Data acquisition and analysis: the strength of methodology in nuclear medicine and molecular imaging

Abstract

Dear Sir, We read with great interest the paper “Data acquisition and analysis: the strength of methodology in nuclear medicine and molecular imaging” by G. Lucignani [1]. The author nicely demonstrates the need for image quantification of positron emission tomography (PET) studies. Typically a PET acquisition consists of an emission scan and a transmission scan. Iterative reconstruction methods yield a better result compared with the filtered back projection method, especially in terms of noise. However, iterative reconstruction methods have a tendency to give rise to noise artefacts when a large number of iterations are performed. Image quality also depends greatly on how correction methods, such as attenuation and scatter correction, are applied, because these methods may introduce additional statistical noise into the reconstructed data sets. The complexity of different reconstruction methods, i.e. different correction methods, and the variation in scanner characteristics make it very difficult to compare the results between institutions. The effect of reconstruction on NEMA contrast values was demonstrated recently [2], and an interlaboratory comparison study of image quality using the NEMA NU 2-2001 procedure for assessment of image quality showed significant differences between different PET scanners [3]. The systematic difference in contrast entails two practical problems: visual image interpretation of scans may become more and more difficult and a systematic difference in measured SUVs may affect the diagnostic accuracy to be compared inter-institutionally. Using a common whole-body PET study and differing OSEM reconstruction parameters (number of iterations and subsets) of the emission data sets, we found no significant differences in the maximum SUV, but there were differences of up to 30% in the homogeneity of the liver. Of course, the SUV has an important clinical significance, but the homogeneity of a study is important for the clinician as well. Changing the reconstruction parameters of transmission data sets did not have any significant impact, because we used a segmented attenuation correction method. This has also been found to reduce the metal artefacts in PET-CT studies [4]. If the mathematical algorithm fails to perform an accurate attenuation correction, the image quality is degraded. Applying a phantom study by varying the transmission acquisition times (600, 300, 200, 100 or 20 s) and using Ge rods for attenuation correction, we observed sudden changes of up to 23% in homogeneity, but very minor differences in maximum SUV. Our preliminary findings show that in order to achieve comparable results of clinical studies from different institutions, the specification of both emission and transmission acquisition parameters has to be as detailed as possible. Eur J Nucl Med Mol Imaging (2007) 34:961–962 DOI 10.1007/s00259-007-0399-0