Compressed sensing reconstruction shortens the acquisition time for myocardial perfusion imaging: a simulation study

Area of Science:

• Nuclear Medicine
• Medical Imaging
• Image Reconstruction

Background:

• Compressed sensing (CS) enhances image quality in single-photon emission tomography (SPECT).
• Its specific impact on myocardial perfusion imaging (MPI) quality parameters requires detailed investigation.
• Current reconstruction methods like filtered back-projection (FBP) and maximum likelihood expectation maximization (ML-EM) have limitations.

Purpose of the Study:

• To compare compressed sensing iterative reconstruction (CS-IR) against FBP and ML-EM for MPI.
• To evaluate the ability of CS-IR to reduce SPECT acquisition time in MPI.
• To assess image quality parameters including uniformity, septal wall thickness, and contrast.

Main Methods:

• A digital phantom simulating the left ventricle myocardium was utilized.
• SPECT projection data were simulated at varying angular sampling (120, 30, 60, 15 directions).
• Images were reconstructed using FBP, ML-EM, and CS-IR; quantitative analysis of uniformity (CV), septal wall thickness, and contrast was performed.

Main Results:

• CS-IR demonstrated lower coefficient of variation (CV) for uniformity compared to FBP and ML-EM across all angular samplings.
• Septal wall thickness with CS-IR was slightly inferior to ML-EM at 360° acquisition (2.5 mm difference).
• Contrast ratios were comparable between ML-EM and CS-IR.
• CS-IR achieved lower CV at quarter acquisition time compared to full acquisition times of other methods.

Conclusions:

• Compressed sensing iterative reconstruction (CS-IR) holds significant potential for reducing MPI acquisition times.
• CS-IR offers improved uniformity and comparable contrast, enabling faster SPECT scans.
• Further investigation is warranted to optimize CS-IR for clinical MPI applications.