RESUMO
Background: Detecting vertebral structural damage in patients with ankylosing spondylitis (AS) is crucial for understanding disease progression and in research settings. Objectives: This study aimed to use deep learning to score the modified Stoke Ankylosing Spondylitis Spinal Score (mSASSS) using lateral X-ray images of the cervical and lumbar spine in patients with AS in Asian populations. Design: A deep learning model was developed to automate the scoring of mSASSS based on X-ray images. Methods: We enrolled patients with AS at a tertiary medical center in Taiwan from August 1, 2001 to December 30, 2020. A localization module was used to locate the vertebral bodies in the images of the cervical and lumbar spine. Images were then extracted from these localized points and fed into a classification module to determine whether common lesions of AS were present. The scores of each localized point were calculated based on the presence of these lesions and summed to obtain the total mSASSS score. The performance of the model was evaluated on both validation set and testing set. Results: This study reviewed X-ray image data from 554 patients diagnosed with AS, which were then annotated by 3 medical experts for structural changes. The accuracy for judging various structural changes in the validation set ranged from 0.886 to 0.985, whereas the accuracy for scoring the single vertebral corner in the test set was 0.865. Conclusion: This study demonstrated a well-trained deep learning model of mSASSS scoring for detecting the vertebral structural damage in patients with AS at an accuracy rate of 86.5%. This artificial intelligence model would provide real-time mSASSS assessment for physicians to help better assist in radiographic status evaluation with minimal human errors. Furthermore, it can assist in a research setting by offering a consistent and objective method of scoring, which could enhance the reproducibility and reliability of clinical studies.
RESUMO
Screening for osteoporosis is crucial for early detection and prevention, yet it faces challenges due to the low accuracy of calcaneal quantitative ultrasound (QUS) and limited access to dual-energy X-ray absorptiometry (DXA) scans. Recent advances in AI offer a promising solution through opportunistic screening using existing medical images. This study aims to utilize deep learning techniques to develop a model that analyzes chest X-ray (CXR) images for osteoporosis screening. This study included the AI model development stage and the clinical validation stage. In the AI model development stage, the combined dataset of 5122 paired CXR images and DXA reports from the patients aged 20 to 98 years at a medical center was collected. The images were enhanced and filtered for hardware retention such as pedicle screws, bone cement, artificial intervertebral discs or severe deformity in target level of T12 and L1. The dataset was then separated into training, validating, and testing datasets for model training and performance validation. In the clinical validation stage, we collected 440 paired CXR images and DXA reports from both the TCVGH and Joy Clinic, including 304 pared data from TCVGH and 136 paired data from Joy Clinic. The pre-clinical test yielded an area under the curve (AUC) of 0.940, while the clinical validation showed an AUC of 0.946. Pearson's correlation coefficient was 0.88. The model demonstrated an overall accuracy, sensitivity, and specificity of 89.0%, 88.7%, and 89.4%, respectively. This study proposes an AI model for opportunistic osteoporosis screening through CXR, demonstrating good performance and suggesting its potential for broad adoption in preliminary screening among high-risk populations.