Michael A. Winkler
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Michael A. Winkler has a total of 140 co-authors. Here are the top co-authors by number of shared publications:
- Driss Raissi
Radiology, University of Kentucky College of Medicine, Lexington, USA
6 shared publications Mohamed Issa
Department of Radiology, University of Kentucky Chandler Medical Center, Lexington, KY, United States of America
5 shared publicationsKevin P. Landwehr
Department of Radiology, University of Kentucky Medical Center, 805 Rose Street, MN-104G, Lexington, KY 40536-0298, USA
3 shared publicationsMichael A. Brooks
Department of Radiology, University of Kentucky, 800 Rose St, Lexington, KY 40536
3 shared publications- Mohamed Issa
Department of Radiology, College of Medicine, University of Kentucky, Lexington, Kentucky, USA
3 shared publications Stephen B. Hobbs
Department of Radiology, University of Kentucky, 800 Rose St, Lexington, KY 40536
3 shared publications
Latest Publications
- 15 September 2026
- Published by Elsevier in Journal of Cardiovascular Computed Tomography
Blooming is more than partial volume averaging
- 15 September 2026
- Published by Elsevier in Journal of Cardiovascular Computed Tomography
Blooming is more than partial volume averaging
- Open Access
- 26 August 2026
- Published by MDPI AG in Diagnostics
- Vol. 16 (17) , 2735
Late Gadolinium Enhancement Entropy as a Novel Imaging Biomarker of Myocardial Tissue Heterogeneity—A Comprehensive Review
Late gadolinium enhancement (LGE) cardiac magnetic resonance is the reference standard for non-invasive myocardial tissue characterization. Conventional LGE analysis focuses on the presence and extent of fibrosis, yet these measures incompletely describe the spatial complexity of myocardial scar that underpins arrhythmogenesis. Entropy, derived from radiomic analysis of LGE signal-intensity distributions, has emerged as a surrogate marker of myocardial tissue heterogeneity. Mechanistically, heterogeneous fibrosis promotes electrical conduction alterations, and entropy serves as a global descriptor of this complex substrate. A growing body of evidence suggests that higher LGE entropy is associated with increased arrhythmogenicity and major adverse cardiac events. Several studies have shown that this association remains significant after adjustment for conventional clinical and imaging predictors. A smaller number of studies have gone further, demonstrating that incorporation of LGE entropy improves the discriminatory or reclassification performance of established risk-prediction models. This narrative review critically synthesizes the current evidence on LGE-derived entropy, compares methodological approaches and clinical applications, and discusses its principal limitations. After standardization and prospective validation, entropy-based phenotyping may prove useful for individualized risk stratification beyond conventional LGE metrics and guide clinical decision-making.
- 26 August 2026
- Published by MDPI AG in Diagnostics
- Vol. 16 (17) , 2735
Open AccessLate Gadolinium Enhancement Entropy as a Novel Imaging Biomarker of Myocardial Tissue Heterogeneity—A Comprehensive Review
Late gadolinium enhancement (LGE) cardiac magnetic resonance is the reference standard for non-invasive myocardial tissue characterization. Conventional LGE analysis focuses on the presence and extent of fibrosis, yet these measures incompletely describe the spatial complexity of myocardial scar that underpins arrhythmogenesis. Entropy, derived from radiomic analysis of LGE signal-intensity distributions, has emerged as a surrogate marker of myocardial tissue heterogeneity. Mechanistically, heterogeneous fibrosis promotes electrical conduction alterations, and entropy serves as a global descriptor of this complex substrate. A growing body of evidence suggests that higher LGE entropy is associated with increased arrhythmogenicity and major adverse cardiac events. Several studies have shown that this association remains significant after adjustment for conventional clinical and imaging predictors. A smaller number of studies have gone further, demonstrating that incorporation of LGE entropy improves the discriminatory or reclassification performance of established risk-prediction models. This narrative review critically synthesizes the current evidence on LGE-derived entropy, compares methodological approaches and clinical applications, and discusses its principal limitations. After standardization and prospective validation, entropy-based phenotyping may prove useful for individualized risk stratification beyond conventional LGE metrics and guide clinical decision-making.
- 29 June 2025
- Published by Springer Nature
- p. 218-234
PySCIPOpt-ML: Embedding Trained Machine Learning Models into Mixed-Integer Programs
- 29 June 2025
- Published by Springer Nature
- p. 218-234
PySCIPOpt-ML: Embedding Trained Machine Learning Models into Mixed-Integer Programs
- Open Access
- 4 September 2024
- Published by Elsevier in JACC: Case Reports
- Vol. 29 (17) , 102496
Maffucci Syndrome May Be a Heritable Thoracic Aortic Disease and a Cause of Aortic Dissection
- 4 September 2024
- Published by Elsevier in JACC: Case Reports
- Vol. 29 (17) , 102496
Open AccessMaffucci Syndrome May Be a Heritable Thoracic Aortic Disease and a Cause of Aortic Dissection
- Open Access
Adaptive Cut Selection in Mixed-Integer Linear Programming
Cutting plane selection is a subroutine used in all modern mixed-integer linear programming solvers with the goal of selecting a subset of generated cuts that induce optimal solver performance. These solvers have millions of parameter combinations, and so are excellent candidates for parameter tuning. Cut selection scoring rules are usually weighted sums of different measurements, where the weights are parameters. We present a parametric family of mixed-integer linear programs together with infinitely many family-wide valid cuts. Some of these cuts can induce integer optimal solutions directly after being applied, while others fail to do so even if an infinite amount are applied. We show for a specific cut selection rule, that any finite grid search of the parameter space will always miss all parameter values, which select integer optimal inducing cuts in an infinite amount of our problems. We propose a variation on the design of existing graph convolutional neural networks, adapting them to learn cut selection rule parameters. We present a reinforcement learning framework for selecting cuts, and train our design using said framework over MIPLIB 2017 and a neural network verification data set. Our framework and design show that adaptive cut selection does substantially improve performance over a diverse set of instances, but that finding a single function describing such a rule is difficult. Code for reproducing all experiments is available at https://github.com/Opt-Mucca/Adaptive-Cutsel-MILP.
Open AccessAdaptive Cut Selection in Mixed-Integer Linear Programming
Cutting plane selection is a subroutine used in all modern mixed-integer linear programming solvers with the goal of selecting a subset of generated cuts that induce optimal solver performance. These solvers have millions of parameter combinations, and so are excellent candidates for parameter tuning. Cut selection scoring rules are usually weighted sums of different measurements, where the weights are parameters. We present a parametric family of mixed-integer linear programs together with infinitely many family-wide valid cuts. Some of these cuts can induce integer optimal solutions directly after being applied, while others fail to do so even if an infinite amount are applied. We show for a specific cut selection rule, that any finite grid search of the parameter space will always miss all parameter values, which select integer optimal inducing cuts in an infinite amount of our problems. We propose a variation on the design of existing graph convolutional neural networks, adapting them to learn cut selection rule parameters. We present a reinforcement learning framework for selecting cuts, and train our design using said framework over MIPLIB 2017 and a neural network verification data set. Our framework and design show that adaptive cut selection does substantially improve performance over a diverse set of instances, but that finding a single function describing such a rule is difficult. Code for reproducing all experiments is available at https://github.com/Opt-Mucca/Adaptive-Cutsel-MILP.
- Open Access
- 1 October 2022
- Published by Oxford University Press (OUP) in European Heart Journal
- Vol. 43 (Supplement)
Clinical significance of the hemodynamic gain index in patients undergoing exercise stress testing and coronary computed tomography angiography
Background: Many hemodynamic parameters are used during exercise stress testing but provide limited information regarding obstructive coronary artery disease (CAD) when exercise is suboptimal. Hemodynamic gain index (HGI) is a recently developed and sensitive indicator of ischemia which has been associated with increased mortality. Purpose: This study sought to evaluate the clinical impact of HGI in patients who underwent concomitant exercise testing and coronary computed tomography angiography (CCTA). Methods: Consecutive patients from the executive health program between 2010 and 2018 were identified. Resting and peak heart rate (HR) and systolic blood pressure (SBP) were recorded. Framingham risk score (FRS), Duke treadmill score (DTS) and HGI [(HRpeak × SBPpeak) − (HRrest × SBPrest)] / (HRrest × SBPrest) were calculated. The latter was divided into quartiles. The presence or absence of any CAD on CCTA was confirmed. Multivariate analysis and artificial neural network were used to determine the independent predictors of obstructive CAD. Results: There were 284 patients (mean age 53 years, 83% male). Mean HGI was 1.74±0.67, with severely blunted HGI cut-off of ≤1.25 (Quartile 4). Patients with blunted HGI were older, had higher FRS, and worse DTS. Patients with obstructive CAD showed lower HGI when compared to control (1.36±0.53 vs. 1.77±0.67, P-value 0.005) and had more prevalence of severely blunted HGI ≤1.25 (44% vs. 22%, P-value 0.019). In the multivariate analysis, after adjusting for traditional risk factors, HGI remained independent predictor of obstructive CAD while severely blunted HGI ≤1.25 was associated with 3-fold increased odds of obstructive CAD (P-value 0.05). Using artificial intelligence analysis, blunted HGI remained an independent predictor of obstructive CAD an area under the curve of 0.83 and 0.96, and normalized importance of HGI of 100% and 63%, respectively for different models. Conclusions: Among patients who underwent concomitant exercise testing and CCTA, blunted HGI ≤1.25 was strongly associated with obstructive CAD after adjusting for traditional risk factors. HGI is robust, practical, and allows clinicians to reclassify patients, change management, and thus improve prognosis. Funding Acknowledgement: Type of funding sources: None.
- 1 October 2022
- Published by Oxford University Press (OUP) in European Heart Journal
- Vol. 43 (Supplement)
Open AccessClinical significance of the hemodynamic gain index in patients undergoing exercise stress testing and coronary computed tomography angiography
Background: Many hemodynamic parameters are used during exercise stress testing but provide limited information regarding obstructive coronary artery disease (CAD) when exercise is suboptimal. Hemodynamic gain index (HGI) is a recently developed and sensitive indicator of ischemia which has been associated with increased mortality. Purpose: This study sought to evaluate the clinical impact of HGI in patients who underwent concomitant exercise testing and coronary computed tomography angiography (CCTA). Methods: Consecutive patients from the executive health program between 2010 and 2018 were identified. Resting and peak heart rate (HR) and systolic blood pressure (SBP) were recorded. Framingham risk score (FRS), Duke treadmill score (DTS) and HGI [(HRpeak × SBPpeak) − (HRrest × SBPrest)] / (HRrest × SBPrest) were calculated. The latter was divided into quartiles. The presence or absence of any CAD on CCTA was confirmed. Multivariate analysis and artificial neural network were used to determine the independent predictors of obstructive CAD. Results: There were 284 patients (mean age 53 years, 83% male). Mean HGI was 1.74±0.67, with severely blunted HGI cut-off of ≤1.25 (Quartile 4). Patients with blunted HGI were older, had higher FRS, and worse DTS. Patients with obstructive CAD showed lower HGI when compared to control (1.36±0.53 vs. 1.77±0.67, P-value 0.005) and had more prevalence of severely blunted HGI ≤1.25 (44% vs. 22%, P-value 0.019). In the multivariate analysis, after adjusting for traditional risk factors, HGI remained independent predictor of obstructive CAD while severely blunted HGI ≤1.25 was associated with 3-fold increased odds of obstructive CAD (P-value 0.05). Using artificial intelligence analysis, blunted HGI remained an independent predictor of obstructive CAD an area under the curve of 0.83 and 0.96, and normalized importance of HGI of 100% and 63%, respectively for different models. Conclusions: Among patients who underwent concomitant exercise testing and CCTA, blunted HGI ≤1.25 was strongly associated with obstructive CAD after adjusting for traditional risk factors. HGI is robust, practical, and allows clinicians to reclassify patients, change management, and thus improve prognosis. Funding Acknowledgement: Type of funding sources: None.
- Open Access
- 25 April 2022
- Published by Georg Thieme Verlag KG in International Journal of Angiology
- Vol. 32 (04) , 258-261
A Lost Bullet in the Coronary Sinus: A Cautionary Tale
We present a case of venous bullet embolism to the right atrium following a gunshot wound (GSW) to the abdomen. A 53-year-old male presented after a GSW to the abdomen. His workup included a computed tomography (CT) scan demonstrating an aortic injury with aortocaval fistula. A radio-opaque object consistent with a bullet was visualized in the right atrium. First, this case details an important decision, choice of surgery versus an interventional approach. After repair of the aortocaval fistula, the patient underwent a planned attempt to extract the bullet through a right lateral thoracotomy approach utilizing cardiopulmonary bypass to facilitate a right atriotomy. Intraoperatively, the team was not able to localize the bullet in the right atrium despite fluoroscopic evaluation. A postoperative CT scan demonstrated that the bullet had migrated into the coronary sinus. Lastly, this case demonstrates successful positioning maneuvers to dislodge the bullet out of the heart and into the inferior vena cava, allowing for the endovascular extraction of the bullet.
- 25 April 2022
- Published by Georg Thieme Verlag KG in International Journal of Angiology
- Vol. 32 (04) , 258-261
Open AccessA Lost Bullet in the Coronary Sinus: A Cautionary Tale
We present a case of venous bullet embolism to the right atrium following a gunshot wound (GSW) to the abdomen. A 53-year-old male presented after a GSW to the abdomen. His workup included a computed tomography (CT) scan demonstrating an aortic injury with aortocaval fistula. A radio-opaque object consistent with a bullet was visualized in the right atrium. First, this case details an important decision, choice of surgery versus an interventional approach. After repair of the aortocaval fistula, the patient underwent a planned attempt to extract the bullet through a right lateral thoracotomy approach utilizing cardiopulmonary bypass to facilitate a right atriotomy. Intraoperatively, the team was not able to localize the bullet in the right atrium despite fluoroscopic evaluation. A postoperative CT scan demonstrated that the bullet had migrated into the coronary sinus. Lastly, this case demonstrates successful positioning maneuvers to dislodge the bullet out of the heart and into the inferior vena cava, allowing for the endovascular extraction of the bullet.
- Open Access
- 1 January 2022
- Published by Oxford University Press (OUP) in BJR|case reports
- Vol. 8 (1) , 20210069
Swiss cheese interventricular septum presenting with catastrophic stroke: the potential role of ECG-gated CTA
Ventricular septal defect is a common congenital cardiac condition that presents in a variety of morphologies. Less commonly, when an individual patient is found to have multiple ventricular septal defects, the term “Swiss cheese ventricular septal defect” is applied. Although not routinely utilized in clinical practice, electrocardiogram (ECG)-gated computed tomographic angiography (CTA) has been shown to provide utility in detecting intracardiac shunts, demonstrating promise in preventing acute strokes secondary to a paradoxical embolus from occurring; this is especially important when atypical cardiac septa are suspected. This case seeks to illustrate how usage of ECG-gated CTA can assist in early detection and prevention of adverse outcomes resulting from an atypical presentation of a ventricular septal defect.
- 1 January 2022
- Published by Oxford University Press (OUP) in BJR|case reports
- Vol. 8 (1) , 20210069
Open AccessSwiss cheese interventricular septum presenting with catastrophic stroke: the potential role of ECG-gated CTA
Ventricular septal defect is a common congenital cardiac condition that presents in a variety of morphologies. Less commonly, when an individual patient is found to have multiple ventricular septal defects, the term “Swiss cheese ventricular septal defect” is applied. Although not routinely utilized in clinical practice, electrocardiogram (ECG)-gated computed tomographic angiography (CTA) has been shown to provide utility in detecting intracardiac shunts, demonstrating promise in preventing acute strokes secondary to a paradoxical embolus from occurring; this is especially important when atypical cardiac septa are suspected. This case seeks to illustrate how usage of ECG-gated CTA can assist in early detection and prevention of adverse outcomes resulting from an atypical presentation of a ventricular septal defect.
- 25 August 2021
- Published by Georg Thieme Verlag KG in International Journal of Angiology
- Vol. 30 (03) , 212-220
Three Technologies That Will Guide Revascularization of Chronic Coronary Syndrome Patients into the 21st Century: A Review
Although medical therapy is the preferred first-line treatment for patients with chronic coronary syndrome (CCS), revascularization remains an important consideration. We present a review that identifies the three diagnostic technologies most important to guiding the decision to revascularize patients with CCS: (1) cardiac computed tomography, (2) intracoronary imaging, and (3) lesion-specific physiological guidance. MW substantially contributed to the concept and design, drafting of the manuscript, literature review, summary and qualitative synthesis of publications related to noninvasive cardiovascular imaging and intracoronary imaging, and selection and annotation of figures. RP was involved in literature review, some drafting, qualitative synthesis of publications related to lesion specific physiological guidance, and selection and annotation of figures. WF was involved in literature review, some drafting, editing, and manuscript preparation. VA was involved in concept and design, editing, and minor drafting. NW was involved in concept and design, editing, final draft, integrity of the manuscript. Publication Date: 25 August 2021 (online) © 2021. International College of Angiology. This article is published by Thieme. Thieme Medical Publishers, Inc. 333 Seventh Avenue, 18th Floor, New York, NY 10001, USA
- 25 August 2021
- Published by Georg Thieme Verlag KG in International Journal of Angiology
- Vol. 30 (03) , 212-220
Three Technologies That Will Guide Revascularization of Chronic Coronary Syndrome Patients into the 21st Century: A Review
Although medical therapy is the preferred first-line treatment for patients with chronic coronary syndrome (CCS), revascularization remains an important consideration. We present a review that identifies the three diagnostic technologies most important to guiding the decision to revascularize patients with CCS: (1) cardiac computed tomography, (2) intracoronary imaging, and (3) lesion-specific physiological guidance. MW substantially contributed to the concept and design, drafting of the manuscript, literature review, summary and qualitative synthesis of publications related to noninvasive cardiovascular imaging and intracoronary imaging, and selection and annotation of figures. RP was involved in literature review, some drafting, qualitative synthesis of publications related to lesion specific physiological guidance, and selection and annotation of figures. WF was involved in literature review, some drafting, editing, and manuscript preparation. VA was involved in concept and design, editing, and minor drafting. NW was involved in concept and design, editing, final draft, integrity of the manuscript. Publication Date: 25 August 2021 (online) © 2021. International College of Angiology. This article is published by Thieme. Thieme Medical Publishers, Inc. 333 Seventh Avenue, 18th Floor, New York, NY 10001, USA
- 1 February 2021
- Published by Elsevier in JACC: Cardiovascular Interventions
- Vol. 14 (3) , 347-349
Arteriolymphatic Fistula
- 1 February 2021
- Published by Elsevier in JACC: Cardiovascular Interventions
- Vol. 14 (3) , 347-349
Arteriolymphatic Fistula