Impact of pre-reconstruction filtering with butterworth filter on 111In-pentetreotide SPECT image quality and quantitative accuracy: A phantom study

Introduction: This study evaluates the image quality and quantitative accuracy of SPECT images with pre- and post-reconstruction smoothing filters in somatostatin receptor scintigraphy using phantom data.
Methods: We evaluated the spatial resolution, the contrast-to-noise ratio (CNR), and the quantitative accuracy using a NEMA IEC body phantom filled with a 111In solution. SPECT images were obtained with a Siemens Symbia T16 SPECT/CT system. Quantitative accuracy refers to the ability to accurately estimate the radioactive concentration of 111In in the phantom from the image. SPECT reconstructions were performed using three methods: post-reconstruction Gaussian filtering (post-G), pre-reconstruction Gaussian filtering (pre-G), and pre-reconstruction Butterworth filtering (pre-B). To verify each filtering method, the cut-off frequency of the Butterworth filter and the full width at half maximum (FWHM) of the Gaussian filter were each changed to eight different settings.
Results: FWHMs were 21.2, 19.8, and 18.0 mm for post-G, pre-G, and pre-B. CNRs (37-mm sphere) were 47.2, 63.8, and 69.5. Pre-B showed a 12.0 % error rate at 0.40 cycles/cm, while post-G and pre-G showed 20.2 % and 22.0 % at 7.2-mm FWHM. Pre-B outperformed other methods for resolution, CNR, and quantitative accuracy.
Conclusion: For 111In-pentetreotide SPECT images, image reconstruction with a Butterworth filter applied to the projection image before reconstruction was found to be superior to reconstruction with a Gaussian filter in terms of image quality and quantitative accuracy. This method can be easily implemented in routine clinical SPECT imaging workflows and has the potential to improve diagnostic confidence.
Implications for practice: The proposed method with a pre-reconstruction Butterworth filter has great potential to improve the image quality and quantitative accuracy of 111In-SPECT images.