Maximization of the usage of coronary CTA derived plaque information using a machine learning based algorithm to improve risk stratification; insights from the CONFIRM registry

Alexander R. van Rosendael, Gabriel Maliakal, Kranthi K. Kolli, Ashley Beecy, Subhi J. Al'Aref, Aeshita Dwivedi, Gurpreet Singh, Mohit Panday, Amit Kumar, Xiaoyue Ma, Stephan Achenbach, Mouaz H. Al-Mallah, Daniele Andreini, Jeroen J. Bax, Daniel S. Berman, Matthew J. Budoff, Filippo Cademartiri, Tracy Q. Callister, Hyuk Jae Chang, Kavitha ChinnaiyanBenjamin J.W. Chow, Ricardo C. Cury, Augustin DeLago, Gudrun Feuchtner, Martin Hadamitzky, Joerg Hausleiter, Philipp A. Kaufmann, Yong Jin Kim, Jonathon A. Leipsic, Erica Maffei, Hugo Marques, Gianluca Pontone, Gilbert L. Raff, Ronen Rubinshtein, Leslee J. Shaw, Todd C. Villines, Heidi Gransar, Yao Lu, Erica C. Jones, Jessica M. Peña, Fay Y. Lin, James K. Min

Research output: Contribution to journalArticle

20 Citations (Scopus)

Abstract

Introduction: Machine learning (ML) is a field in computer science that demonstrated to effectively integrate clinical and imaging data for the creation of prognostic scores. The current study investigated whether a ML score, incorporating only the 16 segment coronary tree information derived from coronary computed tomography angiography (CCTA), provides enhanced risk stratification compared with current CCTA based risk scores. Methods: From the multi-center CONFIRM registry, patients were included with complete CCTA risk score information and ≥3 year follow-up for myocardial infarction and death (primary endpoint). Patients with prior coronary artery disease were excluded. Conventional CCTA risk scores (conventional CCTA approach, segment involvement score, duke prognostic index, segment stenosis score, and the Leaman risk score) and a score created using ML were compared for the area under the receiver operating characteristic curve (AUC). Only 16 segment based coronary stenosis (0%, 1–24%, 25–49%, 50–69%, 70–99% and 100%) and composition (calcified, mixed and non-calcified plaque) were provided to the ML model. A boosted ensemble algorithm (extreme gradient boosting; XGBoost) was used and the entire data was randomly split into a training set (80%) and testing set (20%). First, tuned hyperparameters were used to generate a trained model from the training data set (80% of data). Second, the performance of this trained model was independently tested on the unseen test set (20% of data). Results: In total, 8844 patients (mean age 58.0 ± 11.5 years, 57.7% male) were included. During a mean follow-up time of 4.6 ± 1.5 years, 609 events occurred (6.9%). No CAD was observed in 48.7% (3.5% event), non-obstructive CAD in 31.8% (6.8% event), and obstructive CAD in 19.5% (15.6% event). Discrimination of events as expressed by AUC was significantly better for the ML based approach (0.771) vs the other scores (ranging from 0.685 to 0.701), P < 0.001. Net reclassification improvement analysis showed that the improved risk stratification was the result of down-classification of risk among patients that did not experience events (non-events). Conclusion: A risk score created by a ML based algorithm, that utilizes standard 16 coronary segment stenosis and composition information derived from detailed CCTA reading, has greater prognostic accuracy than current CCTA integrated risk scores. These findings indicate that a ML based algorithm can improve the integration of CCTA derived plaque information to improve risk stratification.

Original languageEnglish
Pages (from-to)204-209
Number of pages6
JournalJournal of Cardiovascular Computed Tomography
Volume12
Issue number3
DOIs
Publication statusPublished - 2018 May 1

Fingerprint

Registries
Coronary Stenosis
Area Under Curve
Machine Learning
Computed Tomography Angiography
ROC Curve
Reading
Coronary Artery Disease
Pathologic Constriction
Myocardial Infarction

All Science Journal Classification (ASJC) codes

  • Radiology Nuclear Medicine and imaging
  • Cardiology and Cardiovascular Medicine

Cite this

van Rosendael, Alexander R. ; Maliakal, Gabriel ; Kolli, Kranthi K. ; Beecy, Ashley ; Al'Aref, Subhi J. ; Dwivedi, Aeshita ; Singh, Gurpreet ; Panday, Mohit ; Kumar, Amit ; Ma, Xiaoyue ; Achenbach, Stephan ; Al-Mallah, Mouaz H. ; Andreini, Daniele ; Bax, Jeroen J. ; Berman, Daniel S. ; Budoff, Matthew J. ; Cademartiri, Filippo ; Callister, Tracy Q. ; Chang, Hyuk Jae ; Chinnaiyan, Kavitha ; Chow, Benjamin J.W. ; Cury, Ricardo C. ; DeLago, Augustin ; Feuchtner, Gudrun ; Hadamitzky, Martin ; Hausleiter, Joerg ; Kaufmann, Philipp A. ; Kim, Yong Jin ; Leipsic, Jonathon A. ; Maffei, Erica ; Marques, Hugo ; Pontone, Gianluca ; Raff, Gilbert L. ; Rubinshtein, Ronen ; Shaw, Leslee J. ; Villines, Todd C. ; Gransar, Heidi ; Lu, Yao ; Jones, Erica C. ; Peña, Jessica M. ; Lin, Fay Y. ; Min, James K. / Maximization of the usage of coronary CTA derived plaque information using a machine learning based algorithm to improve risk stratification; insights from the CONFIRM registry. In: Journal of Cardiovascular Computed Tomography. 2018 ; Vol. 12, No. 3. pp. 204-209.
@article{974a29efd4ee47cab28ed42ac603ed84,
title = "Maximization of the usage of coronary CTA derived plaque information using a machine learning based algorithm to improve risk stratification; insights from the CONFIRM registry",
abstract = "Introduction: Machine learning (ML) is a field in computer science that demonstrated to effectively integrate clinical and imaging data for the creation of prognostic scores. The current study investigated whether a ML score, incorporating only the 16 segment coronary tree information derived from coronary computed tomography angiography (CCTA), provides enhanced risk stratification compared with current CCTA based risk scores. Methods: From the multi-center CONFIRM registry, patients were included with complete CCTA risk score information and ≥3 year follow-up for myocardial infarction and death (primary endpoint). Patients with prior coronary artery disease were excluded. Conventional CCTA risk scores (conventional CCTA approach, segment involvement score, duke prognostic index, segment stenosis score, and the Leaman risk score) and a score created using ML were compared for the area under the receiver operating characteristic curve (AUC). Only 16 segment based coronary stenosis (0{\%}, 1–24{\%}, 25–49{\%}, 50–69{\%}, 70–99{\%} and 100{\%}) and composition (calcified, mixed and non-calcified plaque) were provided to the ML model. A boosted ensemble algorithm (extreme gradient boosting; XGBoost) was used and the entire data was randomly split into a training set (80{\%}) and testing set (20{\%}). First, tuned hyperparameters were used to generate a trained model from the training data set (80{\%} of data). Second, the performance of this trained model was independently tested on the unseen test set (20{\%} of data). Results: In total, 8844 patients (mean age 58.0 ± 11.5 years, 57.7{\%} male) were included. During a mean follow-up time of 4.6 ± 1.5 years, 609 events occurred (6.9{\%}). No CAD was observed in 48.7{\%} (3.5{\%} event), non-obstructive CAD in 31.8{\%} (6.8{\%} event), and obstructive CAD in 19.5{\%} (15.6{\%} event). Discrimination of events as expressed by AUC was significantly better for the ML based approach (0.771) vs the other scores (ranging from 0.685 to 0.701), P < 0.001. Net reclassification improvement analysis showed that the improved risk stratification was the result of down-classification of risk among patients that did not experience events (non-events). Conclusion: A risk score created by a ML based algorithm, that utilizes standard 16 coronary segment stenosis and composition information derived from detailed CCTA reading, has greater prognostic accuracy than current CCTA integrated risk scores. These findings indicate that a ML based algorithm can improve the integration of CCTA derived plaque information to improve risk stratification.",
author = "{van Rosendael}, {Alexander R.} and Gabriel Maliakal and Kolli, {Kranthi K.} and Ashley Beecy and Al'Aref, {Subhi J.} and Aeshita Dwivedi and Gurpreet Singh and Mohit Panday and Amit Kumar and Xiaoyue Ma and Stephan Achenbach and Al-Mallah, {Mouaz H.} and Daniele Andreini and Bax, {Jeroen J.} and Berman, {Daniel S.} and Budoff, {Matthew J.} and Filippo Cademartiri and Callister, {Tracy Q.} and Chang, {Hyuk Jae} and Kavitha Chinnaiyan and Chow, {Benjamin J.W.} and Cury, {Ricardo C.} and Augustin DeLago and Gudrun Feuchtner and Martin Hadamitzky and Joerg Hausleiter and Kaufmann, {Philipp A.} and Kim, {Yong Jin} and Leipsic, {Jonathon A.} and Erica Maffei and Hugo Marques and Gianluca Pontone and Raff, {Gilbert L.} and Ronen Rubinshtein and Shaw, {Leslee J.} and Villines, {Todd C.} and Heidi Gransar and Yao Lu and Jones, {Erica C.} and Pe{\~n}a, {Jessica M.} and Lin, {Fay Y.} and Min, {James K.}",
year = "2018",
month = "5",
day = "1",
doi = "10.1016/j.jcct.2018.04.011",
language = "English",
volume = "12",
pages = "204--209",
journal = "Journal of Cardiovascular Computed Tomography",
issn = "1934-5925",
publisher = "Elsevier Inc.",
number = "3",

}

van Rosendael, AR, Maliakal, G, Kolli, KK, Beecy, A, Al'Aref, SJ, Dwivedi, A, Singh, G, Panday, M, Kumar, A, Ma, X, Achenbach, S, Al-Mallah, MH, Andreini, D, Bax, JJ, Berman, DS, Budoff, MJ, Cademartiri, F, Callister, TQ, Chang, HJ, Chinnaiyan, K, Chow, BJW, Cury, RC, DeLago, A, Feuchtner, G, Hadamitzky, M, Hausleiter, J, Kaufmann, PA, Kim, YJ, Leipsic, JA, Maffei, E, Marques, H, Pontone, G, Raff, GL, Rubinshtein, R, Shaw, LJ, Villines, TC, Gransar, H, Lu, Y, Jones, EC, Peña, JM, Lin, FY & Min, JK 2018, 'Maximization of the usage of coronary CTA derived plaque information using a machine learning based algorithm to improve risk stratification; insights from the CONFIRM registry', Journal of Cardiovascular Computed Tomography, vol. 12, no. 3, pp. 204-209. https://doi.org/10.1016/j.jcct.2018.04.011

Maximization of the usage of coronary CTA derived plaque information using a machine learning based algorithm to improve risk stratification; insights from the CONFIRM registry. / van Rosendael, Alexander R.; Maliakal, Gabriel; Kolli, Kranthi K.; Beecy, Ashley; Al'Aref, Subhi J.; Dwivedi, Aeshita; Singh, Gurpreet; Panday, Mohit; Kumar, Amit; Ma, Xiaoyue; Achenbach, Stephan; Al-Mallah, Mouaz H.; Andreini, Daniele; Bax, Jeroen J.; Berman, Daniel S.; Budoff, Matthew J.; Cademartiri, Filippo; Callister, Tracy Q.; Chang, Hyuk Jae; Chinnaiyan, Kavitha; Chow, Benjamin J.W.; Cury, Ricardo C.; DeLago, Augustin; Feuchtner, Gudrun; Hadamitzky, Martin; Hausleiter, Joerg; Kaufmann, Philipp A.; Kim, Yong Jin; Leipsic, Jonathon A.; Maffei, Erica; Marques, Hugo; Pontone, Gianluca; Raff, Gilbert L.; Rubinshtein, Ronen; Shaw, Leslee J.; Villines, Todd C.; Gransar, Heidi; Lu, Yao; Jones, Erica C.; Peña, Jessica M.; Lin, Fay Y.; Min, James K.

In: Journal of Cardiovascular Computed Tomography, Vol. 12, No. 3, 01.05.2018, p. 204-209.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Maximization of the usage of coronary CTA derived plaque information using a machine learning based algorithm to improve risk stratification; insights from the CONFIRM registry

AU - van Rosendael, Alexander R.

AU - Maliakal, Gabriel

AU - Kolli, Kranthi K.

AU - Beecy, Ashley

AU - Al'Aref, Subhi J.

AU - Dwivedi, Aeshita

AU - Singh, Gurpreet

AU - Panday, Mohit

AU - Kumar, Amit

AU - Ma, Xiaoyue

AU - Achenbach, Stephan

AU - Al-Mallah, Mouaz H.

AU - Andreini, Daniele

AU - Bax, Jeroen J.

AU - Berman, Daniel S.

AU - Budoff, Matthew J.

AU - Cademartiri, Filippo

AU - Callister, Tracy Q.

AU - Chang, Hyuk Jae

AU - Chinnaiyan, Kavitha

AU - Chow, Benjamin J.W.

AU - Cury, Ricardo C.

AU - DeLago, Augustin

AU - Feuchtner, Gudrun

AU - Hadamitzky, Martin

AU - Hausleiter, Joerg

AU - Kaufmann, Philipp A.

AU - Kim, Yong Jin

AU - Leipsic, Jonathon A.

AU - Maffei, Erica

AU - Marques, Hugo

AU - Pontone, Gianluca

AU - Raff, Gilbert L.

AU - Rubinshtein, Ronen

AU - Shaw, Leslee J.

AU - Villines, Todd C.

AU - Gransar, Heidi

AU - Lu, Yao

AU - Jones, Erica C.

AU - Peña, Jessica M.

AU - Lin, Fay Y.

AU - Min, James K.

PY - 2018/5/1

Y1 - 2018/5/1

N2 - Introduction: Machine learning (ML) is a field in computer science that demonstrated to effectively integrate clinical and imaging data for the creation of prognostic scores. The current study investigated whether a ML score, incorporating only the 16 segment coronary tree information derived from coronary computed tomography angiography (CCTA), provides enhanced risk stratification compared with current CCTA based risk scores. Methods: From the multi-center CONFIRM registry, patients were included with complete CCTA risk score information and ≥3 year follow-up for myocardial infarction and death (primary endpoint). Patients with prior coronary artery disease were excluded. Conventional CCTA risk scores (conventional CCTA approach, segment involvement score, duke prognostic index, segment stenosis score, and the Leaman risk score) and a score created using ML were compared for the area under the receiver operating characteristic curve (AUC). Only 16 segment based coronary stenosis (0%, 1–24%, 25–49%, 50–69%, 70–99% and 100%) and composition (calcified, mixed and non-calcified plaque) were provided to the ML model. A boosted ensemble algorithm (extreme gradient boosting; XGBoost) was used and the entire data was randomly split into a training set (80%) and testing set (20%). First, tuned hyperparameters were used to generate a trained model from the training data set (80% of data). Second, the performance of this trained model was independently tested on the unseen test set (20% of data). Results: In total, 8844 patients (mean age 58.0 ± 11.5 years, 57.7% male) were included. During a mean follow-up time of 4.6 ± 1.5 years, 609 events occurred (6.9%). No CAD was observed in 48.7% (3.5% event), non-obstructive CAD in 31.8% (6.8% event), and obstructive CAD in 19.5% (15.6% event). Discrimination of events as expressed by AUC was significantly better for the ML based approach (0.771) vs the other scores (ranging from 0.685 to 0.701), P < 0.001. Net reclassification improvement analysis showed that the improved risk stratification was the result of down-classification of risk among patients that did not experience events (non-events). Conclusion: A risk score created by a ML based algorithm, that utilizes standard 16 coronary segment stenosis and composition information derived from detailed CCTA reading, has greater prognostic accuracy than current CCTA integrated risk scores. These findings indicate that a ML based algorithm can improve the integration of CCTA derived plaque information to improve risk stratification.

AB - Introduction: Machine learning (ML) is a field in computer science that demonstrated to effectively integrate clinical and imaging data for the creation of prognostic scores. The current study investigated whether a ML score, incorporating only the 16 segment coronary tree information derived from coronary computed tomography angiography (CCTA), provides enhanced risk stratification compared with current CCTA based risk scores. Methods: From the multi-center CONFIRM registry, patients were included with complete CCTA risk score information and ≥3 year follow-up for myocardial infarction and death (primary endpoint). Patients with prior coronary artery disease were excluded. Conventional CCTA risk scores (conventional CCTA approach, segment involvement score, duke prognostic index, segment stenosis score, and the Leaman risk score) and a score created using ML were compared for the area under the receiver operating characteristic curve (AUC). Only 16 segment based coronary stenosis (0%, 1–24%, 25–49%, 50–69%, 70–99% and 100%) and composition (calcified, mixed and non-calcified plaque) were provided to the ML model. A boosted ensemble algorithm (extreme gradient boosting; XGBoost) was used and the entire data was randomly split into a training set (80%) and testing set (20%). First, tuned hyperparameters were used to generate a trained model from the training data set (80% of data). Second, the performance of this trained model was independently tested on the unseen test set (20% of data). Results: In total, 8844 patients (mean age 58.0 ± 11.5 years, 57.7% male) were included. During a mean follow-up time of 4.6 ± 1.5 years, 609 events occurred (6.9%). No CAD was observed in 48.7% (3.5% event), non-obstructive CAD in 31.8% (6.8% event), and obstructive CAD in 19.5% (15.6% event). Discrimination of events as expressed by AUC was significantly better for the ML based approach (0.771) vs the other scores (ranging from 0.685 to 0.701), P < 0.001. Net reclassification improvement analysis showed that the improved risk stratification was the result of down-classification of risk among patients that did not experience events (non-events). Conclusion: A risk score created by a ML based algorithm, that utilizes standard 16 coronary segment stenosis and composition information derived from detailed CCTA reading, has greater prognostic accuracy than current CCTA integrated risk scores. These findings indicate that a ML based algorithm can improve the integration of CCTA derived plaque information to improve risk stratification.

UR - http://www.scopus.com/inward/record.url?scp=85046686593&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85046686593&partnerID=8YFLogxK

U2 - 10.1016/j.jcct.2018.04.011

DO - 10.1016/j.jcct.2018.04.011

M3 - Article

C2 - 29753765

AN - SCOPUS:85046686593

VL - 12

SP - 204

EP - 209

JO - Journal of Cardiovascular Computed Tomography

JF - Journal of Cardiovascular Computed Tomography

SN - 1934-5925

IS - 3

ER -