Abstract:
A computer-aided diagnosis (CAD) algorithm was designed and implemented on an
Altera Cyclone II FPGA to detect cardiac murmurs from recorded heart signals.
Training and evaluation data sets were obtained from the online “Classifying Heart Sounds Challenge”
sponsored by PASCAL. The detection algorithm calculates the Low Energy Rate (LER)
from a recorded heart signal and performs a binary classification of the sample as either normal or murmur.
The FPGA system interfaces with a commercial digital stethoscope to acquire real time data
as well as a VGA-compatible monitor for visualization and metric reporting.
In addition, basic I/O was developed for user input to the system.
The result is a complete embedded system capable of acquiring and analyzing cardiac data
in real time to detect heart murmurs.
Full report (pdf)
