Design And Implementation Of FPGA-based Cloud Accelerator For Detection And Analysis Of Large-scale ECG Data

The current healthcare system is facing the challenge of growing numbers of population, increased demands from patients and shortage in human resource. Traditional healthcare cannot satisfy the medical demands of people. Fortunately, cloud medicine, having involved computer technology, communication technology, big data and medical technology provides a new solution for the challenge. The fast growing of wearable device in recent years brings hope for develop of cloud medicine, which also raises higher requirements for cloud medicine platform in terms of fast signal analysis and intelligent diagnosis ability.According to characteristic and demand of cloud medicine platform, this work proposed a prototype of real time FPGA-based Accelerator for ECG analysis and diagnosis, including wave detection, feature extraction and preliminary diagnosis. The streaming and pipelined architecture employs a novel data mining algorithm to diagnose ECG. The system is composed of ECG feature extraction, association rules mining and feature matching.Achievements of this work are as follows:1) This work proposed a hardware oriented algorithm based on association rules mining to diagnose ECG. The algorithm is suitable for parallel and pipeline processing.2) We implemented a pipelined and parallel architecture which can detect ECG feature and provide diagnosis results in real time. The implementation has a small delay, high throughput and good scalability.This work provides cloud medicine platform with a hardware solution which can process massive ECG signal in real time and lays a foundation for the future research in this field.