Deep learning for automated, interpretable classification of lumbar spinal stenosis and facet arthropathy from axial MRI

Artificial intelligence in musculoskeletal imaging: realistic clinical applications in the next decade

This article will provide a perspective review of the most extensively investigated deep learning (DL) applications for musculoskeletal disease detection that have the best potential to translate into routine clinical practice over the next decade. Deep learning methods for detecting fractures, estimating pediatric bone age, calculating bone measurements such as lower extremity alignment and Cobb angle, and grading osteoarthritis on radiographs have been shown to have high diagnostic performance with many of these applications now commercially available for use in clinical practice. Many studies have also documented the feasibility of using DL methods for detecting joint pathology and characterizing bone tumors on magnetic resonance imaging (MRI). However, musculoskeletal disease detection on MRI is difficult as it requires multi-task, multi-class detection of complex abnormalities on multiple image slices with different tissue contrasts. The generalizability of DL methods for musculoskeletal disease detection on MRI is also challenging due to fluctuations in image quality caused by the wide variety of scanners and pulse sequences used in routine MRI protocols. The diagnostic performance of current DL methods for musculoskeletal disease detection must be further evaluated in well-designed prospective studies using large image datasets acquired at different institutions with different imaging parameters and imaging hardware before they can be fully implemented in clinical practice. Future studies must also investigate the true clinical benefits of current DL methods and determine whether they could enhance quality, reduce error rates, improve workflow, and decrease radiologist fatigue and burnout with all of this weighed against the costs.

Development of a Deep-Learning Model for Diagnosing Lumbar Spinal Stenosis Based on CT Images

STUDY DESIGN

Retrospective study.

OBJECTIVES

This study aimed to develop an initial deep-learning (DL) model based on computerized tomography (CT) scans for diagnosing lumbar spinal stenosis.

SUMMARY OF BACKGROUND DATA

Magnetic resonance imaging is commonly used for diagnosing lumbar spinal stenosis due to its high soft tissue resolution, but CT is more portable, cost-effective, and has wider regional coverage. Using DL models to improve the accuracy of CT diagnosis can effectively reduce missed diagnoses and misdiagnoses in clinical practice.

MATERIALS AND METHODS

Axial lumbar spine CT scans obtained between March 2022 and September 2023 were included. The data set was divided into a training set (62.3%), a validation set (22.9%), and a control set (14.8%). All data were labeled by two spine surgeons using the widely accepted grading system for lumbar spinal stenosis. The training and validation sets were used to annotate the regions of interest by the two spine surgeons. First, a region of interest detection model and a convolutional neural network classifier were trained using the training set. After training, the model was preliminarily evaluated using a validation set. Finally, the performance of the DL model was evaluated on the control set, and a comparison was made between the model and the classification performance of specialists with varying levels of experience.

RESULTS

The central stenosis grading accuracies of DL Model Version 1 and DL Model Version 2 were 88% and 83%, respectively. The lateral recess grading accuracies of DL Model Version 1 and DL Model Version 2 were 75% and 71%, respectively.

CONCLUSIONS

Our preliminarily developed DL system for assessing the degree of lumbar spinal stenosis in CT, including the central canal and lateral recess, has shown similar accuracy to experienced specialist physicians. This holds great value for further development and clinical application.

Artificial Intelligence in Spinal Imaging and Patient Care: A Review of Recent Advances

Artificial intelligence (AI) is transforming spinal imaging and patient care through automated analysis and enhanced decision-making. This review presents a clinical task-based evaluation, highlighting the specific impact of AI techniques on different aspects of spinal imaging and patient care. We first discuss how AI can potentially improve image quality through techniques like denoising or artifact reduction. We then explore how AI enables efficient quantification of anatomical measurements, spinal curvature parameters, vertebral segmentation, and disc grading. This facilitates objective, accurate interpretation and diagnosis. AI models now reliably detect key spinal pathologies, achieving expert-level performance in tasks like identifying fractures, stenosis, infections, and tumors. Beyond diagnosis, AI also assists surgical planning via synthetic computed tomography generation, augmented reality systems, and robotic guidance. Furthermore, AI image analysis combined with clinical data enables personalized predictions to guide treatment decisions, such as forecasting spine surgery outcomes. However, challenges still need to be addressed in implementing AI clinically, including model interpretability, generalizability, and data limitations. Multicenter collaboration using large, diverse datasets is critical to advance the field further. While adoption barriers persist, AI presents a transformative opportunity to revolutionize spinal imaging workflows, empowering clinicians to translate data into actionable insights for improved patient care.

Diagnostic evaluation of deep learning accelerated lumbar spine MRI

BACKGROUND AND PURPOSE

Deep learning (DL) accelerated MR techniques have emerged as a promising approach to accelerate routine MR exams. While prior studies explored DL acceleration for specific lumbar MRI sequences, a gap remains in comprehending the impact of a fully DL-based MRI protocol on scan time and diagnostic quality for routine lumbar spine MRI. To address this, we assessed the image quality and diagnostic performance of a DL-accelerated lumbar spine MRI protocol in comparison to a conventional protocol.

METHODS

We prospectively evaluated 36 consecutive outpatients undergoing non-contrast enhanced lumbar spine MRIs. Both protocols included sagittal T1, T2, STIR, and axial T2-weighted images. Two blinded neuroradiologists independently reviewed images for foraminal stenosis, spinal canal stenosis, nerve root compression, and facet arthropathy. Grading comparison employed the Wilcoxon signed rank test. For the head-to-head comparison, a 5-point Likert scale to assess image quality, considering artifacts, signal-to-noise ratio (SNR), anatomical structure visualization, and overall diagnostic quality. We applied a 15% noninferiority margin to determine whether the DL-accelerated protocol was noninferior.

RESULTS

No significant differences existed between protocols when evaluating foraminal and spinal canal stenosis, nerve compression, or facet arthropathy (all p > .05). The DL-spine protocol was noninferior for overall diagnostic quality and visualization of the cord, CSF, intervertebral disc, and nerve roots. However, it exhibited reduced SNR and increased artifact perception. Interobserver reproducibility ranged from moderate to substantial (κ = 0.50-0.76).

CONCLUSION

Our study indicates that DL reconstruction in spine imaging effectively reduces acquisition times while maintaining comparable diagnostic quality to conventional MRI.

Practical Applications of Artificial Intelligence in Spine Imaging: A Review

Artificial intelligence (AI), a transformative technology with unprecedented potential in medical imaging, can be applied to various spinal pathologies. AI-based approaches may improve imaging efficiency, diagnostic accuracy, and interpretation, which is essential for positive patient outcomes. This review explores AI algorithms, techniques, and applications in spine imaging, highlighting diagnostic impact and challenges with future directions for integrating AI into spine imaging workflow.

AI-Based Measurement of Lumbar Spinal Stenosis on MRI: External Evaluation of a Fully Automated Model

OBJECTIVES

The aim of this study was to clinically validate a fully automated AI model for magnetic resonance imaging (MRI)-based quantifications of lumbar spinal canal stenosis.

MATERIALS AND METHODS

This retrospective study included lumbar spine MRI of 100 consecutive clinical patients (56 ± 17 years; 43 females, 57 males) performed on clinical 1.5 (51 examinations) and 3 T MRI scanners (49 examinations) with heterogeneous clinical imaging protocols. The AI model performed segmentations of the thecal sac on axial T2-weighted sequences. Based on these segmentations, the anteroposterior (AP) and mediolateral (ML) distance, and the area of the thecal sac were measured in a fully automated manner. For comparison, 2 fellowship-trained musculoskeletal radiologists performed the same segmentations and measurements independently. Statistics included 1-sample t tests, the intraclass correlation coefficient (ICC), Bland-Altman plots, and Dice coefficients. A P value of <0.05 was considered statistically significant.

RESULTS

The average measurements of the AI model, reader 1, and reader 2 were 194 ± 72 mm2, 181 ± 71 mm2, and 179 ± 70 mm2 for thecal sac area, 13 ± 3.3 mm, 12.6 ± 3.3 mm, and 12.6 ± 3.2 mm for AP distance, and 19.5 ± 3.9 mm, 20 ± 4.3 mm, and 19.4 ± 4 mm for ML distance, respectively. Significant differences existed for all pairwise comparisons, besides reader 1 versus AI model for the ML distance and reader 1 versus reader 2 for the AP distance (P = 0.1 and P = 0.21, respectively). The pairwise mean absolute errors among reader 1, reader 2, and the AI model ranged from 0.59 mm and 0.75 mm for the AP distance, from 1.16 mm to 1.37 mm for the ML distance, and from 7.9 mm2 to 15.54 mm2 for the thecal sac area. Pairwise ICCs among reader 1, reader 2, and the AI model ranged from 0.91 and 0.94 for the AP distance and from 0.86 to 0.9 for the ML distance without significant differences. For the thecal sac area, the pairwise ICC between both readers and the AI model of 0.97 each was slightly, but significantly lower than the ICC between reader 1 and reader 2 of 0.99. Similarly, the Dice coefficient and Hausdorff distance between both readers and the AI model were significantly lower than the values between reader 1 and reader 2, overall ranging from 0.93 to 0.95 for the Dice coefficients and 1.1 to 1.44 for the Hausdorff distances.

CONCLUSIONS

The investigated AI model is reliable for assessing the AP and the ML thecal sac diameters with human level accuracies. The small differences for measurement and segmentation of the thecal sac area between the AI model and the radiologists are likely within a clinically acceptable range.

Machine Learning Predicts Decompression Levels for Lumbar Spinal Stenosis Using Canal Radiomic Features from Computed Tomography Myelography

BACKGROUND

The accurate preoperative identification of decompression levels is crucial for the success of surgery in patients with multi-level lumbar spinal stenosis (LSS). The objective of this study was to develop machine learning (ML) classifiers that can predict decompression levels using computed tomography myelography (CTM) data from LSS patients.

METHODS

A total of 1095 lumbar levels from 219 patients were included in this study. The bony spinal canal in CTM images was manually delineated, and radiomic features were extracted. The extracted data were randomly divided into training and testing datasets (8:2). Six feature selection methods combined with 12 ML algorithms were employed, resulting in a total of 72 ML classifiers. The main evaluation indicator for all classifiers was the area under the curve of the receiver operating characteristic (ROC-AUC), with the precision-recall AUC (PR-AUC) serving as the secondary indicator. The prediction outcome of ML classifiers was decompression level or not.

RESULTS

The embedding linear support vector (embeddingLSVC) was the optimal feature selection method. The feature importance analysis revealed the top 5 important features of the 15 radiomic predictors, which included 2 texture features, 2 first-order intensity features, and 1 shape feature. Except for shape features, these features might be eye-discernible but hardly quantified. The top two ML classifiers were embeddingLSVC combined with support vector machine (EmbeddingLSVC_SVM) and embeddingLSVC combined with gradient boosting (EmbeddingLSVC_GradientBoost). These classifiers achieved ROC-AUCs over 0.90 and PR-AUCs over 0.80 in independent testing among the 72 classifiers. Further comparisons indicated that EmbeddingLSVC_SVM appeared to be the optimal classifier, demonstrating superior discrimination ability, slight advantages in the Brier scores on the calibration curve, and Net benefits on the Decision Curve Analysis.

CONCLUSIONS

ML successfully extracted valuable and interpretable radiomic features from the spinal canal using CTM images, and accurately predicted decompression levels for LSS patients. The EmbeddingLSVC_SVM classifier has the potential to assist surgical decision making in clinical practice, as it showed high discrimination, advantageous calibration, and competitive utility in selecting decompression levels in LSS patients using canal radiomic features from CTM.

Nerve Root Compression Analysis to Find Lumbar Spine Stenosis on MRI Using CNN

Lumbar spine stenosis (LSS) is caused by low back pain that exerts pressure on the nerves in the spine. Detecting LSS is a significantly important yet difficult task. It is detected by analyzing the area of the anteroposterior diameter of the patient's lumbar spine. Currently, the versatility and accuracy of LSS segmentation algorithms are limited. The objective of this research is to use magnetic resonance imaging (MRI) to automatically categorize LSS. This study presents a convolutional neural network (CNN)-based method to detect LSS using MRI images. Radiological grading is performed on a publicly available dataset. Four regions of interest (ROIs) are determined to diagnose LSS with normal, mild, moderate, and severe gradings. The experiments are performed on 1545 axial-view MRI images. Furthermore, two datasets-multi-ROI and single-ROI-are created. For training and testing, an 80:20 ratio of randomly selected labeled datasets is used, with fivefold cross-validation. The results of the proposed model reveal a 97.01% accuracy for multi-ROI and 97.71% accuracy for single-ROI. The proposed computer-aided diagnosis approach can significantly improve diagnostic accuracy in everyday clinical workflows to assist medical experts in decision making. The proposed CNN-based MRI image segmentation approach shows its efficacy on a variety of datasets. Results are compared to existing state-of-the-art studies, indicating the superior performance of the proposed approach.