Dispositifs médicaux, Équipements médicaux, Intelligence artificielle, Logiciels médicaux, Santé, Sciences de la vie, Sciences et génie, Technologies de l’information et des communications (TIC), Technologies disponibles, Traitements du signal


Approche de caractérisation tissulaire automatique robuste prenant en compte la petite taille des artères et des artefacts OCT.

-Description en anglais seulement-

The need of advanced technologies for coronary artery diseases

  • Coronary Artery Disease (CAD) is becoming one of the most frequent causes of myocardial infarction and sudden death. (with over 17.5 million related deaths annually, which is 31% of global mortality)
  • Progression of pathological formations in arterial wall layers may be followed by acute coronary syndrome
  • Global interventional cardiology market was estimated at $14.69 billion in 2017 and is anticipated to reach $24.96 billion by 2027
  • OCT imaging is a new intracoronary imaging modality
  • Currently, OCT interpretation is reader-dependant with vast inter-observer variation, requiring extensive experience and training
  • Characterization of intracoronary pathological formations is highly challenging due to the artifacts of the imaging system, similar structure of pathological formations, and small available datasets, specifically in children and infants
  • Automatic tissue characterization is important to evaluate coronary artery sequelae and functionality

Automatic tissue characterization model for coronary artery disease

  • Using deep learning for general evaluation of the arterial wall structure for each frame of the OCT pullback
  • Discriminating between the normal arterial wall structure and neo-intimal development due to the coronary artery disease
  • Classifying of coronary artery layers, intima and media, using a Fully Convolutional Network (FCN) in normal arterial wall structure
  • In affected arterial wall with neo-intimal development:
  • Extracting all the coronary artery lesions regardless of the lesion type using a FCN model                     
  • Characterizing the lesion type using Convolutional Neural Networks (CNNs)                                                                       
  • Performing the final decision using majority voting approach
  • Technology developed by Atefeh Abdolmanafi and Prof. Luc Duong (Department of Software and IT Engineering, École de technologie supérieure (ÉTS)), Dr. Nagib Dahdah (Department of Pediatrics, Montreal University) and Prof. Farida Cheriet (Department of Computer Engineering and Software Engineering, Polytechnique Montréal)

Automatic tissue characterization model presents multiple competitive advantages

  • High precision and low computational complexity to be used as a robust indicator of progression of the pathological formations
  • Using majority voting, various detailed tissue information extracted from different networks are considered to characterize lesions
  • Significant impact on efficient clinical diagnosis with a higher level of certainty than current methods
  • Early detection of intracoronary complications is possible using the proposed model to enhance patient outcomes
  • The designed automatic method would have significant impact on efficient clinical diagnosis of coronary artery lesions

Market applications

  • Integrated into intravascular OCT imaging system ( the catheter based OCT)
  • Adapted to any other OCT scanner

Business opportunity

  • PCT has been filed on June 11th 2019
  • Technology available for licensing


If you are interested by this technology, please contact :
Dareen Toumi, Technology Analyst, Engineering
dtoumi@aligo.ca, (514) 618-9297


École de Technologie supérieure (ÉTS)

Main inventors


Atefeh Abdolmanafi, Researcher, Software and IT Engineering

Atefeh Abdolmanafi completed her Ph.D. at École de technologie supérieure in 2018. Due to her background in Physics, she integrated her knowledge in optics, interferometry, scattering, and diffraction with computer science in order to do applied research in advancing medical technologies. She has developed a novel technique for coronary artery tissue characterization for pediatric patients affected by Kawasaki disease, based on deep features and machine learning techniques. She has pursued her research as a postdoctoral fellow at Polytechnique Montréal and CHU Sainte-Justine research center by working on the biophysical modeling of coronary arteries from OCT images to create a computer aided diagnostic framework providing clinicians with an operator-independent diagnosis of histological coronary lesions.


Luc Duong, Professor, Software and IT Engineering

Prof. Luc Duong completed his Ph.D. (2007) at Polytechnique Montréal where he focused on machine learning approaches to guide operative strategy in correcting scoliosis deformities. Then, he was a postdoctoral fellow at Siemens Corporate Research, Princeton, NJ. He contributed to navigation assistance for Chronic Total Occlusion (CTO), in collaboration with the ThoraxCentrum, Rotterdam. Since 2009, Prof Duong has been teaching at the École de technologie supérieure and he is mainly interested in image analysis, computer vision and machine learning. He works closely with the cardiology department of CHU Sainte-Justine in projects related to cardiac interventions for congenital heart defects.


Nagib Dahdah, M.D., Professor, Department of Pediatrics, Montreal University

Dr. Nagib Dahdah received his medical degree from Saint-Joseph University, Beirut, Lebanon, in 1988. He completed his specialization in Pediatric Cardiology from the American Board of Pediatrics, in 1998. He is currently full professor at University of Montreal. His main interest is in pediatric cardiology. Many aspects, such as interventional catheterization, play an important role in his career. More precisely, Kawasaki disease is the center of his clinical research. In his research, he focus on the epidemiological aspect as well as on primary and secondary prevention of the ensuing coronary pathology to prevent the consequences of this disease on the coronary arteries is to establish the diagnosis as early as possible.


Farida Cheriet, Professor, Computer and Software Engineering, Polytechnique Montreal

Prof. Farida Cheriet received the B.Sc. degree in computer science from the University USTHB, Algiers, in 1984, the D.E.A. degree in the field of languages, algorithms and programming from the University of Paris VI, France, in 1986, and the Ph.D. degree in computer science from the University of Montreal, QC, Canada, in 1996. She held a postdoctoral position at the Biomedical Engineering Institute, Polytechnique Montreal, from 1997 to 1999. Since 1999, she has been appointed in the Department of Computer and Software Engineering, Polytechnique Montreal, where she is currently a full Professor. She is an IEEE Senior member and her research interests include three–dimensional (3-D) reconstruction of anatomical structures from medical images, 3-D Augmented Reality systems for minimally invasive surgery and Computer-Assisted Diagnosis Systems in orthopedics, cardiology and ophthalmology.