Depth of interaction analysis method to eliminate parallax error for brain-dedicated positron emission tomography

Abstract

As a part of medical diagnosis that falls in the region of nuclear medicine, positron emission tomography (PET) scanning has been in the limelight since it was able to feature the physiological and biochemical functions in the body as an aid to the diagnosis of disease. As the possibility of using PET is suggested in diagnosing incurable diseases such as early diagnosis of Alzheimer's disease and temporal lobe epilepsy. Brain-dedicated positron emission tomography (B-PET) has been developed for the purpose of utilizing basic and clinical research on these neurological fields. In small PET such as B-PET and micro PET, the deterioration of spatial resolution due to parallax error is noticeable compared to whole-body PET. This dissertation presents a depth of interaction (DOI) analysis method to eliminate parallax errors encountered in small PET. The proposed analysis method determined the DOI by light spread and machine learning algorithm, and the results of the study showed that the DOI information could be discrete properly with a probability of more than 80 %. In addition, we also identified the possibility of elimination of parallax error by continuously determining the DOI information through supplementation of the studies conducted. This study is expected to contribute to improving PET image quality and basic research in the nervous system.