Matthieu De Beule
Matthieu De Beule has a total of 41 co-authors. Here are the top co-authors by number of shared publications:
- Patrick Segers
IBiTech-bioMMeda, Ghent University, Ghent, Belgium
3 shared publications - Benedict Verhegghe
Ghent University, Ghent, Belgium
2 shared publications - Joni Dambre
Ghent University
2 shared publications Nic Debusschere
Department of Electronics and Information Systems, Ghent University, Ghent, Belgium
2 shared publications- Patricio Astudillo
Robovision, Ghent, Belgium
2 shared publications - Annemieke van Haver
More Institute, Monica Orthopaedic Research Institute Antwerp Belgium
1 shared publication
Latest Publications
- Open Access
- 28 May 2020
- Published by Wiley in Journal of Interventional Cardiology
- Vol. 2020, 1-9
Automatic Detection of the Aortic Annular Plane and Coronary Ostia from Multidetector Computed Tomography
Anatomic landmark detection is crucial during preoperative planning of transcatheter aortic valve implantation (TAVI) to select the proper device size and assess the risk of complications. The detection is currently a time-consuming manual process influenced by the image quality and subject to operator variability. In this work, we propose a novel automatic method to detect the relevant aortic landmarks from MDCT images using deep learning techniques. We trained three convolutional neural networks (CNNs) with 344 multidetector computed tomography (MDCT) acquisitions to detect five anatomical landmarks relevant for TAVI planning: the three basal attachment points of the aortic valve leaflets and the left and right coronary ostia. The detection strategy used these three CNN models to analyse a single MDCT image and yield three segmentation volumes as output. These segmentation volumes were averaged into one final segmentation volume, and the final predicted landmarks were obtained during a postprocessing step. Finally, we constructed the aortic annular plane, defined by the three predicted hinge points, and measured the distances from this plane to the predicted coronary ostia (i.e., coronary height). The methodology was validated on 100 patients. The automatic landmark detection was able to detect all the landmarks and showed high accuracy as the median distance between the ground truth and predictions is lower than the interobserver variations (1.5 mm [1.1–2.1], 2.0 mm [1.3–2.8] with a paired difference −0.5 ± 1.3 mm and value <0.001). Furthermore, a high correlation is observed between predicted and manually measured coronary heights (for both R2 = 0.8). The image analysis time per patient was below one second. The proposed method is accurate, fast, and reproducible. Embedding this tool based on deep learning in the preoperative planning routine may have an impact in the TAVI environments by reducing the time and cost and improving accuracy.- 28 May 2020
- Published by Wiley in Journal of Interventional Cardiology
- Vol. 2020, 1-9
Open AccessAutomatic Detection of the Aortic Annular Plane and Coronary Ostia from Multidetector Computed Tomography
Anatomic landmark detection is crucial during preoperative planning of transcatheter aortic valve implantation (TAVI) to select the proper device size and assess the risk of complications. The detection is currently a time-consuming manual process influenced by the image quality and subject to operator variability. In this work, we propose a novel automatic method to detect the relevant aortic landmarks from MDCT images using deep learning techniques. We trained three convolutional neural networks (CNNs) with 344 multidetector computed tomography (MDCT) acquisitions to detect five anatomical landmarks relevant for TAVI planning: the three basal attachment points of the aortic valve leaflets and the left and right coronary ostia. The detection strategy used these three CNN models to analyse a single MDCT image and yield three segmentation volumes as output. These segmentation volumes were averaged into one final segmentation volume, and the final predicted landmarks were obtained during a postprocessing step. Finally, we constructed the aortic annular plane, defined by the three predicted hinge points, and measured the distances from this plane to the predicted coronary ostia (i.e., coronary height). The methodology was validated on 100 patients. The automatic landmark detection was able to detect all the landmarks and showed high accuracy as the median distance between the ground truth and predictions is lower than the interobserver variations (1.5 mm [1.1–2.1], 2.0 mm [1.3–2.8] with a paired difference −0.5 ± 1.3 mm and value <0.001). Furthermore, a high correlation is observed between predicted and manually measured coronary heights (for both R2 = 0.8). The image analysis time per patient was below one second. The proposed method is accurate, fast, and reproducible. Embedding this tool based on deep learning in the preoperative planning routine may have an impact in the TAVI environments by reducing the time and cost and improving accuracy. - 1 December 2019
- Published by Elsevier in Morphologie
- Vol. 103 (343) , 139-147
Towards safe and efficient preoperative planning of transcatheter mitral valve interventions
- 1 December 2019
- Published by Elsevier in Morphologie
- Vol. 103 (343) , 139-147
Towards safe and efficient preoperative planning of transcatheter mitral valve interventions
- 7 October 2018
- Published by Wiley in International Journal for Numerical Methods in Biomedical Engineering
- Vol. 35 (1) , e3151
Kinematic boundary conditions substantially impact in silico ventricular function
Computational cardiac mechanical models, individualized to the patient, have the potential to elucidate the fundamentals of cardiac (patho‐)physiology, enable non‐invasive quantification of clinically significant metrics (eg, stiffness, active contraction, work), and anticipate the potential efficacy of therapeutic cardiovascular intervention. In a clinical setting, however, the available imaging resolution is often limited, which limits cardiac models to focus on the ventricles, without including the atria, valves, and proximal arteries and veins. In such models, the absence of surrounding structures needs to be accounted for by imposing realistic kinematic boundary conditions, which, for prognostic purposes, are preferably generic and thus non‐image derived. Unfortunately, the literature on cardiac models shows no consistent approach to kinematically constrain the myocardium. The impact of different approaches (eg, fully constrained base, constrained epi‐ring) on the predictive capacity of cardiac mechanical models has not been thoroughly studied. For that reason, this study first gives an overview of current approaches to kinematically constrain (bi) ventricular models. Next, we developed a patient‐specific in silico biventricular model that compares well with literature and in vivo recorded strains. Alternative constraints were introduced to assess the influence of commonly used mechanical boundary conditions on both the predicted global functional behavior of the in‐silico heart (cavity volumes, stroke volume, ejection fraction) and local strain distributions. Meaningful differences in global functioning were found between different kinematic anchoring strategies, which brought forward the importance of selecting appropriate boundary conditions for biventricular models that, in the near future, may inform clinical intervention. However, whilst statistically significant differences were also found in local strain distributions, these differences were minor and mostly confined to the region close to the applied boundary conditions.
- 7 October 2018
- Published by Wiley in International Journal for Numerical Methods in Biomedical Engineering
- Vol. 35 (1) , e3151
Kinematic boundary conditions substantially impact in silico ventricular function
Computational cardiac mechanical models, individualized to the patient, have the potential to elucidate the fundamentals of cardiac (patho‐)physiology, enable non‐invasive quantification of clinically significant metrics (eg, stiffness, active contraction, work), and anticipate the potential efficacy of therapeutic cardiovascular intervention. In a clinical setting, however, the available imaging resolution is often limited, which limits cardiac models to focus on the ventricles, without including the atria, valves, and proximal arteries and veins. In such models, the absence of surrounding structures needs to be accounted for by imposing realistic kinematic boundary conditions, which, for prognostic purposes, are preferably generic and thus non‐image derived. Unfortunately, the literature on cardiac models shows no consistent approach to kinematically constrain the myocardium. The impact of different approaches (eg, fully constrained base, constrained epi‐ring) on the predictive capacity of cardiac mechanical models has not been thoroughly studied. For that reason, this study first gives an overview of current approaches to kinematically constrain (bi) ventricular models. Next, we developed a patient‐specific in silico biventricular model that compares well with literature and in vivo recorded strains. Alternative constraints were introduced to assess the influence of commonly used mechanical boundary conditions on both the predicted global functional behavior of the in‐silico heart (cavity volumes, stroke volume, ejection fraction) and local strain distributions. Meaningful differences in global functioning were found between different kinematic anchoring strategies, which brought forward the importance of selecting appropriate boundary conditions for biventricular models that, in the near future, may inform clinical intervention. However, whilst statistically significant differences were also found in local strain distributions, these differences were minor and mostly confined to the region close to the applied boundary conditions.
- Open Access
- 5 August 2016
- Published by Springer Nature in BioMedical Engineering OnLine
- Vol. 15 (1) , 1-16
Coronary fractional flow reserve measurements of a stenosed side branch: a computational study investigating the influence of the bifurcation angle
Coronary hemodynamics and physiology specific for bifurcation lesions was not well understood. To investigate the influence of the bifurcation angle on the intracoronary hemodynamics of side branch (SB) lesions computational fluid dynamics simulations were performed. A parametric model representing a left anterior descending—first diagonal coronary bifurcation lesion was created according to the literature. Diameters obeyed fractal branching laws. Proximal and distal main branch (DMB) stenoses were both set at 60 %. We varied the distal bifurcation angles (40°, 55°, and 70°), the flow splits to the DMB and SB (55 %:45 %, 65 %:35 %, and 75 %:25 %), and the SB stenoses (40, 60, and 80 %), resulting in 27 simulations. Fractional flow reserve, defined as the ratio between the mean distal stenosis and mean aortic pressure during maximal hyperemia, was calculated for the DMB and SB (FFRSB) for all simulations. The largest differences in FFRSB comparing the largest and smallest bifurcation angles were 0.02 (in cases with 40 % SB stenosis, irrespective of the assumed flow split) and 0.05 (in cases with 60 % SB stenosis, flow split 55 %:45 %). When the SB stenosis was 80 %, the difference in FFRSB between the largest and smallest bifurcation angle was 0.33 (flow split 55 %:45 %). By describing the ΔPSB−QSB relationship using a quadratic curve for cases with 80 % SB stenosis, we found that the curve was steeper (i.e. higher flow resistance) when bifurcation angle increases (ΔP = 0.451*Q + 0.010*Q 2 and ΔP = 0.687*Q + 0.017*Q 2 for 40° and 70° bifurcation angle, respectively). Our analyses revealed complex hemodynamics in all cases with evident counter-rotating helical flow structures. Larger bifurcation angles resulted in more pronounced helical flow structures (i.e. higher helicity intensity), when 60 or 80 % SB stenoses were present. A good correlation (R2 = 0.80) between the SB pressure drop and helicity intensity was also found. Our analyses showed that, in bifurcation lesions with 60 % MB stenosis and 80 % SB stenosis, SB pressure drop is higher for larger bifurcation angles suggesting higher flow resistance (i.e. curves describing the ΔPSB−QSB relationship being steeper). When the SB stenosis is mild (40 %) or moderate (60 %), SB resistance is minimally influenced by the bifurcation angle, with differences not being clinically meaningful. Our findings also highlighted the complex interplay between anatomy, pressure drops, and blood flow helicity in bifurcations.
- 5 August 2016
- Published by Springer Nature in BioMedical Engineering OnLine
- Vol. 15 (1) , 1-16
Open AccessCoronary fractional flow reserve measurements of a stenosed side branch: a computational study investigating the influence of the bifurcation angle
Coronary hemodynamics and physiology specific for bifurcation lesions was not well understood. To investigate the influence of the bifurcation angle on the intracoronary hemodynamics of side branch (SB) lesions computational fluid dynamics simulations were performed. A parametric model representing a left anterior descending—first diagonal coronary bifurcation lesion was created according to the literature. Diameters obeyed fractal branching laws. Proximal and distal main branch (DMB) stenoses were both set at 60 %. We varied the distal bifurcation angles (40°, 55°, and 70°), the flow splits to the DMB and SB (55 %:45 %, 65 %:35 %, and 75 %:25 %), and the SB stenoses (40, 60, and 80 %), resulting in 27 simulations. Fractional flow reserve, defined as the ratio between the mean distal stenosis and mean aortic pressure during maximal hyperemia, was calculated for the DMB and SB (FFRSB) for all simulations. The largest differences in FFRSB comparing the largest and smallest bifurcation angles were 0.02 (in cases with 40 % SB stenosis, irrespective of the assumed flow split) and 0.05 (in cases with 60 % SB stenosis, flow split 55 %:45 %). When the SB stenosis was 80 %, the difference in FFRSB between the largest and smallest bifurcation angle was 0.33 (flow split 55 %:45 %). By describing the ΔPSB−QSB relationship using a quadratic curve for cases with 80 % SB stenosis, we found that the curve was steeper (i.e. higher flow resistance) when bifurcation angle increases (ΔP = 0.451*Q + 0.010*Q 2 and ΔP = 0.687*Q + 0.017*Q 2 for 40° and 70° bifurcation angle, respectively). Our analyses revealed complex hemodynamics in all cases with evident counter-rotating helical flow structures. Larger bifurcation angles resulted in more pronounced helical flow structures (i.e. higher helicity intensity), when 60 or 80 % SB stenoses were present. A good correlation (R2 = 0.80) between the SB pressure drop and helicity intensity was also found. Our analyses showed that, in bifurcation lesions with 60 % MB stenosis and 80 % SB stenosis, SB pressure drop is higher for larger bifurcation angles suggesting higher flow resistance (i.e. curves describing the ΔPSB−QSB relationship being steeper). When the SB stenosis is mild (40 %) or moderate (60 %), SB resistance is minimally influenced by the bifurcation angle, with differences not being clinically meaningful. Our findings also highlighted the complex interplay between anatomy, pressure drops, and blood flow helicity in bifurcations.
- 1 November 2015
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 66, 39-46
Tissue prolapse and stresses in stented coronary arteries: A computer model for multi-layer atherosclerotic plaque
- 1 November 2015
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 66, 39-46
Tissue prolapse and stresses in stented coronary arteries: A computer model for multi-layer atherosclerotic plaque
- 20 June 2012
- Published by ASME International
- p. 65-66
Modeling of Coated Biodegradable Stents
A bioresorbable stent supports the stenosed blood vessel during the healing period after coronary angioplasty and then gradually disappears. Unlike permanent stents, the biodegradable stent forms no obstacle for future interventions. Moreover, the degradable stent material presents an ideal vehicle for local drug delivery. Long term side effects inherent to drug eluting stents such as in-stent restenosis and late stent thrombosis might be avoided [1]. To date, several bioresorbable stents are being developed or are currently being tested in clinical trials. Two classes of biomaterials are being used in biodegradable stent technology: biodegradable polymers and bioerodible metal alloys. Polymers can be tailored to have a well-defined degradational behaviour but have relatively poor mechanical properties. Biocorrodible metals such as magnesium alloys have good mechanical characteristics but display a more complex an less predictive degradational behaviour. A biocorrodible metallic stent coated with a biodegradable polymer might be able to combine the benefits of both metallic and polymeric biodegradable stents. Finite element modelling can play an important role in the study of nevel stent designs. To correctly simulate the behaviour of degradable stents a material model must be developed that incorporates the effect of degradation on all material characteristics. In case of a coated biocorrodible magnesium stent this includes corrosion modelling, the effect of the coating and the influence of mechanical loading on the corrosion rate.
- 20 June 2012
- Published by ASME International
- p. 65-66
Modeling of Coated Biodegradable Stents
A bioresorbable stent supports the stenosed blood vessel during the healing period after coronary angioplasty and then gradually disappears. Unlike permanent stents, the biodegradable stent forms no obstacle for future interventions. Moreover, the degradable stent material presents an ideal vehicle for local drug delivery. Long term side effects inherent to drug eluting stents such as in-stent restenosis and late stent thrombosis might be avoided [1]. To date, several bioresorbable stents are being developed or are currently being tested in clinical trials. Two classes of biomaterials are being used in biodegradable stent technology: biodegradable polymers and bioerodible metal alloys. Polymers can be tailored to have a well-defined degradational behaviour but have relatively poor mechanical properties. Biocorrodible metals such as magnesium alloys have good mechanical characteristics but display a more complex an less predictive degradational behaviour. A biocorrodible metallic stent coated with a biodegradable polymer might be able to combine the benefits of both metallic and polymeric biodegradable stents. Finite element modelling can play an important role in the study of nevel stent designs. To correctly simulate the behaviour of degradable stents a material model must be developed that incorporates the effect of degradation on all material characteristics. In case of a coated biocorrodible magnesium stent this includes corrosion modelling, the effect of the coating and the influence of mechanical loading on the corrosion rate.
- 1 December 2010
- Published by Europa Digital & Publishing in EuroIntervention
- Vol. 6 (J) , J53-J60
Coronary bifurcation stenting: insights from in vitro and virtual bench testing
The Official Journal of EuroPCR and the European Association of Percutaneous Cardiovascular Interventions (EAPCI)
- 1 December 2010
- Published by Europa Digital & Publishing in EuroIntervention
- Vol. 6 (J) , J53-J60
Coronary bifurcation stenting: insights from in vitro and virtual bench testing
The Official Journal of EuroPCR and the European Association of Percutaneous Cardiovascular Interventions (EAPCI)
- 1 January 2010
- Published by Springer Nature
- p. 639-642
Patellofemoral Contact during Simulated Weight Bearing Squat Movement: A Cadaveric Study
- 1 January 2010
- Published by Springer Nature
- p. 639-642
Patellofemoral Contact during Simulated Weight Bearing Squat Movement: A Cadaveric Study
- 1 January 2009
- Published by Springer Nature
- p. 1442-1445
3D cephalometry: a new approach for landmark identification and image orientation
- 1 January 2009
- Published by Springer Nature
- p. 1442-1445
3D cephalometry: a new approach for landmark identification and image orientation