Automating image segmentation verification and validation by learning test oracles

An Image Segmentation Algorithm is an algorithm that delineates (an) object(s) of interest in an image. The output of the image segmentation algorithm is referred to as a segmentation. Developing image segmentation algorithms is a manual, iterative process involving repetitive verification and validation tasks. This process is time-consuming and depends on the availability of medical experts, who are a scarce resource. We propose a procedure that uses machine learning to construct an oracle, which can then be used to automatically verify the correctness of image segmentations, thus saving substantial resources. During the initial learning phase, segmentations from the first few (optimally two) revisions of the segmentation algorithm are manually verified by experts. The similarity of successive segmentations of the same images is also measured. This information is then fed to a machine learning algorithm to construct a classifier that distinguishes between consistent and inconsistent segmentation pairs based on the values of the similarity measures associated with each segmentation pair. Once the accuracy of the classifier is deemed satisfactory for the purposes of the application, the classifier is then used to determine whether the segmentation, systems’ output by subsequent versions of the algorithm under test, are (in)consistent with already verified segmentations from previous versions. This information is then used to automatically make conclusions about the (in)correctness of the segmentations. To demonstrate the performance of the Carleton University, TR SCE-09-06 2 approach, the proposed solution was successfully applied to 3D segmentations of the cardiac left ventricle obtained from CT scans.