Organ recognition using Gabor filters

The aim of this research is to investigate the possibility of using medical image information to extract unique features and classify different patients' organ tissues, such as the prostate, based on concepts related to what is already done in iris recognition. This paper therefore presents a new approach in medical imaging, an organ recognition system, tested on a standard database of grey scale prostate images in order to validate its performance. In this research, features of the prostate image were encoded by convolving the normalized organ region with a 2D Gabor filter and then quantizing its phase in order to produce a bit-wise biometric template. Our experiments prove that prostate patterns have a low degree of freedom to be used in organ recognition systems and inter-class and intraclass distributions are highly correlated. However, there are still open issues that need to be addressed for future work on organ recognition, including precise segmentation and intensive computation cost.