Session Abstract

In this talk, I will overview our recent developments in the design of a full-leg prosthesis to replace the missing joint(s) of people having suffered from lower-limb amputation. The bionic ankle ELSA is a compact device that replicates the biomechanical function of a human ankle in daily locomotion tasks. It embeds all technological components within the volume of the shoe, such that it has the potential to be used by people with very low-level of amputation. Our prosthetic knee is fully passive, yet can deliver assistance during walking and sit-to-stand transfer tasks, by means of clever design of elastic and clutching mechanisms. Finally, AURHA is a prototype aiming at replacing a biological hip after hip disarticulation. This extreme amputation level poses unique scientific challenges in terms of mechanical design and control, that we tackled with innovative solutions. The talk will overview the specific challenges of each leg joint, and will explain how we addressed them with a global bio-inspired approach.

Session Abstract

Positron emission tomography (PET), x-ray computed tomography (CT) and magnetic resonance imaging (MRI) and their combinations (PET/CT and PET/MRI) provide powerful multimodality techniques for in vivo imaging. This talk presents the fundamental principles of multimodality imaging and reviews the major applications of artificial intelligence (AI), in particular deep learning approaches, in multimodality medical image analysis. It will inform the audience about a series of advanced development recently carried out at the PET instrumentation & Neuroimaging Lab of Geneva University Hospital and other active research groups. To this end, the applications of deep learning in five generic fields of multimodality medical imaging, including imaging instrumentation design, image denoising (low-dose imaging), image reconstruction quantification and segmentation, radiation dosimetry and computer-aided diagnosis and outcome prediction are discussed. Deep learning algorithms have been widely utilized in various medical image analysis problems owing to the promising results achieved in image reconstruction, segmentation, regression, denoising (low-dose scanning) and radiomics analysis. This talk reflects the tremendous increase in interest in quantitative molecular imaging using deep learning techniques in the past decade to improve image quality and to obtain quantitatively accurate data from dedicated standalone (CT, MRI, SPECT, PET) and combined PET/CT and PET/MRI imaging systems. The deployment of DL-powered methods when exposed to a different test dataset requires ensuring that the developed model has sufficient generalizability. This is an important part of quality control measures prior to implementation in the clinic. Novel deep learning techniques are revolutionizing clinical practice and are now offering unique capabilities to the clinical medical imaging community. Future opportunities and the challenges facing the adoption of deep learning approaches and their role in molecular imaging research are also addressed.

Session Abstract

Cardiovascular diseases (CVDs) have been a major cause of human mortality in recent years. Therefore, early detection of CVDs is of key importance for reducing not only mortality rates but also the economic burden imposed on healthcare systems by these diseases. The first step in the examination of a patient for CVD diagnosis is almost always auscultation. In this procedure, a physician uses a stethoscope to listen to and evaluate heart sounds obtained directly from the patient’s chest. This method is very easy to perform, yet highly effective for rapid cardiac assessment. However, the effectiveness of auscultation strongly depends on the physician’s expertise. Therefore, the development of automatic auscultation methods has become an active and popular research area in recent years. Naturally, the most important task in designing an automatic auscultation system is the development of methods for the accurate analysis of heart sounds for diagnosing various CVDs. In this talk, we will investigate various methods that combine computational intelligence with signal and image processing techniques for CVD diagnosis. We will begin by examining the mechanisms of the heart responsible for generating heart sounds. We will then briefly review the fundamental concepts of phonocardiography. Following this, we will examine major approaches for analyzing heart sound signals by exploiting their different characteristics in the time, frequency, and scale domains. For this purpose, we will provide a comparative evaluation of selected representative methods from these approaches and highlight their advantages and disadvantages. We will conclude the talk with a discussion of future directions in this research area.