Mobile Ultrasound System for Research Beyond Classical Imaging
Handheld ultrasound imaging devices are more commonly used for bed-side diagnostics and emergency imaging. While the commercially available devices already provide excellent imaging quality, they are set up as highly integrated and dedicated systems with limited use in research for new field of applications applying new techniques developed due to lack of extendibility in signal or image processing and special types of ultrasound transducers. The vision of a future and modern version of the stethoscope based on ultrasound imaging seems feasible in lots of applications as the low-cost character of handheld ultrasound systems become more common. We propose an open research system for mobile ultrasound applications that features modern technologies combining the diagnostic imaging with more advanced assistance provided by dedicated machine learning tasks. The handheld and battery powered ultrasound system focusing on low-cost design integrating 32 parallel channels to ensure wide-spread usage and availability to enhance patient care. It features modern ultrafast ultrasound imaging techniques using different types of transducers for abdominal, vessel and MSK imaging. Adaptations to application specific ultrasound probe designs are supported. All transducers can be tracked in position and orientation with modern and low-cost integrated tracking systems providing an option for freehand volumetric imaging. The major advantage of this open ultrasound system is the possibility to program individual imaging and signal processing modes to perform research and evaluation of new technologies in a mobile unit. Such open interfaces for ultrasound signal generation and data processing were usually only provided by cost-intensive and stationary ultrasound research systems which limits their evaluation of new technologies in the field of FAST, critical care applications or even home use applications. We designed and integrated this handheld ultrasound imaging device for use in research and development on modern ultrasound techniques with access to all transmit parameters for ultrasound wave generation and access to raw single transducer element channel data for custom signal processing and receive beamforming reconstruction. The overall processing pipeline also includes image-based analysis and diagnosis assistance based on segmentation and classification tasks. It was successfully used in different applications and research tasks including radio-frequent signal classification monitoring of muscle activity and classification of muscle fatigue in fields like wearable fitness and image-based classification of bladder volume and detection of blood cloths. We are looking forward to using the system in more medical case studies and new fields of applications. To overcome limitations that might arise using just 32 electrical channels to address 32 transducer elements, we are currently working on the integration of application-specific integrated circuits (ASICs) into the transducer itself that perform receive signal multiplexing to address more active elements. This results in a larger field-of-views while maintaining the low-cost aspect of the system design.