Computationally simulated fractional flow reserve from coronary computed tomography angiography based on fractional myocardial mass

Huan Han, Yong Gyun Bae, Seung Tae Hwang, Hyung Yoon Kim, Il Park, Sung Mok Kim, Yeonhyeon Choe, Young June Moon, Jin Ho Choi

Research output: Contribution to journalArticle

Abstract

Computed tomography angiography (CCTA)-based calculations of fractional flow reserve (FFR) can improve the diagnostic performance of CCTA for physiologically significant stenosis but the computational resource requirements are high. This study aimed at establishing a simple and efficient algorithm for computing simulated FFR (S-FFR). A total of 107 patients who underwent CCTA and invasive FFR measurements were enrolled in the study. S-FFR was calculated using 145 evaluable coronary arteries with off-the-shelf softwares. FFR ≤ 0.80 was a reference threshold for diagnostic performance of diameter stenosis (DS) ≥ 50%, DS ≥ 70%, or S-FFR ≤ 0.80. FFR ≤ 0.80 was identified in 78 vessels (54%). In per-vessel analysis, S-FFR showed good correlation (r = 0.83) and agreement (mean difference = 0.02 ± 0.08) with FFR. The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of S-FFR ≤ 0.80 for FFR ≤ 0.80 were 84%, 92%, 92%, 83%, and 88%, respectively. S-FFR ≤ 0.80 showed much higher predictive performance for FFR ≤ 0.80 compared with DS ≥ 50% or DS ≥ 70% (c-statistics = 0.92 vs. 0.58 or 0.65, p < 0.001, all). The classification agreement between FFR and S-FFR was > 80% when the average of FFR and S-FFR was < 0.76 or > 0.86. Per-patient analysis showed consistent results. In this study, a simple and computationally efficient simulated FFR (S-FFR) algorithm is designed and tested using non-proprietary off-the-shelf software. This algorithm may expand the accessibility of clinical applications for non-invasive coronary physiology study.

Original languageEnglish
JournalInternational Journal of Cardiovascular Imaging
DOIs
Publication statusAccepted/In press - 2018 Jan 1

Fingerprint

Pathologic Constriction
Software
Coronary Vessels
Computed Tomography Angiography
Sensitivity and Specificity

Keywords

  • Computational coronary physiology
  • Computed tomography
  • Coronary circulation

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging
  • Cardiology and Cardiovascular Medicine

Cite this

Computationally simulated fractional flow reserve from coronary computed tomography angiography based on fractional myocardial mass. / Han, Huan; Bae, Yong Gyun; Hwang, Seung Tae; Kim, Hyung Yoon; Park, Il; Kim, Sung Mok; Choe, Yeonhyeon; Moon, Young June; Choi, Jin Ho.

In: International Journal of Cardiovascular Imaging, 01.01.2018.

Research output: Contribution to journalArticle

Han, Huan ; Bae, Yong Gyun ; Hwang, Seung Tae ; Kim, Hyung Yoon ; Park, Il ; Kim, Sung Mok ; Choe, Yeonhyeon ; Moon, Young June ; Choi, Jin Ho. / Computationally simulated fractional flow reserve from coronary computed tomography angiography based on fractional myocardial mass. In: International Journal of Cardiovascular Imaging. 2018.
@article{dd2d0508fd484722bbbee864153795a1,
title = "Computationally simulated fractional flow reserve from coronary computed tomography angiography based on fractional myocardial mass",
abstract = "Computed tomography angiography (CCTA)-based calculations of fractional flow reserve (FFR) can improve the diagnostic performance of CCTA for physiologically significant stenosis but the computational resource requirements are high. This study aimed at establishing a simple and efficient algorithm for computing simulated FFR (S-FFR). A total of 107 patients who underwent CCTA and invasive FFR measurements were enrolled in the study. S-FFR was calculated using 145 evaluable coronary arteries with off-the-shelf softwares. FFR ≤ 0.80 was a reference threshold for diagnostic performance of diameter stenosis (DS) ≥ 50{\%}, DS ≥ 70{\%}, or S-FFR ≤ 0.80. FFR ≤ 0.80 was identified in 78 vessels (54{\%}). In per-vessel analysis, S-FFR showed good correlation (r = 0.83) and agreement (mean difference = 0.02 ± 0.08) with FFR. The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of S-FFR ≤ 0.80 for FFR ≤ 0.80 were 84{\%}, 92{\%}, 92{\%}, 83{\%}, and 88{\%}, respectively. S-FFR ≤ 0.80 showed much higher predictive performance for FFR ≤ 0.80 compared with DS ≥ 50{\%} or DS ≥ 70{\%} (c-statistics = 0.92 vs. 0.58 or 0.65, p < 0.001, all). The classification agreement between FFR and S-FFR was > 80{\%} when the average of FFR and S-FFR was < 0.76 or > 0.86. Per-patient analysis showed consistent results. In this study, a simple and computationally efficient simulated FFR (S-FFR) algorithm is designed and tested using non-proprietary off-the-shelf software. This algorithm may expand the accessibility of clinical applications for non-invasive coronary physiology study.",
keywords = "Computational coronary physiology, Computed tomography, Coronary circulation",
author = "Huan Han and Bae, {Yong Gyun} and Hwang, {Seung Tae} and Kim, {Hyung Yoon} and Il Park and Kim, {Sung Mok} and Yeonhyeon Choe and Moon, {Young June} and Choi, {Jin Ho}",
year = "2018",
month = "1",
day = "1",
doi = "10.1007/s10554-018-1432-z",
language = "English",
journal = "International Journal of Cardiovascular Imaging",
issn = "1569-5794",
publisher = "Springer Netherlands",

}

TY - JOUR

T1 - Computationally simulated fractional flow reserve from coronary computed tomography angiography based on fractional myocardial mass

AU - Han, Huan

AU - Bae, Yong Gyun

AU - Hwang, Seung Tae

AU - Kim, Hyung Yoon

AU - Park, Il

AU - Kim, Sung Mok

AU - Choe, Yeonhyeon

AU - Moon, Young June

AU - Choi, Jin Ho

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Computed tomography angiography (CCTA)-based calculations of fractional flow reserve (FFR) can improve the diagnostic performance of CCTA for physiologically significant stenosis but the computational resource requirements are high. This study aimed at establishing a simple and efficient algorithm for computing simulated FFR (S-FFR). A total of 107 patients who underwent CCTA and invasive FFR measurements were enrolled in the study. S-FFR was calculated using 145 evaluable coronary arteries with off-the-shelf softwares. FFR ≤ 0.80 was a reference threshold for diagnostic performance of diameter stenosis (DS) ≥ 50%, DS ≥ 70%, or S-FFR ≤ 0.80. FFR ≤ 0.80 was identified in 78 vessels (54%). In per-vessel analysis, S-FFR showed good correlation (r = 0.83) and agreement (mean difference = 0.02 ± 0.08) with FFR. The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of S-FFR ≤ 0.80 for FFR ≤ 0.80 were 84%, 92%, 92%, 83%, and 88%, respectively. S-FFR ≤ 0.80 showed much higher predictive performance for FFR ≤ 0.80 compared with DS ≥ 50% or DS ≥ 70% (c-statistics = 0.92 vs. 0.58 or 0.65, p < 0.001, all). The classification agreement between FFR and S-FFR was > 80% when the average of FFR and S-FFR was < 0.76 or > 0.86. Per-patient analysis showed consistent results. In this study, a simple and computationally efficient simulated FFR (S-FFR) algorithm is designed and tested using non-proprietary off-the-shelf software. This algorithm may expand the accessibility of clinical applications for non-invasive coronary physiology study.

AB - Computed tomography angiography (CCTA)-based calculations of fractional flow reserve (FFR) can improve the diagnostic performance of CCTA for physiologically significant stenosis but the computational resource requirements are high. This study aimed at establishing a simple and efficient algorithm for computing simulated FFR (S-FFR). A total of 107 patients who underwent CCTA and invasive FFR measurements were enrolled in the study. S-FFR was calculated using 145 evaluable coronary arteries with off-the-shelf softwares. FFR ≤ 0.80 was a reference threshold for diagnostic performance of diameter stenosis (DS) ≥ 50%, DS ≥ 70%, or S-FFR ≤ 0.80. FFR ≤ 0.80 was identified in 78 vessels (54%). In per-vessel analysis, S-FFR showed good correlation (r = 0.83) and agreement (mean difference = 0.02 ± 0.08) with FFR. The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of S-FFR ≤ 0.80 for FFR ≤ 0.80 were 84%, 92%, 92%, 83%, and 88%, respectively. S-FFR ≤ 0.80 showed much higher predictive performance for FFR ≤ 0.80 compared with DS ≥ 50% or DS ≥ 70% (c-statistics = 0.92 vs. 0.58 or 0.65, p < 0.001, all). The classification agreement between FFR and S-FFR was > 80% when the average of FFR and S-FFR was < 0.76 or > 0.86. Per-patient analysis showed consistent results. In this study, a simple and computationally efficient simulated FFR (S-FFR) algorithm is designed and tested using non-proprietary off-the-shelf software. This algorithm may expand the accessibility of clinical applications for non-invasive coronary physiology study.

KW - Computational coronary physiology

KW - Computed tomography

KW - Coronary circulation

UR - http://www.scopus.com/inward/record.url?scp=85051788552&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85051788552&partnerID=8YFLogxK

U2 - 10.1007/s10554-018-1432-z

DO - 10.1007/s10554-018-1432-z

M3 - Article

C2 - 30128848

AN - SCOPUS:85051788552

JO - International Journal of Cardiovascular Imaging

JF - International Journal of Cardiovascular Imaging

SN - 1569-5794

ER -