Introduction

The Imaging Analysis Team aims to develop new and improve existing algorithms for molecular imaging with PET. Careful implementation of these algorithms is essential for obtaining accurate quantitative functional images because of the large noise presented in PET data. It is expected that many of these algorithms will contribute towards various aspects of basic and clinical molecular imaging. We also welcome interdisciplinary collaboration with other areas of molecular imaging sciences.

Research Introduction

1) Development of algorithms for neuroreceptor imaging
Neuroreceptor imaging is one of the most important goals in PET molecular imaging. However, ordinary approaches assume the existence of reference regions where target receptors are sparse, restricting the wider application of receptor study. At the same time, the number of target receptors is increasing due to newly developed radioligands, making the situation even worse. In this research, voxel-by-voxel compartmental analysis is achieved by using MAP estimation and a statistical clustering approach without reference regions.

2) Performance improvement of molecular imaging
The noise in PET data reduces the reliability of molecular images. In this work, the influence of PET noise on functional images is evaluated, and denoising approaches such as Wavelet transformation and statistical clustering are investigated. Existing algorithms are also improved to provide more precise and reliable functional images

3) Omission of arterial blood sampling
The requirement for measuring concentrations of radioligands in arterial blood is a serious problem for quantitative PET applications. It is necessary to develop methods, as our next targets, that extract blood-related information from PET data using statistical analysis rather than arterial sampling.

>> PDF (90KB)

members