Abstract
242533
Introduction: The state-of-the-art clinical PET/CT scanners are designed to optimize the diagnostic accuracy and clinical throughput of whole-body FDG imaging for cancer detection and treatment monitoring. While many organ-specific PET systems have been proposed to enhance the system sensitivity and resolution, they are not designed to image patients who are immobilized or in intensive care. With an increasing number of novel radio-theranostics becoming available each year, there are demands for versatile and accessible molecular imaging tools prompting us to the development of the Point-of-Care (POC) PET technology to support interactive scanning, and provide real-time visualization. This study focuses on the development of a POC-PET system and integrating ultrasound (US) to augment its utility for POC diagnostics.
Methods: Our prototype POC-PET system comprises a robotic arm (Kuka KR 10 R1100 sixx), a rotation stage and 8 PET detector modules (2 front panels controlled by the robotic arm and 6 rear panels fixed on the rotation stage) all mounted on a trolley (Fig. 1A). Each detector (Fig. 1B) contains 2 LSO (Lutetium Oxyorthosilicate) arrays (each with 8x8 elements of 3x3x10mm crystal each), 2 MPPC (Multi-Pixel Photon Counter) arrays (each with 8x8 elements of 3x3mm sensing area), and front-end-electronics (PETsys TOFPET2 ASIC). An optional US transducer can be sandwiched between the 2 front PET detectors to form a PET/US probe (Fig. 1A) for simultaneous PET/US imaging. The benefit of using a robotic arm is the freedom with the positions of the PET/US probe (both in terms of physical location and number of positions). The POC-PET system acquires data in list mode; sorts coincidence events containing crystal ID, timing, and energy information; and reconstructs PET images using a GPU-based list-mode MLEM (Maximal-Likelihood Expectation-Maximization) reconstruction framework.
To evaluate the imaging capability, we scanned a tumor phantom (Fig. 1C Right) containing 6 groups of spherical lesions of different diameters (ranging from 3.3 - 11.4mm) following the Derenzo patterns, with a tumor-to-background activity concentration ratio of 20:1. Reference PET/CT images were acquired using a Siemens Biograph Vision PET/CT.
We also imaged a silicone phantom containing 64Cu solution using the POC-PET/US probe (Fig. 1D) and the Biograph Vision to compare the PET/US images with the reference PET/CT.
Results: Fig. 1C Row #1 shows the 6 positions of the front and rear PET detector panels when imaging a phantom. The cumulated list-mode data was reconstructed using the cumulated sensitivity images (Fig. 1C Row #2) to obtain the PET images in Fig. 1C Row #3. The PET images improve progressively as more data is acquired. With the visual feedback, an operator may opt to halt the acquisition after the first 3-4 angles when the image quality starts to approach that of a clinical PET/CT image (Fig. 1C Right Column).
Similarly, the PET and US images from our prototype provide structural and molecular images similar to those from a clinical PET/CT scanner (Fig. 1D).
Conclusions: The proposed POC-PET technology and the integration with US represent a paradigm shift in dual-modal imaging technology that offers complementary diagnostic information at the patient's bedside. Combining the flexibility of a robotic arm with high-performance PET detectors, we have demonstrated the feasibility of high-quality PET and US images at bedside, achieving image resolution comparable to clinical PET/CT scanners. In the future, we aim to augment the system with real-time image reconstruction and fusion with US to enable interactive PET/US imaging.
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.