Region grow approach to detect anomalous shapes of spread of cell transformations from radiological images
T. R. Gopalakrishnan Nair, Richa Sharma · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2011
Segmentation of biological/medical images is a fundamental approach to isolate the regions of anomalous cell transformation in many high level applications. Segmentation carried out using seeded region growing technique has become popular because of its ability to involve the knowledge of anatomical structures in seed creation and selection process [13]. In this paper two different algorithms are carried out to build the regions of anomaly. This paper has implemented a 4 neighbor and a 8 neighbor region growing technique and observed that 4 neighbor approaches tend to fail to reach the actual boundary of the anomalous shape in certain cases. It calls for the extra care needed in generating approaches for such regions of interest for further research. It was also investigated that how a semi adaptive approach is using region mean to modify the threshold.