Optimisation of quantitative lung SPECT applied to mild COPD: a software phantom simulation study

Impacts of different reconstruction methods on the image quality of cadmium-zinc-telluride-based single photon emission computed tomography/computed tomography pulmonary perfusion imaging

OBJECTIVE

The objective was to evaluate the impacts of different reconstruction methods [filtered back projection (FBP) and ordered subset expectation maximization (OSEM)] and different filters (Butterworth filter and Gaussian filter) on the image quality in cadmium-zinc-telluride (CZT)-based single photon emission computed tomography (SPECT)/computed tomography (CT) pulmonary perfusion imaging.

METHODS

A combinations including FBP with Butterworth filter, OSEM with Butterworth filter (OSEM + Butterworth filter ), and OSEM with Gaussian filter (OSEM + Gaussian filter) were used during SPECT image reconstruction. Visual and quantitative parameters [root mean square (RMS) noise, contrast and contrast-to-noise ratio (CNR)] were used to evaluate image quality.

RESULTS

The OSEM + Gaussian filter had better RMS noise and CNR than those of the FBP + Butterworth filter or OSEM + Butterworth filter, while the OSEM + Butterworth filter had the best contrast. The highest visual scores were obtained by OSEM + Gaussian filter ( P  < 0.0001). In the lesion size <2 cm group, the contrast ( P < 0.01) and visual scores ( P < 0.001) of OSEM + Butterworth filter were better than those of the other two groups. In the lesion size ≥2 cm group, the RMS noise and visual scores of OSEM + Gaussian filter were better than those of the other two groups.

CONCLUSION

In CZT SPECT/CT pulmonary perfusion imaging, this study recommended the clinical use of the OSEM + Gaussian filter combination for reconstruction in both conventional and larger lesions, the OSEM + Butterworth filter image postprocessing method might be advantageous in small lesions.

Normal Dual Isotope V/Q SPECT Model for Monte-Carlo Studies

Background: There is currently no reliable or validated tool to delineate and quantify functional lung volumes with ventilation/perfusion (V/Q) SPECT/CT. The main challenges encountered include the physiological non-uniformity of lung function, such as the anterior-to-posterior gradient on perfusion images, and the lack of ground truth to assess the accuracy of delineation algorithms. In that respect, Monte-Carlo simulations would be an interesting tool. Thus, the aim of this study was to develop a realistic model of dual-isotope lung V/Q SPECT-CT Monte-Carlo simulations, integrating the anterior to posterior gradient on perfusion.

Methods: Acquisitions and simulations parameters were set in accordance to nuclear medicine guidelines for V/Q lung SPECT-CT. Projections were acquired and simulated, then the reconstructions [with and without attenuation correction (AC)] were compared. A model was built from a patient's CT scan. To model the anterior to posterior gradient, the lungs were divided into sixteen coronal planes, where a rising radioactivity concentration was set. To assess the realism of simulations, they were compared to a normal co-registered normal cases database in terms of pixelwize Z-score map.

Results: For ventilation images, mean (SD) Zscores on Zscore maps were −0.2 (0.7) and −0.2 (0.7) for AC and noAC images, respectively. For perfusion images, mean (SD) Zscores were −0.2 (0.6) and −0.1 (0.6) for AC and noAC images, respectively.

Conclusion: We developed a model for dual isotopes lung V/Q SPECT-CT, integrating the anterior-to-posterior gradient on perfusion images. This model could be used to build a catalog of clinical scenarios, in order to test delineation methods of functional lung volumes.