The effect of Butterworth and Metz reconstruction filters on volume and ejection fraction calculations with 99Tcm gated myocardial SPECT

Pix2Pix generative adversarial network for low dose myocardial perfusion SPECT denoising

BACKGROUND

Myocardial perfusion (MP) SPECT is a well-established method for diagnosing cardiac disease, yet its radiation risk poses safety concern. This study aims to apply and evaluate the use of Pix2Pix generative adversarial network (Pix2Pix GAN) in denoising low dose MP SPECT images.

METHODS

One hundred male and female patients with different 99mTc-sestamibi activity distributions, organ and body sizes were simulated by a population of digital 4D Extended Cardiac Torso (XCAT) phantoms. Realistic noisy SPECT projections of full dose of 987 MBq injection and 16 min acquisition, and low dose ranged from 1/20 to 1/2 of the full dose, were generated by an analytical projector from the right anterior oblique (RAO) to the left posterior oblique (LPO) positions. Additionally, twenty patients underwent ~1,184 MBq 99mTc-sestamibi stress SPECT/CT scan were also retrospectively recruited for the study. For each patient, low dose SPECT images (7/10 to 1/10 of full dose) were generated from the full dose list mode data. Our Pix2Pix GAN model was trained with full dose and low dose reconstructed SPECT image pairs. Normalized mean square error (NMSE), structural similarity index (SSIM), coefficient of variation (CV), full-width-at-half-maximum (FWHM) and relative defect size differences (RSD) of Pix2Pix GAN processed images were evaluated along with a reference convolutional auto encoder (CAE) network and post-reconstruction filters.

RESULTS

NMSE values of 0.0233±0.004 vs. 0.0249±0.004 and 0.0313±0.007 vs. 0.0579±0.016 were obtained on 1/2 and 1/20 dose level for Pix2Pix GAN and CAE in the simulation study, while they were 0.0376±0.010 vs. 0.0433±0.010 and 0.0907±0.020 vs. 0.1186±0.025 on 7/10 and 1/10 dose level in the clinical study. Similar results were also obtained from the SSIM, CV, FWHM and RSD values. Overall, the use of Pix2Pix GAN was superior to other denoising methods in all physical indices, particular in the lower dose levels in the simulation and clinical study.

CONCLUSIONS

The Pix2Pix GAN method is effective to reduce the noise level of low dose MP SPECT. Further studies on clinical performance are warranted to demonstrate its full clinical effectiveness.

Effects of filtration on right ventricular function by the gated blood pool SPECT

OBJECTIVE

Gated blood po ol single photon emission computed tomography (GBPS) offers the possibility of obtaining additional functional information from blood pool studies, including evaluation of left and right ventricular function simultaneously. The calculation of ventricular volumes based on the identification of the endocardial surface would be influenced by the spatial resolution in the reconstructed images. This study was performed to evaluate the effect of different filters on the right ventricular function.

METHODS

The normal four-dimensional (4-D) NURBS-based cardiac-torso (NCAT) phantom with known right ventricular volume and ejection fraction was generated. The SIMIND Monte Carlo program was used to create projections. The studies were reconstructed by FBP and post-processing filtration such as Butterworth, Hanning, Shepp-Logan, Metz and Wiener in different statuses (cutoff and order). Using the Cedars-Sinai QBS (quantitative blood pool SPECT) package, the ventricular functional parameters were computed. The calculated values were analyzed and compared with the normal NCAT results.

RESULTS

The results implied that the calculated right ventricular end diastolic volume (RVEDV) by Butterworth filtration (cutoff frequency = 0.3) agreed more with the NCAT Phantom characteristics [relative difference percentage (RDP) = 1.2 %], while the maximum accordance in the calculation of the RV ejection fraction (EF) (RDP = 3 %) was observed by Metz filter (FWHM 20 pixel). Also, the results of this study demonstrate that the Butterworth filter provided the most stable values (cutoff frequency = 0.4-0.5) in the estimation of RVEDV (RDP = 7.5 %). The Hanning and Shepp-Logan filters produced a much larger RDP, particularly in low frequency (41.1 and 21.5 %, respectively) compared to other filters.

CONCLUSIONS

This study demonstrated that the operation of different filters has a severe effect in computing right ventricular volume. The resolution recovery and Butterworth filters tend to give more comparable ventricular volumes with the actual normal NCAT value. Further evaluation using a large clinical database is underway to evaluate the optimum protocol in a clinical setting.

Filters in 2D and 3D Cardiac SPECT Image Processing

Nuclear cardiac imaging is a noninvasive, sensitive method providing information on cardiac structure and physiology. Single photon emission tomography (SPECT) evaluates myocardial perfusion, viability, and function and is widely used in clinical routine. The quality of the tomographic image is a key for accurate diagnosis. Image filtering, a mathematical processing, compensates for loss of detail in an image while reducing image noise, and it can improve the image resolution and limit the degradation of the image. SPECT images are then reconstructed, either by filter back projection (FBP) analytical technique or iteratively, by algebraic methods. The aim of this study is to review filters in cardiac 2D, 3D, and 4D SPECT applications and how these affect the image quality mirroring the diagnostic accuracy of SPECT images. Several filters, including the Hanning, Butterworth, and Parzen filters, were evaluated in combination with the two reconstruction methods as well as with a specified MatLab program. Results showed that for both 3D and 4D cardiac SPECT the Butterworth filter, for different critical frequencies and orders, produced the best results. Between the two reconstruction methods, the iterative one might be more appropriate for cardiac SPECT, since it improves lesion detectability due to the significant improvement of image contrast.

UK audit of quantification of left-ventricular function using gated myocardial perfusion imaging

PURPOSE

The aim of the study was to evaluate UK-wide interinstitutional reproducibility of left-ventricular functional parameters, end-systolic volume, end-diastolic volume and ejection fraction, obtained from gated myocardial perfusion imaging (GMPI) studies using technetium-99m-labelled radiopharmaceuticals. The study was carried out by the UK Institute of Physics and Engineering in Medicine Nuclear Medicine Software Quality Group.

MATERIALS AND METHODS

Ten anonymized clinical GMPI studies, five with normal perfusion and five with perfusion defects, were made available in DICOM and proprietary formats for download and through manufacturers' representatives. Two of the studies were duplicated in order to assess intraoperator repeatability, giving a total of 12 studies. Studies were made available in 8 and 16 frames/cycle.

RESULTS

A total of 58 institutions across England, Scotland, Wales and Northern Ireland participated in this study using six different computer packages. Studies were processed at centres using their normal clinical computers and software. The overall mean±SD ejection fraction for all centres was 58.5±3%; the mean end-diastolic volume was 114±12 ml and the mean end-systolic volume was 54±6 ml. The results were affected by the number of frames per cycle and by the postprocessing computer package, but not by the reconstruction filter in the filtered back-projection.

CONCLUSION

Calculation of functional parameters from GMPI using technetium-99m-labelled radiopharmaceuticals is reliable and shows limited variability across the UK.

Late gadolinium uptake demonstrated with magnetic resonance in patients where automated PERFIT analysis of myocardial SPECT suggests irreversible perfusion defect

BACKGROUND

Myocardial perfusion single photon emission computed tomography (MPS) is frequently used as the reference method for the determination of myocardial infarct size. PERFIT(R) is a software utilizing a three-dimensional gender specific, averaged heart model for the automatic evaluation of myocardial perfusion. The purpose of this study was to compare the perfusion defect size on MPS, assessed with PERFIT, with the hyperenhanced volume assessed by late gadolinium enhancement magnetic resonance imaging (LGE) and to relate their effect on the wall motion score index (WMSI) assessed with cine magnetic resonance imaging (cine-MRI) and echocardiography (echo).

METHODS

LGE was performed in 40 patients where clinical MPS showed an irreversible uptake reduction suggesting a myocardial scar. Infarct volume, extent and major coronary supply were compared between MPS and LGE as well as the relationship between infarct size from both methods and WMSI.

RESULTS

MPS showed a slightly larger infarct volume than LGE (MPS 29.6 +/- 23.2 ml, LGE 22.1 +/- 16.9 ml, p = 0.01), while no significant difference was found in infarct extent (MPS 11.7 +/- 9.4%, LGE 13.0 +/- 9.6%). The correlation coefficients between methods in respect to infarct size and infarct extent were 0.71 and 0.63 respectively. WMSI determined with cine-MRI correlated moderately with infarct volume and infarct extent (cine-MRI vs MPS volume r = 0.71, extent r = 0.71, cine-MRI vs LGE volume r = 0.62, extent r = 0.60). Similar results were achieved when wall motion was determined with echo. Both MPS and LGE showed the same major coronary supply to the infarct area in a majority of patients, Kappa = 0.84.

CONCLUSION

MPS and LGE agree moderately in the determination of infarct size in both absolute and relative terms, although infarct volume is slightly larger with MPS. The correlation between WMSI and infarct size is moderate.