Osteoporosis, a condition characterized by weakened bones and decreased bone mass, is a growing concern as the global population increases. The current methods used to diagnose osteoporosis, such as central dual-energy X-ray absorptiometry (DXA), have limitations that contribute to the underdiagnosis and undertreatment of the condition. However, a team of researchers from Korea University College of Medicine has developed a groundbreaking tool called DL-BMD, which utilizes deep learning technology to automate bone mineral density (BMD) measurements and revolutionize osteoporosis screening.
DL-BMD is a highly efficient and precise solution that leverages routine computed tomography (CT) scans to measure bone mineral density on the lumbar spine. Unlike traditional approaches that require specialized imaging techniques, DL-BMD enables opportunistic screening for osteoporosis by utilizing existing CT scans, making it more accessible and accurate.
The DL-BMD tool is built upon a segmentation network called U-Net, which is designed to locate the lumbar spine. The researchers also incorporated additional techniques, such as field of view augmentation and CT denoising, to improve the reliability of the tool in different scan settings. The model was trained using a diverse dataset of CT scans from various sources, and data augmentation and pre-processing steps were implemented to enhance its generalizability.
During the screening process, DL-BMD uses a region of interest (ROI) placement algorithm to create an elliptical ROI that excludes cortical bone and avoids the basivertebral vein. The selected slices, typically including the L1 and L2 vertebrae, undergo a calculation of Hounsfield unit (HU) values within the ROI. The accuracy of DL-BMD relies on the conversion of these HU values into bone mineral density (BMD), which is calibrated against the European Spine Phantom for precise measurements.
The introduction of DL-BMD represents a significant advancement in osteoporosis screening, using deep learning techniques to enhance the accuracy of diagnostic evaluations. By addressing the limitations of traditional methods, DL-BMD enables more efficient and accessible opportunistic screening through routine CT scans. This breakthrough has the potential to improve early identification and proactive prevention of osteoporotic fractures, leading to enhanced bone health on a larger scale.
The source of the article is from the blog trebujena.net