Strategy of SPECT reconstruction for absolute quantitation of cerebral function image

Introduction

Recent advance in SPECT could incorporate accurate corrections for attenuation and scatter. Despite various equipment conditions of SPECT, such as different collimator design and/or energy window settings, a novel SPECT reconstruction package (QSPECT) may provide quantitative images of in vivo distribution of radio-tracers in clinical studies. In this work, we have evaluated accuracy and reproducibility of reconstructed images among different scanners from different manufacturers installed in different institutions.

Methods

Six institutions were involved in this study using one of the following 4 SPECT cameras, namely E. CAM dual-headed from Toshiba-Siemens (n=1), PRISM IRIX triple-headed from Shimazu (n=1), MillenniumVG dual-headed from GE (n=1), and GCA-9300A triple-headed from Toshiba (n=3). Absolute sensitivity (Becquerel per mL in reconstructed images, or Becquerel Calibration Factor (BCF)) was calibrated using a syringe of I-123 solution supplied from a radio-pharmaceutical company (Nihon-Medi Physics, Tokyo, Japan). Uniform cylindrical phantom of known radioactivity concentration and multiple rod phantoms were scanned in each institution. Images were reconstructed using a software package, QSPECT, with a scatter correction by means of the transmission-dependent convolution subtraction technique, and OSEM procedures including the attenuation correction process. Cross calibration factor (CCF) that illustrates the relative sensitivity between external well-counter and SPECT, was estimated. Accuracy and the inter-institutional reproducibility were then evaluated for those reconstructed images. CBF study was performed with split-injection of I-123 IMP, on patients with cerebrovascular disease, and cerebral function was evaluated using the image obtained by this approach.

Results

BCF values were dependent on to collimator design and the number of detector. The three institutions that had the same SPECT camera (GCA-9300A) demonstrated a good agreement of the BCF value (COV of 2.6%). The pixel values in the reconstructed images, which should represent the absolute radioactivity concentration in units of Bq/mL, were in a good agreement with the true radioactivity concentration for the uniform cylindrical phantom, i.e. the difference was 10.5 ± 1.7% for all 6 institutions. CCF values that represent the sensitivity of well-counter and should be less than 1, were in a narrow range (from 0.5 to 0.8), which supported the robustness of this approach in radioactivity measurement. The pixel values of reconstructed IMP image showed about 8% of injected dose on the normal region, which was consistent with previous reports of IMP uptake in brain tissue.

Conclusion

The presented SPECT reconstruction could provide quantitative images that represent regional radioactivity concentration in units of Bq/mL. These quantitative values appeared to be consistent and reproducible among different hardware setup supplied from different manufacturers, which resulted in the improved evaluation of cerebral function. Selection of adequate methodologies for the accurate scatter and attenuation correction procedures was essential to appropriate measurement of radioactivity. Thus, this approach can be feasible for quantitation of physiological functions, as has been done only by PET, and may be useful in multi-center trials.