Scholarlink

Objective <p>To assess the feasibility of estimating myocardial flow reserve (MFR) from static standardized uptake values (SUVs) derived using stress and rest <sup>13</sup>N-ammonia positron emission tomography (PET) images acquired with a silicon photomultiplier-based PET/computed tomography (CT) system. The goal was to simplify MFR assessment by avoiding dynamic acquisition while maintaining diagnostic accuracy.</p> Methods <p>We retrospectively included 121 consecutive patients who underwent <sup>13</sup>N-ammonia PET myocardial perfusion imaging. The mean SUVs of the left and right ventricles were measured on transverse-axis slices corresponding to the largest cross-sectional area, and the stress-to-rest SUV ratio was calculated. A logarithmic transformation was applied to the MFR, and linear regression models between the log-transformed MFR and the stress-to-rest left ventricular-SUV ratio were developed using six-fold cross-validation across 121 patients. The predicted MFR values from the test folds were then combined to evaluate model performance across all patients with a comparable MFR distribution. Model performance was assessed using the root mean square error (RMSE) between the estimated and true MFR values, and the diagnostic performance for identifying MFRs &lt; 2.0 by receiver operating characteristic (ROC) curve analysis with 95% confidence intervals.</p> Results <p>The RMSE of the predicted MFR calculated using the linear regression model was 0.39. Using an optimal cut-off value of 1.79 for the predicted MFR, the model achieved an area under the ROC curve of 0.89 (95% confidence interval: 0.83–0.95), with a sensitivity of 75% and a specificity of 88%. The proposed simplified SUV-based approach exhibited clinically acceptable error for MFR estimation and could effectively identify patients with an impaired MFR.</p> Conclusion <p>This method may facilitate quantitative MFR assessment in clinical settings where dynamic PET protocols or dedicated software are not available, thereby broadening access to physiological evaluation of myocardial perfusion.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Myocardial flow reserve estimation from standardized uptake values of 13N-ammonia imaging with a silicon photomultiplier positron emission tomography/computed tomography system

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月8日

查看全文
  • Masateru Kawakubo,
  • Yoko Kaimoto,
  • Atsushi Yamamoto,
  • Masaki Watanabe,
  • Akiko Sakai,
  • Michinobu Nagao

摘要

Objective

To assess the feasibility of estimating myocardial flow reserve (MFR) from static standardized uptake values (SUVs) derived using stress and rest 13N-ammonia positron emission tomography (PET) images acquired with a silicon photomultiplier-based PET/computed tomography (CT) system. The goal was to simplify MFR assessment by avoiding dynamic acquisition while maintaining diagnostic accuracy.

Methods

We retrospectively included 121 consecutive patients who underwent 13N-ammonia PET myocardial perfusion imaging. The mean SUVs of the left and right ventricles were measured on transverse-axis slices corresponding to the largest cross-sectional area, and the stress-to-rest SUV ratio was calculated. A logarithmic transformation was applied to the MFR, and linear regression models between the log-transformed MFR and the stress-to-rest left ventricular-SUV ratio were developed using six-fold cross-validation across 121 patients. The predicted MFR values from the test folds were then combined to evaluate model performance across all patients with a comparable MFR distribution. Model performance was assessed using the root mean square error (RMSE) between the estimated and true MFR values, and the diagnostic performance for identifying MFRs < 2.0 by receiver operating characteristic (ROC) curve analysis with 95% confidence intervals.

Results

The RMSE of the predicted MFR calculated using the linear regression model was 0.39. Using an optimal cut-off value of 1.79 for the predicted MFR, the model achieved an area under the ROC curve of 0.89 (95% confidence interval: 0.83–0.95), with a sensitivity of 75% and a specificity of 88%. The proposed simplified SUV-based approach exhibited clinically acceptable error for MFR estimation and could effectively identify patients with an impaired MFR.

Conclusion

This method may facilitate quantitative MFR assessment in clinical settings where dynamic PET protocols or dedicated software are not available, thereby broadening access to physiological evaluation of myocardial perfusion.