Observer Studies of Image Quality of Denoising Reduced-count Cardiac Single Photon Emission Computed Tomography Myocardial Perfusion Imaging by Three-dimensional Gaussian Post-reconstruction Filtering and Deep Learning

Overview

Journal J Nucl Cardiol

Specialty Cardiology & Vascular Diseases

Date 2023 May 23

PMID 37221409

Authors

P Hendrik Pretorius

Junchi Liu

Kesava S Kalluri

Yulei Jiang

Jeffery A Leppo

Seth T Dahlberg

Janusz Kikut

Matthew W Parker

Friederike K Keating

Robert Licho

Benjamin Auer

Clifford Lindsay

Arda Konik

Yongyi Yang

Miles N Wernick

Michael A King

Affiliations

Soon will be listed here.

Abstract

Background: The aim of this research was to asses perfusion-defect detection-accuracy by human observers as a function of reduced-counts for 3D Gaussian post-reconstruction filtering vs deep learning (DL) denoising to determine if there was improved performance with DL.

Methods: SPECT projection data of 156 normally interpreted patients were used for these studies. Half were altered to include hybrid perfusion defects with defect presence and location known. Ordered-subset expectation-maximization (OSEM) reconstruction was employed with the optional correction of attenuation (AC) and scatter (SC) in addition to distance-dependent resolution (RC). Count levels varied from full-counts (100%) to 6.25% of full-counts. The denoising strategies were previously optimized for defect detection using total perfusion deficit (TPD). Four medical physicist (PhD) and six physician (MD) observers rated the slices using a graphical user interface. Observer ratings were analyzed using the LABMRMC multi-reader, multi-case receiver-operating-characteristic (ROC) software to calculate and compare statistically the area-under-the-ROC-curves (AUCs).

Results: For the same count-level no statistically significant increase in AUCs for DL over Gaussian denoising was determined when counts were reduced to either the 25% or 12.5% of full-counts. The average AUC for full-count OSEM with solely RC and Gaussian filtering was lower than for the strategies with AC and SC, except for a reduction to 6.25% of full-counts, thus verifying the utility of employing AC and SC with RC.

Conclusion: We did not find any indication that at the dose levels investigated and with the DL network employed, that DL denoising was superior in AUC to optimized 3D post-reconstruction Gaussian filtering.

Citing Articles

Deep learning-based multi-frequency denoising for myocardial perfusion SPECT.

Du Y, Sun J, Li C, Yang B, Wu T, Mok G EJNMMI Phys. 2024; 11(1):80.

PMID: 39356406 PMC: 11447183. DOI: 10.1186/s40658-024-00680-w.

DEMIST: A Deep-Learning-Based Detection-Task-Specific Denoising Approach for Myocardial Perfusion SPECT.

Rahman M, Yu Z, Laforest R, Abbey C, Siegel B, Jha A IEEE Trans Radiat Plasma Med Sci. 2024; 8(4):439-450.

PMID: 38766558 PMC: 11101197. DOI: 10.1109/trpms.2024.3379215.

Need for Objective Task-Based Evaluation of Image Segmentation Algorithms for Quantitative PET: A Study with ACRIN 6668/RTOG 0235 Multicenter Clinical Trial Data.

Liu Z, Mhlanga J, Xia H, Siegel B, Jha A J Nucl Med. 2024; 65(3).

PMID: 38360049 PMC: 10924158. DOI: 10.2967/jnumed.123.266018.

References

METZ , Pan . "Proper" Binormal ROC Curves: Theory and Maximum-Likelihood Estimation. J Math Psychol. 1999; 43(1):1-33. DOI: 10.1006/jmps.1998.1218. View

Bonardel G, Dupont A, Decazes P, Queneau M, Modzelewski R, Coulot J . Clinical and phantom validation of a deep learning based denoising algorithm for F-18-FDG PET images from lower detection counting in comparison with the standard acquisition. EJNMMI Phys. 2022; 9(1):36. PMC: 9095795. DOI: 10.1186/s40658-022-00465-z. View

Henzlova M, Duvall W, Einstein A, Travin M, Verberne H . ASNC imaging guidelines for SPECT nuclear cardiology procedures: Stress, protocols, and tracers. J Nucl Cardiol. 2016; 23(3):606-39. DOI: 10.1007/s12350-015-0387-x. View

Liu J, Yang Y, Wernick M, Pretorius P, King M . Deep learning with noise-to-noise training for denoising in SPECT myocardial perfusion imaging. Med Phys. 2020; 48(1):156-168. PMC: 7902415. DOI: 10.1002/mp.14577. View

Metz C, Herman B, Shen J . Maximum likelihood estimation of receiver operating characteristic (ROC) curves from continuously-distributed data. Stat Med. 1998; 17(9):1033-53. DOI: 10.1002/(sici)1097-0258(19980515)17:9<1033::aid-sim784>3.0.co;2-z. View

Pagnanelli R, Borges-Neto S . Technical aspects of resolution recovery reconstruction. J Nucl Cardiol. 2015; 23(1):149-52. DOI: 10.1007/s12350-015-0345-7. View

Zoccarato O, Scabbio C, De Ponti E, Matheoud R, Leva L, Morzenti S . Comparative analysis of iterative reconstruction algorithms with resolution recovery for cardiac SPECT studies. A multi-center phantom study. J Nucl Cardiol. 2013; 21(1):135-48. DOI: 10.1007/s12350-013-9821-0. View

Shanbhag A, Miller R, Pieszko K, Lemley M, Kavanagh P, Feher A . Deep Learning-Based Attenuation Correction Improves Diagnostic Accuracy of Cardiac SPECT. J Nucl Med. 2022; 64(3):472-478. PMC: 10071806. DOI: 10.2967/jnumed.122.264429. View

Hudson H, Larkin R . Accelerated image reconstruction using ordered subsets of projection data. IEEE Trans Med Imaging. 1994; 13(4):601-9. DOI: 10.1109/42.363108. View

10.

Svane B, Bone D, Holmgren A . Coronary angiography and thallium-201 single photon emission computed tomography in single vessel coronary artery disease. Acta Radiol. 1990; 31(3):237-44. View

11.

Alskaf E, Dutta U, Scannell C, Chiribiri A . Deep learning applications in myocardial perfusion imaging, a systematic review and meta-analysis. Inform Med Unlocked. 2022; 32:101055. PMC: 9514037. DOI: 10.1016/j.imu.2022.101055. View

12.

Jha A, Myers K, Obuchowski N, Liu Z, Rahman M, Saboury B . Objective Task-Based Evaluation of Artificial Intelligence-Based Medical Imaging Methods:: Framework, Strategies, and Role of the Physician. PET Clin. 2021; 16(4):493-511. DOI: 10.1016/j.cpet.2021.06.013. View

13.

Juan Ramon A, Yang Y, Pretorius P, Johnson K, King M, Wernick M . Improving Diagnostic Accuracy in Low-Dose SPECT Myocardial Perfusion Imaging With Convolutional Denoising Networks. IEEE Trans Med Imaging. 2020; 39(9):2893-2903. PMC: 9472754. DOI: 10.1109/TMI.2020.2979940. View

14.

Duvall W, Slomka P, Gerlach J, Sweeny J, Baber U, Croft L . High-efficiency SPECT MPI: comparison of automated quantification, visual interpretation, and coronary angiography. J Nucl Cardiol. 2013; 20(5):763-73. PMC: 3820488. DOI: 10.1007/s12350-013-9735-x. View

15.

Jha A, Bradshaw T, Buvat I, Hatt M, Kc P, Liu C . Nuclear Medicine and Artificial Intelligence: Best Practices for Evaluation (the RELAINCE Guidelines). J Nucl Med. 2022; 63(9):1288-1299. PMC: 9454473. DOI: 10.2967/jnumed.121.263239. View

16.

Narayanan M, King M, Pretorius P, Dahlberg S, Spencer F, Simon E . Human-observer receiver-operating-characteristic evaluation of attenuation, scatter, and resolution compensation strategies for (99m)Tc myocardial perfusion imaging. J Nucl Med. 2003; 44(11):1725-34. View

17.

Metz C . Basic principles of ROC analysis. Semin Nucl Med. 1978; 8(4):283-98. DOI: 10.1016/s0001-2998(78)80014-2. View

18.

Slomka P, Nishina H, Berman D, Akincioglu C, Abidov A, Friedman J . Automated quantification of myocardial perfusion SPECT using simplified normal limits. J Nucl Cardiol. 2005; 12(1):66-77. DOI: 10.1016/j.nuclcard.2004.10.006. View

19.

Pretorius P, King M, Gifford H, Dahlberg S, Spencer F, Simon E . Myocardial perfusion SPECT reconstruction: receiver operating characteristic comparison of CAD detection accuracy of filtered backprojection reconstruction with all of the clinical imaging information available to readers and solely stress slices.... J Nucl Cardiol. 2005; 12(3):284-93. DOI: 10.1016/j.nuclcard.2005.01.012. View

20.

Juan Ramon A, Yang Y, Pretorius P, Johnson K, King M, Wernick M . Personalized Models for Injected Activity Levels in SPECT Myocardial Perfusion Imaging. IEEE Trans Med Imaging. 2018; 38(6):1466-1476. PMC: 6582653. DOI: 10.1109/TMI.2018.2885319. View