Validation of fast-RAMLA in clinical PET

Images using the fast row action maximum likelihood algorithm (fast-RAMLA), which employs half-interpolated sinograms of conventional 3DRAMLA, are immediately generated following positron emission tomography (PET) scanning and are invariably produced in the process of line-of-response RAMLA (LOR-RAMLA) reconstruction. We quantitatively and visually compared the clinical validity of dual time point [(18)F]-FDG imaging with fast-RAMLA and LOR-RAMLA.

An International Electrotechnical Commission (IEC) phantom was established in which the ratio of the activities in the hot sphere was set up and a background of 3.8:1 was scanned and reconstructed using both algorithms. The contrast recovery coefficient was then calculated. The clinical study retrospectively analyzed 35 patients (25 men and 10 women; age range 30-84 years; mean age 63.9 years) with confirmed specific pathological lesions or clinical follow-up; 21 of the patients had 51 malignant lesions and 15 had 23 benign lesions. The maximum standard uptake value (SUV(max)) was measured in all lesions using LOR-RAMLA. The maximal counts of all lesions determined manually were divided by the average count of bilateral ventricles and the aortic arch for standardization on fast-RAMLA, and the values were compared with the SUV(max) of LORRAMLA. Inter-observer variation in detection was determined among three radiologists who blindly reviewed and scored 70 maximum intensity projection images from 35 patients reconstructed using LORRAMLA and fast-RAMLA.

We identified a quantitative correlation and determined the visual quality of lesion detection between fast-RAMLA and LOR-RAMLA and indicated usefulness and improvement point on fast-RAMLA.

Fast-RAMLA can improve the strategy for using dual time point [(18)F] fluorodeoxyglucose positron emission tomography ([(18)F]-FDG-PET) and increase the efficiency of the [(18)F]-FDG-PET scanner.