Initial human study of virtual-pinhole PET technology for head and neck cancer imaging

Objectives We have developed a prototype virtual-pinhole PET (VP-PET) insert device and integrated it into a Siemens Biograph-40 PET-CT scanner to provide high-resolution (2-3 mm FWHM) human imaging capability within a reduced (24cm diameter by 6cm axial) imaging field-of-view (FOV). We evaluated the performance and limitations of the prototype for head-and-neck cancer imaging through initial human study.

Methods We recruited patients who have histologically confirmed squamous cell carcinoma of the head and neck with the primary tumor in the oropharynx or nasopharynx with suspected lymph node metastasis. The research protocol includes the following steps: (1) inject 259-370 MBq of FDG to the subject and wait 45-60 minutes for uptake; (2) image the head-and-neck region with the standard brain imaging protocol (no VP-PET insert, single-bed, 10-minute acquisition); reconstruct using standard Siemens software with maximal matrix size and minimal smoothing; (3) move the VP-PET insert into the FOV, image the same region in 2 bed-positions of 10-minute each; reconstruct using algorithms developed specifically for VP-PET systems, stitch two bed-positions into one image volume; (4) compare 2 datasets.

Results Images acquired by the VP-PET prototype provide higher tumor-to-background contrast and better delineation of the tumor-tissue boundaries. The half-ring geometry limits the use of the prototype for imaging tumors in the lower neck region when the gantry hits the shoulders of the subject. Additional patient volunteers are being recruited to increase the sample size.

Conclusions We have demonstrated that VP-PET insert technology can improve the resolution of a clinical PET-CT scanner, which can enhance tumor contrast and (potentially) detectability for head-and-neck cancer imaging. New geometry for the VP-PET insert devices is required in order to make such technology more versatile and clinically usable.

Research Support This work was supported in part by the National Cancer Institute (R01-CA136554, R33-CA110011, and P30-CA91842), the Siteman Cancer Center PSRS Award, the Washington University Center for High Performance Computing (funded by NIH grant NCRR 1S10RR022984-01A1), and the Mallinckrodt Institute of Radiology Internal Allotment (MIR 11-014)