使用超音波影像之腎結石影像偵測與辨識

Abstract

腎結石在現代已經成為一種很普遍的疾病，一種最常用的治療方式是震波碎石治療。然而，現行的震波碎石治療由於缺乏一自動偵測腎結石位置的機制，使得震波常未準確聚焦在結石位置，不但降低治療效率，更會對人體組織產生不可忽視的傷害。本研究提出一偵測演算法，利用結石於超音波影像中具有高亮度及具有陰影之特性，先以亮點偵測於畫面中找出可能為結石的亮點位置，再以對比估測來判定一亮點是否具有陰影，可以自動於超音波影像中偵測出結石位置，以期能夠改善震波碎石治療的效果並減少副作用。

Kidney stone is a popular disease in recent time. One kind of popular treatment is Extracorporeal Shock Wave Lithotripter (ESWL). Existing ESWL is lack of an automatic method to detect the position of kidney stone, so the shock wave miss-shot often happens. This situation often causes damage in human body. This research presents a practicable algorithm, using the property that stone in ultrasonic image has high gray level and shadow. First we find some possible positions of stone by spot detection, then we use contrast estimation to determine whether a spot is a stone or not. We can detect position of stone automatically in ultrasonic image. We hope this algorithm can improve ESWL treatment and reduce side effect