Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A detection-task-specific deep-learning method to improve the quality of sparse-view myocardial perfusion SPECT images

Zezhang Yang, Zitong Yu, Nuri Choi, Abhinav K. Jha

TL;DR

The paper tackles the challenge of shortening MPI SPECT scanning time by adopting sparse-view acquisitions without compromising perfusion defect detection. It introduces a detection-task-specific deep-learning framework that combines a fidelity loss with an observer-based loss grounded in anthropomorphic channel features, aiming to preserve clinically relevant information. Evaluation on a retrospective, IRB-approved dataset demonstrates that the method yields higher defect-detection performance (AUC) than sparse-view protocols and a task-agnostic baseline, with qualitative improvements including restoration of left-ventricle wall structure. These findings suggest task-specific loss functions informed by human observer models can enhance the clinical utility of sparsely sampled MPI SPECT, meriting broader multi-center validation.

Abstract

Myocardial perfusion imaging (MPI) with single-photon emission computed tomography (SPECT) is a widely used and cost-effective diagnostic tool for coronary artery disease. However, the lengthy scanning time in this imaging procedure can cause patient discomfort, motion artifacts, and potentially inaccurate diagnoses due to misalignment between the SPECT scans and the CT-scans which are acquired for attenuation compensation. Reducing projection angles is a potential way to shorten scanning time, but this can adversely impact the quality of the reconstructed images. To address this issue, we propose a detection-task-specific deep-learning method for sparse-view MPI SPECT images. This method integrates an observer loss term that penalizes the loss of anthropomorphic channel features with the goal of improving performance in perfusion defect-detection task. We observed that, on the task of detecting myocardial perfusion defects, the proposed method yielded an area under the receiver operating characteristic (ROC) curve (AUC) significantly larger than the sparse-view protocol. Further, the proposed method was observed to be able to restore the structure of the left ventricle wall, demonstrating ability to overcome sparse-sampling artifacts. Our preliminary results motivate further evaluations of the method.

A detection-task-specific deep-learning method to improve the quality of sparse-view myocardial perfusion SPECT images

TL;DR

The paper tackles the challenge of shortening MPI SPECT scanning time by adopting sparse-view acquisitions without compromising perfusion defect detection. It introduces a detection-task-specific deep-learning framework that combines a fidelity loss with an observer-based loss grounded in anthropomorphic channel features, aiming to preserve clinically relevant information. Evaluation on a retrospective, IRB-approved dataset demonstrates that the method yields higher defect-detection performance (AUC) than sparse-view protocols and a task-agnostic baseline, with qualitative improvements including restoration of left-ventricle wall structure. These findings suggest task-specific loss functions informed by human observer models can enhance the clinical utility of sparsely sampled MPI SPECT, meriting broader multi-center validation.

Abstract

Myocardial perfusion imaging (MPI) with single-photon emission computed tomography (SPECT) is a widely used and cost-effective diagnostic tool for coronary artery disease. However, the lengthy scanning time in this imaging procedure can cause patient discomfort, motion artifacts, and potentially inaccurate diagnoses due to misalignment between the SPECT scans and the CT-scans which are acquired for attenuation compensation. Reducing projection angles is a potential way to shorten scanning time, but this can adversely impact the quality of the reconstructed images. To address this issue, we propose a detection-task-specific deep-learning method for sparse-view MPI SPECT images. This method integrates an observer loss term that penalizes the loss of anthropomorphic channel features with the goal of improving performance in perfusion defect-detection task. We observed that, on the task of detecting myocardial perfusion defects, the proposed method yielded an area under the receiver operating characteristic (ROC) curve (AUC) significantly larger than the sparse-view protocol. Further, the proposed method was observed to be able to restore the structure of the left ventricle wall, demonstrating ability to overcome sparse-sampling artifacts. Our preliminary results motivate further evaluations of the method.

Paper Structure

This paper contains 8 sections, 1 equation, 3 figures.

Table of Contents

  1. Introduction
  2. Methods
  3. Proposed method
  4. Objective evaluation of the proposed method
  5. Results
  6. Discussion
  7. Conclusion and future work
  8. ACKNOWLEDGMENTS

Figures (3)

  • Figure 1: 3D neural network architecture of the proposed method
  • Figure 2: AUC values for the full-view and sparse-view protocol images, as well as the images processed using the proposed method and the task-agnostic DL method.
  • Figure 3: Two representative test cases qualitatively demonstrate the performance of the proposed method. In each case, the sparse-view level was set to 5. In the uppercase and lowercase cases, defects were located in the inferior and anterior walls, respectively. For all cases, the defects had an extent of 30 degrees and a severity of 25%.