Detecting uptake changes with lesion size uncertainty: Comparison of SUV methods and reconstruction

Le résumé fourni par la source

336 Objectives We have previously shown that, when object size is known and fixed, SUVpeak gives higher confidence than SUVmax in detecting changes of uptake. For a more clinically realistic scenario, we measure uptake changes where object size is uncertain, as changes in tumour size during therapy are rarely known exactly. Methods Images of a NEMA image quality phantom with 8:1, 6:1 and 4:1 sphere contrasts acquired on a Siemens Biograph mCT were reconstructed with OSEM, PSF, TOF and PSF+TOF algorithms. 10 replicate images were produced for each phantom. Filters were chosen to align background image noise across reconstructions. SUVmax and SUVpeak were measured for each sphere. To model uncertainty in sizes, 10 independent random samples of 10 measurements from adjacent sphere sizes were taken from each contrast phantom. Unpaired t-tests were used to assess separability of sphere contrasts. Lower p values indicating improved separability, hence confidence in detecting changes in uptake. Results SUVpeak provided better confidence than SUVmax for 22 to 37mm spheres (mean p values: 2.5x10-10 and 4.9 x10-5 for SUVpeak and SUVmax respectively). SUVmax provided better confidence for 10 to 17mm spheres (0.04 and 0.01 for SUVpeak and SUVmax respectively), suggesting a transition around 20mm. This is due to the stronger dependence of SUVpeak recovery on size for the small spheres. Findings were consistent across all reconstructions. Confidence was only improved with TOF and PSF+TOF reconstruction compared with OSEM due to greater variance on individual measurements with PSF. Conclusions Applying these findings to clinical practice suggests that SUVmax is more suitable than SUVpeak for detecting uptake changes for tumours smaller than 20mm, reinforcing PERCIST guidelines that state SUVpeak should only be used on tumours above 20mm. Furthermore, reconstructions with TOF can further increase confidence by reducing variability on individual uptake measures.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.