Multiorgan detection: Deep learning based techniques and research directions

Abstract

Automatic organ segmentation plays an important role in clinical procedures such as planning of radiation therapies and in computer-aided diagnostic systems. Several state-of –art techniques are available for multiorgan segmentation, however deep learning methods are doing exceptionally well and become the methodology of choice to analyze medical images. This intensively carried out work is conducted for deep learning methods applied on various organs in abdominal CT images. Firstly, this paper formulates segmentation, semantic segmentation problem and their methods. Secondly, multiorgan detection techniques based on deep learning along with their contributions, chosen datasets and gaps are discussed. It presents the metrics used to evaluate these methods. Finally, interesting conclusions has been drawn which will add to do future work using deep learning.