Many types of natural (including our bodies) phenomena can be observed by sensors and turned into continuous electrical signal, either by direct observation (like heart rates) or through external excitation (like ultrasound). The sampled and digitized signal can be studied and analyzed by the digital signal processing (DSP) to produce meaningful data and images to assist physician in proper diagnosis.
DSP implementation can be done on a dedicated ASIC, FPGA, GPU, DSP chips, embedded processor with DSP extensions or general purpose processors. Each has its advantages and trade-offs in terms of performance, power, production cost, development resources and flexibility. A dedicated ASIC can give the best possible performance, power and production cost, ideally for a mass market portable medical device. However it demands very high development resources while provide the least flexibility. A GPU might provide high performance and great flexibility but the power consumption and cost might be prohibitive for many applications. But for a high end ultrasound machine, where new algorithms are still been developed, GPU is a perfect candidate. At A&W, our engineers have implemented algorithms for all of platform mentioned. We can do a comprehensive study and provide you with the best implementation option.
DSP algorithms are developed in environment like MATLAB, which provides a rich library of design, visualization and analysis tools. Under this environment developer can quickly try out different DSP techniques to obtain the desired signal or image. However such environment is cumbersome to use during the implementation phase and with the real time data collection. At A&W, we have developed a light weight environment called ADE (algorithm development environment) that includes many of the essential tools needed for visualization and data analysis. It can be packaged as an executable that can be deployed quickly to a clinician laptop and works well with the portable medical data collection hardware.
Medical signal processing revolutions are continuously discovered in research institutions and large medical companies. Many of these innovations are improvements on the current modalities, providing better resolutions at faster data rate. While others center around new modalities and new measurement techniques. One of the game changing innovation is machine learning with medical data and images. Algorithm breakthrough not limited providing better data or image, but to teach computer understand the medical data and images. At A&W we are building a cooperation network with partners that can bring these innovations to you.