| 摘要: | Spondylolisthesis, a common spinal disorder, is the relative displacement between the two vertebra due to one being shifted away from the spine's smooth curvature. It usually occurs in the lower part of the spine and is more prevalent among women aged older than 60 years. Age-related diseases, like degenerative spondylolisthesis, place a substantial burden on healthcare systems of any society, and require care from a specialized network of health professionals and support services. Here, we proposed LumbarNet, a computer-aided algorithm to diagnose lumbar vertebral slippage detection from X-ray images, and assessed its efficiency. By collaborating U-Net, a feature fusion module (FFM) and utilizing P-grade, piecewise slope detection (PSD), as well as dynamic shift (DS), our model achieved 0.88 in mean intersection over union (mIOU) for vertebral segmentation and 88.83% in spondylolisthesis detection for lumbar spine. These results showed that LumbarNet outcompeted U-Net in delineating the complex structure of lumbar spine from lateral radiographic images and could be applied as a potential method to detect spondylolisthesis on clinical practice.
Furthermore, neural decompression and intervertebral instrumented fusion were the standard procedure for spondylolisthesis patients failed to conservative treatment. With the advancements of surgical instruments, minimally invasive surgery (MIS) was applied widely in spinal surgery. However, we still lack of the evidence to confirm which procedure is optimal treatment. Therefore, we conducted a systematic review and meta-analysis to evaluate the clinical outcomes of Endo-TLIF and MIS-TLIF in spondylolisthesis treatment. After collecting suitable reports on electronic databases and using Revman 5.4 software to analyze, our results indicated that Endo-TLIF could be a potential method in spinal surgery with significant difference in blood loss, ambulation time, hospitalization time, post-operative 2 weeks and 3 months VAS back pain. |
