Design of Hardware Accelerators for Optimized and Quantized Neural Networks to Detect Atrial Fibrillation in Patch ECG Device with RISC-V.

Atrial Fibrillation (AF) is one of the most common heart arrhythmias. It is known to cause up to 15% of all strokes. In current times, modern detection systems for arrhythmias, such as single-use patch electrocardiogram (ECG) devices, have to be energy efficient, small, and affordable. In this work, specialized hardware accelerators were developed. First, an artificial neural network (NN) for the detection of AF was optimized. Special attention was paid to the minimum requirements for the inference on a RISC-V-based microcontroller. Hence, a 32-bit floating-point-based NN was analyzed. To reduce the silicon area needed, the NN was quantized to an 8-bit fixed-point datatype (Q7). Based on this datatype, specialized accelerators were developed. Those accelerators included single-instruction multiple-data (SIMD) hardware as well as accelerators for activation functions such as sigmoid and hyperbolic tangents. To accelerate activation functions that require the e-function as part of their computation (e.g., softmax), an e-function accelerator was implemented in the hardware. To compensate for the losses of quantization, the network was expanded and optimized for run-time and memory requirements. The resulting NN has a 7.5% lower run-time in clock cycles (cc) without the accelerators and 2.2 percentage points (pp) lower accuracy compared to a floating-point-based net, while requiring 65% less memory. With the specialized accelerators, the inference run-time was lowered by 87.2% while the F1-Score decreased by 6.1 pp. Implementing the Q7 accelerators instead of the floating-point unit (FPU), the silicon area needed for the microcontroller in 180 nm-technology is below 1 mm2.

A New Configurable Wireless Sensor System for Biomedical Applications with ISO 18000-3 Interface in 0.35 µm CMOS.

This article presents a new configurable wireless sensor system. The system is used to perform amperometric measurements and send the measurement data to a handheld reader using a wireless transponder interface. The two-chip sensor system was implemented in a 0.35 µm CMOS technology. The system consists of an integrated nano-potentiostat that performs the actual measurements and an ISO 18000-3 compliant frontend that enables wireless telemetric data transmission and powering of the entire sensor system. The system was manufactured in combination with a chronoamperometric glucose sensor which allows the measurement of the glucose content in tear fluid and thus a non-invasive determination of the blood sugar level. For a range of sensor currents from 0.1 µA to 10 µA, the potentiostat achieved an accuracy of better than 5 % with a total power dissipation of less than 600 µW. With the realized antenna geometry a wireless communication distance of more than 7 cm has been achieved.

The technological basis for adaptive ion beam therapy at MedAustron: Status and outlook.

The ratio of patients who need a treatment adaptation due to anatomical variations at least once during the treatment course is significantly higher in light ion beam therapy (LIBT) than in photon therapy. The ballistic behaviour of ion beams makes them more sensitive to changes. Hence, the delivery of LIBT has always been supported by state of art image guidance. On the contrary CBCT technology was adapted for LIBT quite late. Adaptive concepts are being implemented more frequently in photon therapy and also efficient workflows are needed for LIBT. The MedAustron Ion Beam Therapy Centre was designed to allow the clinical implementation of adaptive image-guided concepts. The aim of this paper is to describe the current status and the potential future use of the technology installed at MedAustron. Specifically addressed is the beam delivery system, the patient alignment system, the treatment planning system as well as the Record & Verify system. Finally, an outlook is given on how high quality X-ray imaging, MR image guidance, fast and automated treatment planning as well as in vivo range verification methods could be integrated.