Computer Vision Centre - The Interactive and Augmented Modelling Group

The Interactive and Augmented Modelling (IAM) group from the Computer Vision Centre (CVC) is a multidisciplinary team of mathematicians, physics and engineers, experts in mathematical and statistical modelling of heterogeneous data (especially from images and videos) for augmented reality intelligent environments in biomedical applications and health. Current research interests include the use of geometric modelling, graphs and predictive statistics for the extraction of anatomical information and cellular patterns for bronchoscopy guidance and in-vivo diagnosis of lung cancer, modelling of radiogenomic patterns for personalized treatments in interactive graphical environments able to issue recommendations without altering clinical protocols and driver-vehicle interactions in highly automated systems. The group actively cooperates with national and international institutions including Hospital Universitari de Bellvitge and Uni Vest Timisoara (since 2009) in in-vivo diagnosis systems, Hospital Sant Pau-BSC (since 2007) in personalized models of cardiac biomechanics, Vall Hebron Oncologic Institute (since 2017) in cancer personalized treatments and University of Essex in driving performance evaluation. The members of the group have more than 300 publications in indexed peer-review journals and international conferences and they have also participated in several European and national research projects. More information about the research group: http://iam.cvc.uab.es/.

Project description: Assistance systems for Health applications have some specific requirements that demand of new methods for data gathering, analysis and modelling able to deal with heterogeneous SmallData. Systems should dynamically collect data from, both, the environment and the user itself to provide personalized recommendations and actions. Data analysis should be performed with a limited number of samples in heterogeneous views prone to include biased influential cases. Finally, algorithms should run in real time with possibly limited computational resources and fluctuant internet access. Electronic medical devices can enhance the process of data gathering and analysis in several ways. By acquiring multiple data which can be combined to extract more meaningful features leading to better predictions. Also, devices can help to detect the quality of the sampling process and avoid including noise on the new samples. Finally, multiprocessor embedded systems can provide efficient implementation of SmallData methods with good compromise between processing speed and power consumption. The project focus on the complementary role of Cyber-Physical devices and analysis of SmallData in the design of personalized models for Health applications. We address design of Cyber-Physical Systems in Health for Personalized Assistance in two specific application cases:

a) A Smart Assisted Driving System for dynamical assessment of the driving capabilities of Mild Cognitive Impaired people

b) An Intelligent Operating Room for improving the yield of bronchoscopic interventions for in-vivo lung cancer diagnosis

The candidate should complement IAM expertise on mathematical and statistical modelling of heterogeneous data with design Cyber-Physical systems and FPGA-GPU parallel programming of AI methods.