A deep learning-based technique for the diagnosis of epidural spinal cord compression on thoracolumbar CT.

PURPOSE: To develop a deep learning (DL) model for epidural spinal cord compression (ESCC) on CT, which will aid earlier ESCC diagnosis for less experienced clinicians. METHODS: We retrospectively collected CT and MRI data from adult patients with suspected ESCC at a tertiary referral institute from 2007 till 2020. A total of 183 patients were used for training/validation of the DL model. A separate test set of 40 patients was used for DL model evaluation and comprised 60 staging CT and matched MRI scans performed with an interval of up to 2 months. DL model performance was compared to eight readers: one musculoskeletal radiologist, two body radiologists, one spine surgeon, and four trainee spine surgeons. Diagnostic performance was evaluated using inter-rater agreement, sensitivity, specificity and AUC. RESULTS: Overall, 3115 axial CT slices were assessed. The DL model showed high kappa of 0.872 for normal, low and high-grade ESCC (trichotomous), which was superior compared to a body radiologist (R4, κ = 0.667) and all four trainee spine surgeons (κ range = 0.625-0.838)(all p < 0.001). In addition, for dichotomous normal versus any grade of ESCC detection, the DL model showed high kappa (κ = 0.879), sensitivity (91.82), specificity (92.01) and AUC (0.919), with the latter AUC superior to all readers (AUC range = 0.732-0.859, all p < 0.001). CONCLUSION: A deep learning model for the objective assessment of ESCC on CT had comparable or superior performance to radiologists and spine surgeons. Earlier diagnosis of ESCC on CT could reduce treatment delays, which are associated with poor outcomes, increased costs, and reduced survival.

Improved Productivity Using Deep Learning-assisted Reporting for Lumbar Spine MRI.

Background Lumbar spine MRI studies are widely used for back pain assessment. Interpretation involves grading lumbar spinal stenosis, which is repetitive and time consuming. Deep learning (DL) could provide faster and more consistent interpretation. Purpose To assess the speed and interobserver agreement of radiologists for reporting lumbar spinal stenosis with and without DL assistance. Materials and Methods In this retrospective study, a DL model designed to assist radiologists in the interpretation of spinal canal, lateral recess, and neural foraminal stenoses on lumbar spine MRI scans was used. Randomly selected lumbar spine MRI studies obtained in patients with back pain who were 18 years and older over a 3-year period, from September 2015 to September 2018, were included in an internal test data set. Studies with instrumentation and scoliosis were excluded. Eight radiologists, each with 2-13 years of experience in spine MRI interpretation, reviewed studies with and without DL model assistance with a 1-month washout period. Time to diagnosis (in seconds) and interobserver agreement (using Gwet κ) were assessed for stenosis grading for each radiologist with and without the DL model and compared with test data set labels provided by an external musculoskeletal radiologist (with 32 years of experience) as the reference standard. Results Overall, 444 images in 25 patients (mean age, 51 years ± 20 [SD]; 14 women) were evaluated in a test data set. DL-assisted radiologists had a reduced interpretation time per spine MRI study, from a mean of 124-274 seconds (SD, 25-88 seconds) to 47-71 seconds (SD, 24-29 seconds) (P < .001). DL-assisted radiologists had either superior or equivalent interobserver agreement for all stenosis gradings compared with unassisted radiologists. DL-assisted general and in-training radiologists improved their interobserver agreement for four-class neural foraminal stenosis, with κ values of 0.71 and 0.70 (with DL) versus 0.39 and 0.39 (without DL), respectively (both P < .001). Conclusion Radiologists who were assisted by deep learning for interpretation of lumbar spinal stenosis on MRI scans showed a marked reduction in reporting time and superior or equivalent interobserver agreement for all stenosis gradings compared with radiologists who were unassisted by deep learning. © RSNA, 2022 Online supplemental material is available for this article. See also the editorial by Hayashi in this issue.

Deep learning model for automated diagnosis of degenerative cervical spondylosis and altered spinal cord signal on MRI.

BACKGROUND CONTEXT: A deep learning (DL) model for degenerative cervical spondylosis on MRI could enhance reporting consistency and efficiency, addressing a significant global health issue. PURPOSE: Create a DL model to detect and classify cervical cord signal abnormalities, spinal canal and neural foraminal stenosis. STUDY DESIGN/SETTING: Retrospective study conducted from January 2013 to July 2021, excluding cases with instrumentation. PATIENT SAMPLE: Overall, 504 MRI cervical spines were analyzed (504 patients, mean=58 years±13.7[SD]; 202 women) with 454 for training (90%) and 50 (10%) for internal testing. In addition, 100 MRI cervical spines were available for external testing (100 patients, mean=60 years±13.0[SD];26 women). OUTCOME MEASURES: Automated detection and classification of spinal canal stenosis, neural foraminal stenosis, and cord signal abnormality using the DL model. Recall(%), inter-rater agreement (Gwet's kappa), sensitivity, and specificity were calculated. METHODS: Utilizing axial T2-weighted gradient echo and sagittal T2-weighted images, a transformer-based DL model was trained on data labeled by an experienced musculoskeletal radiologist (12 years of experience). Internal testing involved data labeled in consensus by 2 musculoskeletal radiologists (reference standard, both with 12-years-experience), 2 subspecialist radiologists, and 2 in-training radiologists. External testing was performed. RESULTS: The DL model exhibited substantial agreement surpassing all readers in all classes for spinal canal (κ=0.78, p<.001 vs. κ range=0.57-0.70 for readers) and neural foraminal stenosis (κ=0.80, p<.001 vs. κ range=0.63-0.69 for readers) classification. The DL model's recall for cord signal abnormality (92.3%) was similar to all readers (range: 92.3-100.0%). Nearly perfect agreement was demonstrated for binary classification (grades 0/1 vs. 2/3) (κ=0.95, p<.001 for spinal canal; κ=0.90, p<.001 for neural foramina). External testing showed substantial agreement using all classes (κ=0.76, p<.001 for spinal canal; κ=0.66, p<.001 for neural foramina) and high recall for cord signal abnormality (91.9%). The DL model demonstrated high sensitivities (range:83.7%-92.4%) and specificities (range:87.8%-98.3%) on both internal and external datasets for spinal canal and neural foramina classification. CONCLUSIONS: Our DL model for degenerative cervical spondylosis on MRI showed good performance, demonstrating substantial agreement with the reference standard. This tool could assist radiologists in improving the efficiency and consistency of MRI cervical spondylosis assessments in clinical practice.

Application of Machine Learning for Differentiating Bone Malignancy on Imaging: A Systematic Review.

An accurate diagnosis of bone tumours on imaging is crucial for appropriate and successful treatment. The advent of Artificial intelligence (AI) and machine learning methods to characterize and assess bone tumours on various imaging modalities may assist in the diagnostic workflow. The purpose of this review article is to summarise the most recent evidence for AI techniques using imaging for differentiating benign from malignant lesions, the characterization of various malignant bone lesions, and their potential clinical application. A systematic search through electronic databases (PubMed, MEDLINE, Web of Science, and clinicaltrials.gov) was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. A total of 34 articles were retrieved from the databases and the key findings were compiled and summarised. A total of 34 articles reported the use of AI techniques to distinguish between benign vs. malignant bone lesions, of which 12 (35.3%) focused on radiographs, 12 (35.3%) on MRI, 5 (14.7%) on CT and 5 (14.7%) on PET/CT. The overall reported accuracy, sensitivity, and specificity of AI in distinguishing between benign vs. malignant bone lesions ranges from 0.44-0.99, 0.63-1.00, and 0.73-0.96, respectively, with AUCs of 0.73-0.96. In conclusion, the use of AI to discriminate bone lesions on imaging has achieved a relatively good performance in various imaging modalities, with high sensitivity, specificity, and accuracy for distinguishing between benign vs. malignant lesions in several cohort studies. However, further research is necessary to test the clinical performance of these algorithms before they can be facilitated and integrated into routine clinical practice.

Deep learning assessment compared to radiologist reporting for metastatic spinal cord compression on CT.

Introduction: Metastatic spinal cord compression (MSCC) is a disastrous complication of advanced malignancy. A deep learning (DL) algorithm for MSCC classification on CT could expedite timely diagnosis. In this study, we externally test a DL algorithm for MSCC classification on CT and compare with radiologist assessment. Methods: Retrospective collection of CT and corresponding MRI from patients with suspected MSCC was conducted from September 2007 to September 2020. Exclusion criteria were scans with instrumentation, no intravenous contrast, motion artefacts and non-thoracic coverage. Internal CT dataset split was 84% for training/validation and 16% for testing. An external test set was also utilised. Internal training/validation sets were labelled by radiologists with spine imaging specialization (6 and 11-years post-board certification) and were used to further develop a DL algorithm for MSCC classification. The spine imaging specialist (11-years expertise) labelled the test sets (reference standard). For evaluation of DL algorithm performance, internal and external test data were independently reviewed by four radiologists: two spine specialists (Rad1 and Rad2, 7 and 5-years post-board certification, respectively) and two oncological imaging specialists (Rad3 and Rad4, 3 and 5-years post-board certification, respectively). DL model performance was also compared against the CT report issued by the radiologist in a real clinical setting. Inter-rater agreement (Gwet's kappa) and sensitivity/specificity/AUCs were calculated. Results: Overall, 420 CT scans were evaluated (225 patients, mean age=60 ± 11.9[SD]); 354(84%) CTs for training/validation and 66(16%) CTs for internal testing. The DL algorithm showed high inter-rater agreement for three-class MSCC grading with kappas of 0.872 (p<0.001) and 0.844 (p<0.001) on internal and external testing, respectively. On internal testing DL algorithm inter-rater agreement (κ=0.872) was superior to Rad 2 (κ=0.795) and Rad 3 (κ=0.724) (both p<0.001). DL algorithm kappa of 0.844 on external testing was superior to Rad 3 (κ=0.721) (p<0.001). CT report classification of high-grade MSCC disease was poor with only slight inter-rater agreement (κ=0.027) and low sensitivity (44.0), relative to the DL algorithm with almost-perfect inter-rater agreement (κ=0.813) and high sensitivity (94.0) (p<0.001). Conclusion: Deep learning algorithm for metastatic spinal cord compression on CT showed superior performance to the CT report issued by experienced radiologists and could aid earlier diagnosis.

Application of Artificial Intelligence Methods for Imaging of Spinal Metastasis.

Spinal metastasis is the most common malignant disease of the spine. Recently, major advances in machine learning and artificial intelligence technology have led to their increased use in oncological imaging. The purpose of this study is to review and summarise the present evidence for artificial intelligence applications in the detection, classification and management of spinal metastasis, along with their potential integration into clinical practice. A systematic, detailed search of the main electronic medical databases was undertaken in concordance with the PRISMA guidelines. A total of 30 articles were retrieved from the database and reviewed. Key findings of current AI applications were compiled and summarised. The main clinical applications of AI techniques include image processing, diagnosis, decision support, treatment assistance and prognostic outcomes. In the realm of spinal oncology, artificial intelligence technologies have achieved relatively good performance and hold immense potential to aid clinicians, including enhancing work efficiency and reducing adverse events. Further research is required to validate the clinical performance of the AI tools and facilitate their integration into routine clinical practice.

Diagnostic Accuracy of CT for Metastatic Epidural Spinal Cord Compression.

BACKGROUND: Early diagnosis of metastatic epidural spinal cord compression (MESCC) is vital to expedite therapy and prevent paralysis. Staging CT is performed routinely in cancer patients and presents an opportunity for earlier diagnosis. METHODS: This retrospective study included 123 CT scans from 101 patients who underwent spine MRI within 30 days, excluding 549 CT scans from 216 patients due to CT performed post-MRI, non-contrast CT, or a gap greater than 30 days between modalities. Reference standard MESCC gradings on CT were provided in consensus via two spine radiologists (11 and 7 years of experience) analyzing the MRI scans. CT scans were labeled using the original reports and by three radiologists (3, 13, and 14 years of experience) using dedicated CT windowing. RESULTS: For normal/none versus low/high-grade MESCC per CT scan, all radiologists demonstrated almost perfect agreement with kappa values ranging from 0.866 (95% CI 0.787-0.945) to 0.947 (95% CI 0.899-0.995), compared to slight agreement for the reports (kappa = 0.095, 95%CI -0.098-0.287). Radiologists also showed high sensitivities ranging from 91.51 (95% CI 84.49-96.04) to 98.11 (95% CI 93.35-99.77), compared to 44.34 (95% CI 34.69-54.31) for the reports. CONCLUSION: Dedicated radiologist review for MESCC on CT showed high interobserver agreement and sensitivity compared to the current standard of care.

Deep Learning Model for Classifying Metastatic Epidural Spinal Cord Compression on MRI.

Background: Metastatic epidural spinal cord compression (MESCC) is a devastating complication of advanced cancer. A deep learning (DL) model for automated MESCC classification on MRI could aid earlier diagnosis and referral. Purpose: To develop a DL model for automated classification of MESCC on MRI. Materials and Methods: Patients with known MESCC diagnosed on MRI between September 2007 and September 2017 were eligible. MRI studies with instrumentation, suboptimal image quality, and non-thoracic regions were excluded. Axial T2-weighted images were utilized. The internal dataset split was 82% and 18% for training/validation and test sets, respectively. External testing was also performed. Internal training/validation data were labeled using the Bilsky MESCC classification by a musculoskeletal radiologist (10-year experience) and a neuroradiologist (5-year experience). These labels were used to train a DL model utilizing a prototypical convolutional neural network. Internal and external test sets were labeled by the musculoskeletal radiologist as the reference standard. For assessment of DL model performance and interobserver variability, test sets were labeled independently by the neuroradiologist (5-year experience), a spine surgeon (5-year experience), and a radiation oncologist (11-year experience). Inter-rater agreement (Gwet's kappa) and sensitivity/specificity were calculated. Results: Overall, 215 MRI spine studies were analyzed [164 patients, mean age = 62 ± 12(SD)] with 177 (82%) for training/validation and 38 (18%) for internal testing. For internal testing, the DL model and specialists all showed almost perfect agreement (kappas = 0.92-0.98, p < 0.001) for dichotomous Bilsky classification (low versus high grade) compared to the reference standard. Similar performance was seen for external testing on a set of 32 MRI spines with the DL model and specialists all showing almost perfect agreement (kappas = 0.94-0.95, p < 0.001) compared to the reference standard. Conclusion: A DL model showed comparable agreement to a subspecialist radiologist and clinical specialists for the classification of malignant epidural spinal cord compression and could optimize earlier diagnosis and surgical referral.

Deep Learning Model for Grading Metastatic Epidural Spinal Cord Compression on Staging CT.

Background: Metastatic epidural spinal cord compression (MESCC) is a disastrous complication of advanced malignancy. Deep learning (DL) models for automatic MESCC classification on staging CT were developed to aid earlier diagnosis. Methods: This retrospective study included 444 CT staging studies from 185 patients with suspected MESCC who underwent MRI spine studies within 60 days of the CT studies. The DL model training/validation dataset consisted of 316/358 (88%) and the test set of 42/358 (12%) CT studies. Training/validation and test datasets were labeled in consensus by two subspecialized radiologists (6 and 11-years-experience) using the MRI studies as the reference standard. Test sets were labeled by the developed DL models and four radiologists (2−7 years of experience) for comparison. Results: DL models showed almost-perfect interobserver agreement for classification of CT spine images into normal, low, and high-grade MESCC, with kappas ranging from 0.873−0.911 (p < 0.001). The DL models (lowest κ = 0.873, 95% CI 0.858−0.887) also showed superior interobserver agreement compared to two of the four radiologists for three-class classification, including a specialist (κ = 0.820, 95% CI 0.803−0.837) and general radiologist (κ = 0.726, 95% CI 0.706−0.747), both p < 0.001. Conclusion: DL models for the MESCC classification on a CT showed comparable to superior interobserver agreement to radiologists and could be used to aid earlier diagnosis.