MDLinxArtificial Intelligence Tool May Enhance Usability of Medical Images

Artificial Intelligence Tool Enhances Usability of Medical Images by Improving Detection Accuracy

A new AI-driven tool, DEMIST, developed by researchers at Washington University in St. Louis, shows promise in improving the clinical utility of myocardial perfusion imaging (MPI) single-photon emission computed tomography (SPECT) images. By employing a detection-task-specific deep-learning approach, DEMIST enhances image quality without compromising the critical features needed for detecting cardiac defects, thus potentially allowing for reduced radiation doses and shorter scan times.

Key Points:

• AI Tool Development: DEMIST, a deep-learning-based denoising tool, was developed to improve the quality of MPI SPECT images.
• Clinical Task Focus: Unlike traditional methods that prioritize visual appeal, DEMIST focuses on enhancing the images’ clinical utility, particularly for detecting cardiac defects.
• Study Context: The study, published in IEEE Transactions on Radiation and Plasma Medical Sciences, involved evaluating DEMIST on anonymized clinical data from 338 patients across two different scanners.
• Radiation Dose and Scan Time: The tool aims to reduce the radiation dose and acquisition time required for MPI SPECT scans, benefiting patients by reducing exposure and discomfort.
• Image Quality Improvement: DEMIST significantly improved the detection of cardiac defects compared to both low-dose scans and a commonly used task-agnostic denoising method.
• Wide Applicability: The effectiveness of DEMIST was consistent across male and female patients, different defect types, and varied scanning equipment.
• Mathematical Validation: Additional analyses confirmed that DEMIST preserved features crucial for detection tasks, enhancing observer performance.
• Potential for Clinical Use: The findings support the potential for clinical evaluation and future implementation of DEMIST in routine MPI SPECT imaging, aiming to improve diagnostic accuracy and patient care.

“I am excited about these findings since we are seeing that AI may have the potential to enhance the usability of medical images. By providing the possibility to reduce radiation dose and acquisition time, DEMIST offers possibilities to enhance the accuracy and efficiency of detecting myocardial perfusion defects, ultimately benefiting patient care and treatment outcomes.”
— Abhinav Jha, Biomedical Engineer & Assistant Professor in the McKelvey School of Engineering and of Radiology at Mallinckrodt Institute of Radiology in the School of Medicine, Washington University

More in Radiology