A METHODOLOGY FOR EVALUATION OF BOUNDARY DETECTION ALGORITHMS ON MEDICAL IMAGES

Image segmentation is the partition of an image into a set of nonoverl apping regions whose union is the entire image, The image is decompose d into meaningful parts which are uniform with respect to certain char acteristics, such as gray level or texture. In this paper, we propose a methodology for evaluating medical image segmentation algorithms whe rein the only information available is boundaries outlined by multiple expert observers. In this case, the results of the segmentation algor ithm can be evaluated against the multiple observers' outlines. We hav e derived statistics to enable us to find whether the computer-generat ed boundaries agree with the observers' hand-outlined boundaries as mu ch as the different observers agree with each other, We illustrate the use of this methodology by evaluating image segmentation algorithms o n two different applications in ultrasound imaging, In the first appli cation, we attempt to find the epicardial and endocardial boundaries f rom cardiac ultrasound images, and in the second application, our goal is to find the fetal skull and abdomen boundaries from prenatal ultra sound images.