In clinical infertility treatments, assessment of ovarian follicle growth by ultrasonography is important. In order to measure the geometrical characteristics of the human follicle, such as the area of the cross-sectional image and the volume inside the follicle, a method based on manual tracing of the follicle contour from the ultrasonic B mode image is widely used. However, the observable ultrasonic B mode images are sometimes imperfect and some parts of the follicle border are missing due to the existence of the acoustic shadow. In this paper, a method that interpolates the missing part of the follicle border from the known part is proposed. This method uses a priori information of the follicle, which is usually known in actual cases: (1) the follicle's surface is so smooth that its border is assumed to be a smooth closed curve; and (2) the position of the follicle's center is roughly predicted in advance in the ultrasonic B-mode image. In the proposed method, the missing part of the human follicle border is interpolated from the known part by applying an iterative revision so as to satisfy the smoothness condition of the follicle. This method is also applied to three-dimensional image reconstruction of the human follicle.
在临床不孕症治疗中,通过超声检查评估卵巢卵泡生长是很重要的。为了测量人类卵泡的几何特征,如横截面图像的面积和卵泡内部的体积,一种基于从超声B模式图像手动跟踪卵泡轮廓的方法被广泛使用。然而,可观察到的超声B模式图像有时是不完整的,由于声影的存在,卵泡边界的一些部分可能会缺失。本文提出了一种从已知部分插值卵泡边界缺失部分的方法。该方法利用了卵泡的先验信息,通常在实际情况下是已知的:(1)卵泡表面非常光滑,其边界被假定为光滑的闭合曲线;(2)卵泡中心的位置在超声B模式图像中大致预测。在提出的方法中,通过应用迭代修订,从已知部分插值人类卵泡边界的缺失部分,以满足卵泡的平滑条件。该方法还应用于人类卵泡的三维图像重建。