TOF PET uses high-temporal-resolution detectors that enable annihilation events to be localized to within a 10 cm segment between the two detectors intercepting a photon pair. The enhanced detector technology produces substantial improvements in signal-to-noise ratio (SNR) over conventional PET imaging.

The researchers, from Harvard Medical School (Boston, MA) and the University of Pennsylvania (Philadelphia, PA), used 100 patient scans to quantitate improvements in lesion detection associated with TOF imaging. In particular, they examined the influence of lesion location, lesion contrast, body mass index (BMI) and scanning time.

To provide unequivocal knowledge of lesion presence and location, the researchers used a series of artificial lesions. Ten millimetre diameter plastic spheres filled with 5–50 MBq of radiopharmaceutical tracer were imaged in isolation, on the scanning couch, at known locations in the scanner co-ordinate system. The researchers fused the lesion images with patient data sets, creating patient data sets that contained artificial lesions of known size and location. Attenuation correction was applied to the lesion image prior to fusion with the patient data, mimicking the attenuation of counts exhibited by an in situ lesion. Lesion locations in the lungs and liver provided high and low tracer uptake backgrounds respectively.

Analysis by human observers of the 36,000 images generated by the study was unfeasible. Instead, an automated "model" Channelized Hotelling Observer (CHO) was used to evaluate the lesion detection SNRs; a greater SNR indicating superior lesion detection.

While TOF consistently outperformed standard PET, there were scenarios where TOF showed especially high gains. These included the detection of lesions exhibiting lower contrast: TOF produced an SNR 20.3% greater than that for standard PET for low lesion contrast (2.0:1), compared to only 7.5% greater for high lesion contrast (5.7:1). Improvement in lesion detection with TOF was also particularly marked for larger individuals, a patient subgroup associated with suboptimal image quality. The greatest gains were obtained for patients with a BMI of more than 30, where the TOF SNR was 11.1% greater than that for standard PET.

The researchers report that the advantages of TOF PET can be exploited in three main ways. Firstly, the superior image quality can be exploited to detect lesions which may not be visible using conventional PET. Secondly, image quality can be traded off for reduced radiopharmaceutical administration and an associated reduction in patients and staff exposure to ionising radiation. Thirdly, image quality can be traded off for reduced scan times, improving patient comfort and reducing motion artefacts.

The latter strategy has been applied clinically by the authors. "Patient scan times are already being adjusted based on 2–3 groups of BMI. At our institution, we routinely image patients for 1.5, 2 and 3 minutes per bed position based on three categories of patient BMI. This study reinforces the need for efficient patient scanning without sacrificing clinical image quality," explained Joel Karp, professor of radiologic physics in radiology at Penn Medicine.

What next for the researchers? Karp told medicalphysicsweb: "The current study demonstrated the benefit of TOF imaging in lesion detection using numerical observers. Since that work, we decided to get one step closer to a real clinical imaging environment where a nuclear medicine clinician reads a patient image and tries to detect a lesion at an unknown location. The results from this follow-up work show improved lesion detection, as well as localization capability with TOF. The study, which was led by Suleman Surti, will appear in a Journal of Nuclear Medicine publication shortly."