The development of the next generation of mobile health monitoring systems requires the development of advanced technologies that can tackle a large variety of biosensors: from ECG to 3D/4D interactive ultrasound. Furthermore, there is a requirement for power management solutions that can be used continuously, adapting performance to emergency events and varying channel conditions.

Current image and video processing and communication systems do not address the diagnostic requirements for making efficient use of bandwidth, power, and computational resources. Without re-evaluating and re-designing these systems from a diagnostic quality perspective, there is no guarantee that they will meet the growing requirements of mobile health systems. We examine two different systems that provide us with an appreciation of the engineering complexity in terms of software and hardware.

The first engineering challenge problem looks at the development of new diagnostic software systems for atherosclerosis. From the early diagnosis perspective, we look at the development of relating intima-media texture features to age and sex. Beyond the current use of intima-media thickness (IMT), we demonstrate the measurement of structural changes as measured by multi-scale instantaneous amplitude and instantaneous frequency changes. After plaque formation, we are interested in the development of a stratified stroke risk assessment based on plaque strain estimates from ultrasound videos. To visualize the diagnostic-quality of the extracted video motion, we develop robust motion estimation methods. Beyond the current use of invasive, non-optimized motion estimation, we present a non-invasive, global-optimization framework that is used to provide high-resolution estimates of the strain tensor over atherosclerotic plaques. After stroke occurs, we are interested in communicating emergency ultrasound video through noisy communications channels. The focus on diagnostically relevant regions of interest allows us to develop effective methods that can deliver quality H.264 video at significantly reduced bitrates and high packet drop rates.

The second engineering challenge looks at the development of a new diagnostic hardware approach for emergency systems. For 24-hour health monitoring, we first require the use of a low-power monitoring during normal activity. After an abnormal event, we expect the system to become an effective communications device with high-precision recording and communications. Instead of the current static architecture approaches, we propose the use of a real-time reconfigurable computing architecture that allows dynamic power control and dynamic arithmetic with dynamic precision.

Host: Brendt Wohlberg, 7-6886