Detecting the neural processes like axons and dendrites needs high quality SEM images. This paper proposes an approach using perceptual grouping via a graph cut and its combinations with Convolutional Neural Network (CNN) to achieve improved segmentation of SEM images. Experimental results demonstrate improved computational efficiency with linear running time.
Machine learning system refers to the one which automatically accumulates knowledge about new environments based on experience to recognize complex patterns. This ability to learn from experience, analytical observation, and other means, results in a system that can improve its own speed and performance. In this work, Convolutional Neural Network is used for learning how to segment images. Convolutional Neural Networks (CNN) extract features directly from pixel images with minimal preprocessing. It can even able to recognize a pattern which has not been presented before, provided it resembles one of the training patterns. After learning (from ground-truth image), CNN automatically generate a good affinity graph from raw SEM images. This affinity graph can be then paired with any standard partitioning algorithm, such as N-cut, connected component to achieve improved segmentation. In this paper, we demonstrate the use of combined approach, where a Convolutional Neural Network and Connected Component algorithm(CC) are used to segment SEM images. F-score of this algorithm was found to be 78%.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.