Jun Zhang PhD, Presenter: Nothing to Disclose

Katherine Binzel BS, Abstract Co-Author: Nothing to Disclose

Xiaoli Liu, Abstract Co-Author: Nothing to Disclose

Wesley Thio, Abstract Co-Author: Nothing to Disclose

Nathan C. Hall MD, PhD, Abstract Co-Author: Consultant, Enlyton, Ltd

Michael Vinzenz Knopp MD, PhD, Abstract Co-Author: Nothing to Disclose

To demonstrate the feasibility of an SUV lookup table (LUT) based approach for estimating equivalent SUV and establishing comparable, normalized SUV readouts when patients are imaged on multiple PET/CT systems for clinical care

Identical 18F-FDG Flangeless Jaszczak PET phantoms with 6 hollow spheres in descending sizes (16, 8, 4, 2, 1 and 0.5mL) simulating lesions in ascending lesion-to-background ratios (LBR=2.2, 4.5, 9.0, 13.5, 22.5 and 31.5) were cross-acquired among current analog detector PET/CT systems (mCT (Siemens), Gemini TF 64 (Philips) ) and next generation Vereos Digital PET/CT (Philips). SUV and SBR were measured. An SUV LUT in relationship to sphere sizes, LBRs (1-100) and reconstruction methods with and without PSF between scanners was established. Subsequently, 146 malignant lesions from 60 patients were analyzed with equivalent SUVs estimated according to the established LUT

An initial 6 (lesion sizes) by 9 (LBRs) dimensional ‘GRID’ LUT of SUV conversion factors (CF) between systems was established based on phantom data assessment and subsequently extended to a larger dimensional LUT via data interpolations. Using the LUT, SUVs measured on one PET system were converted to equivalent SUVs on another PET system. CF values were different in combination with lesion size, LBR and PET recon protocols.  For instance it revealed SUV CFs (Gemini/mCT) from 0.43 to 0.82 for lesion sizes from 0.5 to 4ml and remained constant at 0.86 for lesion sizes ≥8ml under LBR of 29 using default PET recons. Veroes PSF PET showed the best results (5%±7%) for small lesions (≤4mL) with high LBRs (≥9) than others (-38%±22% to -10%±14%), and has the least RC needed. For the 146 tumor lesions (LBR of 3 to 83, lesion sizes of 0.25 to 66 ml) imaged on different systems, a variance of 12%-60% in therapy response was found between before & after SUV corrections using the LUT

The developed and demonstrated the feasibility of a LUT approach for SUV conversion between different PET/CT scanners to provide comparable, normalized SUV assessment in oncologic PET. Multiple factors influence SUV readouts such as lesion size, LBR and PET recon protocols, all appeared manageable by the LUT 

PET/CT scans of same patients performed on different scanners exist today in oncology reality causing inaccurate diagnosis. Such inconsistency can be minimized using the proposed LUT approach

Zhang, J, Binzel, K, Liu, X, Thio, W, Hall, N, Knopp, M, An Approach for Comparable SUV Quantification in Clinical Oncological PET/CT When Patient Acquisitions are Performed on Multiple Systems.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14015738.html