Using Image Quality Metrics to Optimize the Design of Integrated Medical Ultrasound ADCs.

Emerging handheld and wearable ultra-sound devices enable diagnosis and long-term monitoring outside clinical settings. They require a low-power, highly complex, locally integrated system to process the RF data. The analog-to-digital converter (ADC) is a critical building block in the receive chain of these systems as it enables digital beamforming and image reconstruction. However, the ADCs currently used in cart-based imaging systems are bulky and consume too much power to be integrated into battery-powered devices. This paper investigates how the area and power consumption of the commonly-used successive approximation register (SAR) ADC can be reduced without negatively affecting B-mode and color-Doppler image quality. A Monte-Carlo (MC) simulation study was performed in which RF data acquired with a phased-array transducer in Field II was digitized using a model of a non-ideal ADC. Five different non-idealities were applied to four commonly used SAR-ADC architectures. B-mode and color-Doppler images were reconstructed from the digitized RF data. The impact of the non-idealities on the image quality was evaluated by means of three image quality metrics: peak signal-to-noise ratio (PSNR), structural similarity index (SSIM), and contrast-to-noise ratio (CNR). The effectiveness of error correction and ways of calibration are also discussed. The results show that both B-mode imaging and color-Doppler imaging are inherently resilient to non-idealities, particularly capacitor mismatch, leading to relaxed ADC requirements and paving the way for more practical in-probe digitization.