results of the three models in terms of key feature recognition, respectively. The RF model's subject working characteristic curve shows area under curve (AUC) of 0.783 for the training set (95% confidence interval (CI): 0.783 to 0.821), and AUC of 0.699 for the testing set, 95% CI: 0.625 to 0.699). SVM also demonstrates fairly strong diagnostic performance. The AUC for the training set is 0.803, with 95% CI ranging from 0.800 to 0.816, whereas the AUC for the test set is 0.719, with 95% CI from 0.614 to 0.719. In the training set, the AUC for KNN is 0.693 (ranging from 0.693 to 0.727), while in the test set, it is 0.671 (ranging from 0.591 to 0.671). Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e provides a summary of the prediction model's parameters.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eModel diagnosis performance parameters\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFigure\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTraining cohort\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eTest cohort\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRandom forest\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAUC(95%CI)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.783(0.783,0.821)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.699(0.625,0.699)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePrecision\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.761\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.670\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRecall\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.783\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.699\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAccuracy\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.809\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.756\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eF1 score\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.770\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.680\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSupport vector machine\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAUC(95%CI)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.803(0.800,0.816)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.719(0.614,0.719)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePrecision\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.890\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.738\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRecall\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.803\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.719\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAccuracy\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.881\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.817\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eF1 score\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.833\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.728\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eK-nearest neighbor\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAUC(95%CI)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.693(0.693,0.727)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.671(0.591,0.671)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePrecision\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.776\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.690\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRecall\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.693\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.671\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAccuracy\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.805\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.787\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eF1 score\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.715\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.679\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eROC analysis of test set, comparison of results of random forest, support vector machine and K nearest neighbor\u003c/p\u003e\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis research involved the collection and processing of lumbar CT images, performed ROI segmentation, extracted radiomic features, and constructed three machine-learning models for diagnosing FJOA. All models demonstrated promising diagnostic performance in both training set and validation set, highlighting the substantial potential of CT-based radiomic features in assessing FJOA severity. The developed radiomics-based model may thus serve as a valuable clinical tool for screening and stratifying patients with FJOA. Patients identified by the model as having a high risk of FJOA are recommended for specialist spinal consultation and advanced diagnostic evaluation.\u003c/p\u003e\u003cp\u003eMachine-learning approaches excel at identifying subtle imaging features beyond human perception [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. In this context, LoG filtering was employed to enhance image characteristics, thereby preserving critical radiomic signatures [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Finally, 28 features were left, including six first-order features, one second-order feature, and twenty-one high-order features. Wavelet domain images, altered using high (H) or low (L) filters in each of the three dimensions of CT images, encapsulate advanced features [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. The filtering process will generate a total of 8 wavelet filtered images: Wavelet LHL, Wavelet LHH, Wavelet HHH, Wavelet LHL, Wavelet HHL, and Wavelet LLL [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. In our study, the wavelet family with the most representative differences is Wavelet LHL (retaining 6 differential features). The results from these models underscore the substantial potential of machine learning in clinical imaging, although further improvements in advanced image processing techniques may yield additional predictive value.\u003c/p\u003e\u003cp\u003eClinically, the treatment of FJOA typically follows a hierarchical management strategy comprising conservative pharmacological therapy, physical rehabilitation, image-guided intra-articular injections, radiofrequency ablation, and, in advanced cases, surgical interventions such as spinal fusion [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. Recent international studies have validated the efficacy of percutaneous radiofrequency ablation combined with spinal stabilization exercises in reducing pain and improving functional outcomes [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. Similarly, domestic multicenter retrospective analyses confirm these findings [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e], supporting the need for early and accurate identification of FJOA severity to guide appropriate interventions. Integrating radiomics-derived predictive tools into current diagnostic workflows could facilitate personalized management strategies and streamline patient care pathways.\u003c/p\u003e\u003cp\u003eSeveral limitations in this study still need to be enhanced. Initially, the dataset was limited to 164 patients from one institution, which might restrict the applicability of our findings; future validation should involve larger, multi-center cohorts. Second, although the validation cohort achieved an AUC of approximately 0.8, the accuracy remains insufficient to replace standard clinical diagnosis fully, emphasizing the need for model refinement and integration with other imaging modalities.\u003c/p\u003e\u003cp\u003eUltimately, we created and confirmed a CT-based radiomics model that can assess the severity of lumbar FJOA. As an adjunct diagnostic tool, radiomics provides clinically relevant insights, facilitating targeted patient management and supporting precision treatment strategies.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eBSC, JMC and ZMC contributed to the study design, participated in the review process and prepared the manuscript. YYS and JMC contributed to collecting the relevant literature, generating figures and critical interpretation. TQW and GHX processed and analyzed data. BSC, ZMC and GHX conceived the paper and modified the manuscript. All authors read and approved the final manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003ePang H, Chen S, Klyne DM, Harrich D, Ding W, Yang S, Han FY (2023) Low back pain and osteoarthritis pain: a perspective of estrogen. Bone Res 11(1):42\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePerolat R, Kastler A, Nicot B, Pellat JM, Tahon F, Attye A, Heck O, Boubagra K, Grand S, Krainik A (2018) Facet joint syndrome: from diagnosis to interventional management. Insights Imaging 9(5):773\u0026ndash;789\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang J, Yang Z, He X, Wang Y, Luo D, Xu W, Zhang H, Zhou X (2024) DNM3OS/miR-127-5p/CDH11, activates Wnt3a/beta-catenin/LEF-1 pathway to form a positive feedback and aggravate spine facet joint osteoarthritis. Noncoding RNA Res 9(2):294\u0026ndash;306\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang J, Lu Q, Mackay MJ, Liu X, Feng Y, Burton DC, Asher MA (2022) Spontaneous Facet Joint Osteoarthritis in NFAT1-Mutant Mice: Age-Dependent Histopathologic Characteristics and Molecular Mechanisms. J Bone Joint Surg Am 104(10):928\u0026ndash;940\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBreivik H, Collett B, Ventafridda V, Cohen R, Gallacher D (2006) Survey of chronic pain in Europe: prevalence, impact on daily life, and treatment. Eur J Pain 10(4):287\u0026ndash;333\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJiang J, Zhang J, Wu C, Guo X, Chen C, Bao G, Sun Y, Chen J, Xue P, Xu G, Cui Z (2018) Up-regulation of TRAF2 inhibits chondrocytes apoptosis in lumbar facet joint osteoarthritis. Biochem Biophys Res Commun 503(3):1659\u0026ndash;1665\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKwee RM, Kwee TC (2021) Imaging of facet joint diseases. Clin Imaging 80:167\u0026ndash;179\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChen C, Bao GF, Xu G, Sun Y, Cui ZM (2018) Altered Wnt and NF-kappaB Signaling in Facet Joint Osteoarthritis: Insights from RNA Deep Sequencing. Tohoku J Exp Med 245(1):69\u0026ndash;77\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLakadamyali H, Tarhan NC, Ergun T, Cakir B, Agildere AM (2008) STIR sequence for depiction of degenerative changes in posterior stabilizing elements in patients with lower back pain. AJR Am J Roentgenol 191(4):973\u0026ndash;979\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChen C, Cui S, Li W, Jin H, Fan J, Sun Y, Cui Z (2020) Ingenuity pathway analysis of human facet joint tissues: Insight into facet joint osteoarthritis. Exp Ther Med 19(4):2997\u0026ndash;3008\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang A, Wang T, Zang L, Yuan S, Fan N, Du P, Wu Q (2022) Quantitative Radiological Characteristics of the Facet Joints in Patients with Lumbar Foraminal Stenosis. J Pain Res 15:2363\u0026ndash;2371\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGillies RJ, Kinahan PE, Hricak H (2016) Radiomics: Images Are More than Pictures, They Are Data. Radiology 278(2):563\u0026ndash;577\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKlontzas ME, Manikis GC, Nikiforaki K, Vassalou EE, Spanakis K, Stathis I, Kakkos GA, Matthaiou N, Zibis AH, Marias K, Karantanas AH (2021) Radiomics and Machine Learning Can Differentiate Transient Osteoporosis from Avascular Necrosis of the Hip. Diagnostics (Basel) 11(9)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBianchi J, Goncalves JR, de Oliveira Ruellas AC, Ashman LM, Vimort JB, Yatabe M, Paniagua B, Hernandez P, Benavides E, Soki FN, Ioshida M, Cevidanes LHS (2021) Quantitative bone imaging biomarkers to diagnose temporomandibular joint osteoarthritis. Int J Oral Maxillofac Surg 50(2):227\u0026ndash;235\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCui T, Liu R, Jing Y, Fu J, Chen J (2023) Development of machine learning models aiming at knee osteoarthritis diagnosing: an MRI radiomics analysis. J Orthop Surg Res 18(1):375\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWeishaupt D, Zanetti M, Boos N, Hodler J (1999) MR imaging and CT in osteoarthritis of the lumbar facet joints. Skeletal Radiol 28(4):215\u0026ndash;219\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHirvasniemi J, Klein S, Bierma-Zeinstra S, Vernooij MW, Schiphof D, Oei EHG (2021) A machine learning approach to distinguish between knees without and with osteoarthritis using MRI-based radiomic features from tibial bone. Eur Radiol 31(11):8513\u0026ndash;8521\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTang VH, Duong STM, Nguyen CDT, Huynh TM, Duc VT, Phan C, Le H, Bui T, Truong SQH (2023) Wavelet radiomics features from multiphase CT images for screening hepatocellular carcinoma: analysis and comparison. Sci Rep 13(1):19559\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLi MD, Cheng MQ, Chen LD, Hu HT, Zhang JC, Ruan SM, Huang H, Kuang M, Lu MD, Li W, Wang W (2022) Reproducibility of radiomics features from ultrasound images: influence of image acquisition and processing. Eur Radiol 32(9):5843\u0026ndash;5851\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSaeedi E, Dezhkam A, Beigi J, Rastegar S, Yousefi Z, Mehdipour LA, Abdollahi H, Tanha K (2019) Radiomic Feature Robustness and Reproducibility in Quantitative Bone Radiography: A Study on Radiologic Parameter Changes. J Clin Densitom 22(2):203\u0026ndash;213\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZheng M, Zhu G, Chen D, Xiao Q, Lei T, Ye C, Pan C, Miao S, Ye L (2023) T1-weighted images-based radiomics for structural lesions evaluation in patients with suspected axial spondyloarthritis. Radiol Med 128(11):1398\u0026ndash;1406\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKwak DG, Kwak SG, Lee AY, Chang MC (2019) Outcome of intra-articular lumbar facet joint corticosteroid injection according to the severity of facet joint arthritis. Exp Ther Med 18(5):4132\u0026ndash;4136\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSancar KS, Seref C, Hamit G, Yavuz AE, Saziye S (2025) Does the severity of facet joint osteoarthritis affect facet medial branch radiofrequency thermocoagulation results? Neurosciences (Riyadh) 30(2):144\u0026ndash;149\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAnaya JEC, Coelho SRN, Taneja AK, Cardoso FN, Skaf AY, Aihara AY (2021) Differential Diagnosis of Facet Joint Disorders. Radiographics 41(2):543\u0026ndash;558\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"european-spine-journal","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"esjo","sideBox":"Learn more about [European Spine Journal](http://link.springer.com/journal/586)","snPcode":"586","submissionUrl":"https://submission.springernature.com/new-submission/586/3","title":"European Spine Journal","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7540310/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7540310/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003ePurpose Develop machine learning models utilizing computed tomography (CT) and the weishaupt grading criteria to assess the degeneration severity of facet joint of osteoarthritis (FJOA).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eMethods The machine learning model utilizes features extracted from patient Lumbar CT at the First People's Hospital of Nantong. Use 3D Slicer software to perform semi-automatic image segmentation on CT images and extract radiological features from the segmented regions. Preliminary screening of radiomic features extracted by radiomics using t-test and rank sum test with p\u0026lt;0.05 as the standard. Based on the core features selected by Lasso regression, construct random forest (RF), K-nearest neighbor (KNN), and support vector machine (SVM) models. Use receiver operating characteristic (ROC) curves to evaluate the model's performance, considering metrics such as accuracy, recall, precision, F1 score, and area under curve (AUC).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eResults The radiomics package of 3D Slicer extracted 1037 radiomic features from ROI. The T-test combined with rank sum test preliminarily screened 589 radiomics features with statistical differences. Subsequently, Lasso regression was used to identify 28 core features. Develop machine learning models based on 28 core feature selections of RF, SVM, and KNN. The AUCs of RF model, SVM model and KNN model in the training set were 0.783, 0.803 and 0.693 respectively, and those in the validation set were 0.699, 0.719 and 0.671 respectively.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eConclusion The machine learning model utilizing lumbar CT images can effectively assess lumbar facet joint degeneration. Through this model, diseases can be classified and diagnosed, and doctors can develop personalized treatment plans.\u003c/p\u003e","manuscriptTitle":"Utilization of radiomics model derived from lumbar CT images for grading the diagnosis of osteoarthritis in facet joints","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-13 08:15:49","doi":"10.21203/rs.3.rs-7540310/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewersInvited","content":"","date":"2025-11-03T07:58:36+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-09-09T21:16:22+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-09-08T05:58:27+00:00","index":"","fulltext":""},{"type":"submitted","content":"European Spine Journal","date":"2025-09-05T03:19:13+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"european-spine-journal","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"esjo","sideBox":"Learn more about [European Spine Journal](http://link.springer.com/journal/586)","snPcode":"586","submissionUrl":"https://submission.springernature.com/new-submission/586/3","title":"European Spine Journal","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"9b0a9fce-f61e-4bd1-9166-078e253415d2","owner":[],"postedDate":"November 13th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-11-13T08:15:49+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-13 08:15:49","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7540310","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7540310","identity":"rs-7540310","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}