TY - JOUR
T1 - Integrated prediction of lesion-specific ischaemia from quantitative coronary CT angiography using machine learning
T2 - a multicentre study
AU - Dey, Damini
AU - Gaur, Sara
AU - Ovrehus, Kristian A
AU - Slomka, Piotr J
AU - Betancur, Julian
AU - Goeller, Markus
AU - Hell, Michaela M
AU - Gransar, Heidi
AU - Berman, Daniel S
AU - Achenbach, Stephan
AU - Botker, Hans Erik
AU - Jensen, Jesper Moller
AU - Lassen, Jens Flensted
AU - Norgaard, Bjarne Linde
PY - 2018/6/1
Y1 - 2018/6/1
N2 - Objectives We aimed to investigate if lesion-specific ischaemia by invasive fractional flow reserve (FFR) can be predicted by an integrated machine learning (ML) ischaemia risk score from quantitative plaque measures from coronary computed tomography angiography (CTA). Methods In a multicentre trial of 254 patients, CTA and invasive coronary angiography were performed, with FFR in 484 vessels. CTA data sets were analysed by semi-automated software to quantify stenosis and non-calcified (NCP), low-density NCP (LD-NCP, < 30 HU), calcified and total plaque volumes, contrast density difference (CDD, maximum difference in luminal attenuation per unit area) and plaque length. ML integration included automated feature selection and model building from quantitative CTA with a boosted ensemble algorithm, and tenfold stratified cross-validation. Results Eighty patients had ischaemia by FFR (FFR ≤ 0.80) in 100 vessels. Information gain for predicting ischaemia was highest for CDD (0.172), followed by LD-NCP (0.125), NCP (0.097), and total plaque volumes (0.092). ML exhibited higher area-under-the-curve (0.84) than individual CTA measures, including stenosis (0.76), LD-NCP volume (0.77), total plaque volume (0.74) and pre-test likelihood of coronary artery disease (CAD) (0.63); p < 0.006. Conclusions Integrated ML ischaemia risk score improved the prediction of lesion-specific ischaemia by invasive FFR, over stenosis, plaque measures and pre-test likelihood of CAD.
AB - Objectives We aimed to investigate if lesion-specific ischaemia by invasive fractional flow reserve (FFR) can be predicted by an integrated machine learning (ML) ischaemia risk score from quantitative plaque measures from coronary computed tomography angiography (CTA). Methods In a multicentre trial of 254 patients, CTA and invasive coronary angiography were performed, with FFR in 484 vessels. CTA data sets were analysed by semi-automated software to quantify stenosis and non-calcified (NCP), low-density NCP (LD-NCP, < 30 HU), calcified and total plaque volumes, contrast density difference (CDD, maximum difference in luminal attenuation per unit area) and plaque length. ML integration included automated feature selection and model building from quantitative CTA with a boosted ensemble algorithm, and tenfold stratified cross-validation. Results Eighty patients had ischaemia by FFR (FFR ≤ 0.80) in 100 vessels. Information gain for predicting ischaemia was highest for CDD (0.172), followed by LD-NCP (0.125), NCP (0.097), and total plaque volumes (0.092). ML exhibited higher area-under-the-curve (0.84) than individual CTA measures, including stenosis (0.76), LD-NCP volume (0.77), total plaque volume (0.74) and pre-test likelihood of coronary artery disease (CAD) (0.63); p < 0.006. Conclusions Integrated ML ischaemia risk score improved the prediction of lesion-specific ischaemia by invasive FFR, over stenosis, plaque measures and pre-test likelihood of CAD.
KW - Journal Article
UR - http://www.scopus.com/inward/record.url?scp=85040640232&partnerID=8YFLogxK
U2 - 10.1007/s00330-017-5223-z
DO - 10.1007/s00330-017-5223-z
M3 - Journal article
C2 - 29352380
SN - 0938-7994
VL - 28
SP - 2655
EP - 2664
JO - European Radiology
JF - European Radiology
IS - 6
ER -