FPGA Implementation of CNN Algorithm for Detecting Malaria Diseased Blood Cells

Abstract

In this study, Field Programmable Gate Array (FPGA) implementation of Convolutional Neural Network (CNN) for classification of malaria diseased cell is done. The hardware is designed and implemented on Xilinx Zynq-7000 FPGA using Very High–Speed Integrated Circuit Hardware Description Language (VHDL). In accordance with this purpose, Convolutional Neural Network (CNN) classification method used by image processing to make it easier for experts to comment on diseased cells. The classification method allows us to make a simpler interpretation by classifying complex images. Thanks to this research, it facilitates early diagnosis using image processing in the medical field as soon as reduces death and treatment costs. According to the experimental results, the accuracy rate for finding malaria diseased cell using CNN method is 94.76% for 200 8x8 binary images. The average execution time of CNN algorithm using Matlab on desktop PC is 174 microseconds. On the other hand, the maximum achievable frequency on Zynq FPGA is 168MHz (i.e. the longest critical path is 5.93 nanoseconds). The occupied area of CNN on Xilinx Zynq FPGA is only 783 six-input Look Up Tables (LUTs) of 17600, which is about 4.34% of Xilinx Zynq-7000 (XC7Z010-1CLG400C) FPGA.