Image-selective segmentation model for multi-regions within the object of interest with application to medical disease

Detection and extraction of an object of interest and accurate boundaries segmentation in a given image has been of interest in the last decades due to its application in different fields. To successfully segment a single object, interactive/selective segmentation techniques has been developed as a supplement to the existing global segmentation techniques. Even though existing interactive/selective segmentation techniques perform well in segmenting the images with prominent edges, those methods are less efficient or even fail in segmenting images having multi-regions of different intensity scale. In this paper, we design a new variational selective segmentation model which incorporates the idea of area-based fitting term along with a signed pressure force function based on a generalized average into a variational energy function. The new model is capable to capture the object of interest which can be single or multi-region within the object of interest. To evaluate the performance of our new model, we compare our results with state of the art models by showing same efficiency and reliability on detecting single-region and an outperforming for multi-region selective segmentation. Comparison tests were carried out on synthetic and real data images.