Enhanced imaging with multi-pinhole single-photon emission computed tomography

点击翻译按钮获取中文摘要

Single-photon emission computed tomography (SPECT) is a commonly used imaging modality. Cardiac-specific multi-pinhole cameras with solid-state detectors are now in use that have improved sensitivity and improved spatial, energy and temporal resolution allowing dynamic 3D imaging. However, pinhole collimators produce variable resolution, magnification, and sensitivity across the camera's field of view. These factors lead to image artifacts that may cause inaccurate interpretation and misdiagnosis. They also make it more difficult to correct for image degrading factors such as attenuation, scatter, patient motion, which is essential for accurate qualitative and quantitative evaluation. Multi-pinhole SPECT is currently interpreted without any of these corrections, reducing image quality and test accuracy. The long-term goal of my program is to improve the quantitative accuracy and precision of imaging with multi-pinhole cardiac SPECT(/CT). My short-term objectives are: Aim 1) Develop novel methods for co-registration of transmission and emission datasets using data consistency conditions for the purpose of attenuation correction. Aim 2) Use accurate scatter modeling to integrate the signal from lower energy windows and thereby increase the signal-to-noise ratio (SNR) of pinhole SPECT images. Aim 3) Develop novel methods of motion correction (MC) for dynamic SPECT imaging to improve the accuracy, precision and repeatability of MBF measurements. The general approach of my laboratory is to develop and validate advanced image reconstruction and analysis methods using analytic or Monte Carlo computer simulations, followed by evaluation using physical phantom experiments and datasets from real patient studies. Attenuation, scatter and motion depend strongly on body shape and size. Thus in selecting data sets for evaluation, we use a mixture of scans to ensure a diverse sampling of sexes, body shapes and sizes. Extra-cardiac signals violate the assumptions of the data consistency conditions, so Aim 1 will use cardiac contractile motion to separate the heart from static background signals. Aims 2 and 3 will use machine-learning (ML) to estimate scatter over a range of energies and to register images and estimate motion from dynamic imaging. The proposed work will exploit the energy-dimension and time-dimension of the acquired SPECT data to enhance and expand the information available for medical imaging. My lab provides a rich training environment for undergraduate and graduate student education. As part of the CAMPEP accredited Ottawa-Carleton joint program in medical physics, students interact with others engaged in a wide variety of medical physics projects. As part of the University of Ottawa Heart Institute, students are also exposed to the medical motivation that drives much of this field of research as well as to the broad multi-disciplinary nature of this field with exposure to research from microbiology to radiochemistry and tissue engineering.