A - Imaging systems currently in development:

1. Advanced positron emission tomography (PET) system dedicated to breast cancer imaging
2. High resolution PET system for imaging small animal models of disease
3. PET detector that can be used simultaneously in an MR system
4. Hand held gamma ray camera for surgical cancer staging

B - Research topics involved:

1. High resolution photon sensors
2. Readout electronics and data acquisition
3. Signal and image processing algorithms
4. Tomographic image reconstruction on graphics hardware
5. Computer simulation and system performance analysis

A1 - Advanced positron emission tomography (PET) system dedicated to breast cancer imaging

A2 - High resolution PET system for imaging small animal models of disease

A3 - PET detector that can be used simultaneously in an MR system

A4 - Hand held gamma ray camera for surgical cancer staging

A compact, hand-held gamma camera with excellent intrinsic and extrinsic performance has been developed for the rapid identification and localization of the sentinel lymph node during the surgical staging of cancer. A goal for this device is an image acquisition time of five seconds to allow the surgeon to easily search for points of interest without excessive motion blurring. The camera comprises a 5x5 cm2 field of view NaI (Tl) pixellated crystal array, a high sensitivity (2.0 cm thick) hexagonal parallel-hole collimator, a position sensitive photomultiplier tube (PSPMT), and a novel highly multiplexed electrical readout. The novel software signal processing algorithms incorporate optical flow-based adaptive exposure control and motion-compensated filtering.

B1 - High resolution photon sensors

B2 - Readout electronics and data acquisition

B3 - Signal and image processing algorithms

B4 - Tomographic image reconstruction on graphics hardware

The number of lines-of-response (LOR) in modern positron emission tomography systems has increased by orders of magnitude. This trend has been driven by the use of smaller detector crystals, more accurate depth-of-interaction positioning and fully-3D acquisition. This has boosted the resolution and the sensitivity of PET systems. However, it has made the task of reconstructing images from the collected data more difficult. The demand in computation power and memory storage as exploded, outpacing the advances in memory capacity and processor performance.

We are investigating practical ways to perform fully-3D OSEM reconstruction using programmable graphics hardware known as graphics processing unit (GPU). Primarily designed to deliver high-definition graphics for video games in real-time, GPUs are now increasingly being used as cost-effective high-performance co-processors for scientific computing. GPUs are characterized by extremely high processing parallelism, fast clock-rate, high-bandwidth memory access and built-in optimized geometrical functions. However, they have a quite limited amount of memory (512 Mb). Nevertheless, these characteristics make them particularly well suited for ’on-the-fly’ schemes with low memory profile but high computational intensity. Implementation of 3D-OSEM on the GPU is challenging because the graphics programming interface is not designed to handle general-purpose computation. Yet, we show that the two main components of 3D-OSEM (line back-projection and line forward projection) can be reformulated as pseudo-rendering tasks that can be run extremely efficiently on the GPU.

B5 - Computer simulation and system performance analysis