ABSTRACT
Epilepsy is the commonest neurological disorder in the world. Neurologists analyze seizure activity from recorded electroencephalography (EEG) signals over a long period. This chapter describes detection of epileptic seizures using the random forest (RF) algorithm. Classification is done in software using MATLAB R2019a, then feature extraction, RF training and inference modules are implemented on the EDGE ZYNQ SoC FPGA development board using Verilog HDL in XILINX VIVADO 2019.2. An axis parallel-based decision tree is used to design the RF modules. Accuracy of 99.87%, and 97%, respectively, is achieved in software and hardware implementation. The latency time required to detect epileptic seizures from MATLAB and Verilog HDL is 9 seconds and 144 micro-seconds, respectively. The FPGA-based epileptic seizures detection system is found to work faster and more efficiently than the software implementation.
