Validity and value of quantitative spect reconstruction in a multi-center clinical study using I-123 radiopharmaceuticals

Introduction: SPECT has a potential to provide parametric functional images in a "quantitative" manner, as has been widely done with PET in vivo, using several tracers typically labeled with I-123. Due to availability of equipments, as well as well-established delivery/transportation of radio-ligands, SPECT has an advantage in applying it to large-scale clinical studies. It has however been considered that accuracy and inter-institutional reproducibility of SPECT are limited for quantitative assessment of functional parametric images, largely attributed to a lack of general consensus of reconstruction procedures (attenuation/scatter correction) as well as, in case of I-123, significant amount of penetration through collimator from high-energy photons. We have recently developed a novel method to reconstruct SPECT images from existing projection data including appropriate corrections for scatter and attenuation in the object, head contour detection, as well as for collimator-dependent penetration. This study was aimed at evaluating validity and impact of our approach for Multicenter clinical study. We particularly investigated consistency and reproducibility of quantitative values for I-123, at different SPECT systems from different manufacturers. Materials and methods: The Windows program can handle original projection data to analyze and/or DICOM format obtained from 5 manufactures with essentially 7 different operation systems. Twenty-five institutions provided a series of phantom data for 28 collimators according to a given protocol. Nine are symmetric fanbeam, and rest parallel beam collimators. Experiments were done for a radioactivitycalibrated point source of I-123 in air to determine the absolute sensitivity of the system and to determine the penetration factor for each collimator. Additional experiments were done using a uniform cylindrical phantom filled with the radioactivity mentioned above, and single/multiple rod sources placed in cylindrical water. The scatter/penetration correction was based on the previously proposed transmission dependent convolution subtraction method, and the attenuation correction was implemented during the ordered-subset EM reconstruction. Dynamic SPECT studies were carried out in 9 institutions on 64 patients using I-123 iodo-amphetamine (IMP) to assess quantitative CBF images at rest and during Diamox challenge. Inter- and intra-institutional reproducibility was then evaluated. Results and discussion: Scatter fraction was consistent among different collimators from different manufacturers as predicted, but the penetration varied largely dependent on the collimator design and manufacturers. Our approach successfully compensated for the penetration and the scatter as well as the attenuation, as demonstrated in reconstructed images of the water-filled single/multiple rod source phantoms, except for one fanbeam collimator. Parallel beam collimator tends to have small penetration than fanbeam. Absolute concentration of the uniform cylindrical phantom was also consistent among institutions, although it was significantly smaller than the true value (0.88+/-0.05). Clinical data also demonstrated that CBF values obtained with I-123 IMP at rest and during Diamox were reproducible among institutions. Normal CBF values obtained in 2 institutions showed also no significant difference. Conclusion: These data suggested that QSPECT could be a useful tool for quantitative mapping using I-123 compounds, allowing a large scale clinical evaluation using SPECT.