在諸多老化的議題中,骨質疏鬆是高齡化社會中普遍性的疾病。在台灣每年約有六萬人因骨質疏鬆而造成脊椎骨折,而患者可以透過人工骨水泥手術來進行改善。但是根據過去學者的研究發現,患者經過骨水泥手術治療後,可能會導致動手術之腰椎其上、下節之椎體骨質變型、鬆垮。 因此,本研究先透過主動形狀模型切割演算法的技術,將X光影像中的腰椎正確的分割出來後,並利用自動化的方式萃取每個椎體的特徵值角度。藉由實驗萃取出來的角度幫助醫生或護理人員判別病患在完成手術後的一年內,是否有新骨折發生之情形,以提升工作效率及減少花費的時間成本。在實驗結果中我們展示了所提出方法之分割結果,並與醫生手動圈選的結果進行比較,顯示了我們提出的方法在腰椎的分割與特徵值角度萃取上是有效與正確的。
In recent years, osteoporosis is one of the important and universality disease in many issues about aging society. Around sixty thousand people have vertebral fracture result from the osteoporosis in Taiwan every year, and percutaneous vertebroplasty can relieve pain from vertebral compression fractures. However, according to the past a few clinical researches, we find out that the patients who take percutaneous vertebroplasty treatment is able to suffer from subsequent fracture within one year, especially on the adjacent fracture. In the past, physician have to measure the angle of lumbar spine manually, then assess whether patient have a subsequent fracture within one year. In this study, we apply active shape model algorithm to segment lumbar spine X-ray image. Next, we extract the angle feature between two endplates automatically. The angle we extract can effectively assist physicians in making professional assessment. This can upgrade the work efficiency and reduce time costs. In the experiments, the segmentation results with proposed method are illustrated and in comparison with those which are segmented manually by physicians. The results indicate that our methods achieve satisfactory results.