A quantitative comparison between manual segmentation and threshold-based segmentation of CLSM recorded images

Segmentation is the process of defining distinct objects in an image. Object segmentation of two-dimensional images is often accomplished by a time consuming manual process where trained persons trace a line along the boundary of the object. Significant effort has been directed towards various computer segmentation algorithms to reduce the time required to segment each object. Often the question arises as to the accuracy of the computer segmentation results. This paper makes a quantitative comparison between the segmented object from a threshold-based computer segmentation process and the manual segmentation results from a group of volunteers. This comparison is based on the fact that humans have an intuitive capability to recognize objects. The image sample used in this report is a portion of the brain of the common housefly, Musca domestica. The very small size of the object makes it impractical to compare the computer segmentation results to the actual object.