[Paper Review] A physics and learning-based transmission-less attenuation compensation method for SPECT

Zitong Yu, Ashequr Rahman+3 more|arXiv (Cornell University)|Feb 10, 2021

Medical Imaging Techniques and Applications45 references4 citations

TL;DR

This paper proposes a physics and learning-based transmission-less attenuation compensation method for SPECT that uses scatter and photopeak emission data to estimate attenuation maps without a transmission scan. By combining a physics-based initial attenuation map reconstruction with a convolutional neural network for region segmentation and assigned attenuation coefficients, the method achieves image quality and defect detection performance statistically indistinguishable from using true attenuation maps in a realistic myocardial perfusion SPECT simulation study.

ABSTRACT

Attenuation compensation (AC) is a pre-requisite for reliable quantification and beneficial for visual interpretation tasks in single-photon emission computed tomography (SPECT). Typical AC methods require the availability of an attenuation map obtained using a transmission scan, such as a CT scan. This has several disadvantages such as increased radiation dose, higher costs, and possible misalignment between SPECT and CT scans. Also, often a CT scan is unavailable. In this context, we and others are showing that scattered photons in SPECT contain information to estimate the attenuation distribution. To exploit this observation, we propose a physics and learning-based method that uses the SPECT emission data in the photopeak and scatter windows to perform transmission-less AC in SPECT. The proposed method uses data acquired in the scatter window to reconstruct an initial estimate of the attenuation map using a physics-based approach. A convolutional neural network is then trained to segment this initial estimate into different regions. Pre-defined attenuation coefficients are assigned to these regions, yielding the reconstructed attenuation map, which is then used to reconstruct the activity map using an ordered subsets expectation maximization-based reconstruction approach. We objectively evaluated the performance of this method using a highly realistic simulation study conducted on the clinically relevant task of detecting perfusion defects in myocardial perfusion SPECT. Our results showed no statistically significant differences between the performance achieved using the proposed method and that with the true attenuation maps. Visually, the images reconstructed using the proposed method looked similar to those with the true attenuation map. Overall, these results provide evidence of the capability of the proposed method to perform transmission-less AC and motivate further evaluation.

Motivation & Objective

• To address the limitations of traditional attenuation compensation in SPECT that rely on transmission scans, such as increased radiation dose and misregistration.
• To explore whether scattered photons in SPECT emission data contain sufficient information to estimate attenuation maps without a transmission scan.
• To develop a hybrid method combining physics-based reconstruction and deep learning for accurate, transmission-less attenuation compensation.
• To evaluate the method's performance on the clinically relevant task of detecting myocardial perfusion defects.

Proposed method

• The method uses emission data from the scatter window to reconstruct an initial attenuation map using a physics-based iterative reconstruction approach.
• A convolutional neural network is trained to segment the initial attenuation map into anatomical regions based on image features.
• Pre-defined attenuation coefficients (e.g., soft tissue, lung, heart) are assigned to each segmented region to generate a final attenuation map.
• The final attenuation map is used in an ordered subsets expectation maximization (OSEM) reconstruction algorithm to produce the activity map.
• The approach avoids the need for a transmission scan, reducing radiation dose and eliminating misregistration risks.
• The method is evaluated using a realistic simulation framework tailored to myocardial perfusion SPECT.

Experimental results

Research questions

• RQ1 Can scattered photons in SPECT emission data provide sufficient information to estimate attenuation maps without a transmission scan?
• RQ2 How well can a physics-based reconstruction of the attenuation map from scatter data perform when enhanced by deep learning segmentation?
• RQ3 Does the proposed transmission-less attenuation compensation method achieve diagnostic performance comparable to that of using true attenuation maps in perfusion defect detection?
• RQ4 Can the method maintain image quality and quantitative accuracy without requiring additional imaging modalities?

Key findings

• The proposed method achieved no statistically significant difference in perfusion defect detection performance compared to the gold standard using true attenuation maps.
• Visual quality of images reconstructed with the proposed method was indistinguishable from those reconstructed with true attenuation maps.
• The method effectively leveraged scatter data and deep learning to produce accurate attenuation maps without requiring a transmission scan.
• The combination of physics-based reconstruction and CNN segmentation enabled robust attenuation map estimation in a clinically relevant simulation setting.