TITLE:
Flexible Analytic Wavelet Transformation Based Brain Signals Extraction and Classification
AUTHORS:
Shams Al Ajrawi, Mahasweta Sarkar, Ramesh Rao, Sarah Jawad, Mohammed Mohammed Raoof
KEYWORDS:
Brain Computer Interfaces, ECoG Signals, Cross Correntropy, LS-SVM, FAWT, Log Entropy, SURE Entropy
JOURNAL NAME:
Journal of Information Security,
Vol.17 No.3,
July
30,
2026
ABSTRACT: Brain computer interfaces (BCIs) have emerged as communication technologies to re-mobilize individuals with paralysis, neurological diseases, injuries or limb loss. BCI systems detect brain signals and use extraction and classification algorithms to convert the detected signals into useful outputs that can be used to control external devices. Designing and developing BCI systems face several challenges, such as session-to-session transfer, subject-to-subject transfer, non-stationary signals required for adaptivity, idle signals for motionless subjects, and continuous data collection that can be used to expand BCI systems for a broader spectrum of subjects. This paper presents a novel approach to detect and classify the Electrocorticography (ECoG) brain signals. The Flexible Analytic Wavelet Transformation (FAWT) is used to extract entropy features and the Least-squares Support-Vector Machines (LS-SVM) is utilized to classify the activities. ECoG signals are obtained by directly recording from the surface of the cerebral cortex. This provides high-resolution data and stability in signals, in contrast to Electroencephalography (EEG) signals. Three entropy features were extracted from the ECoG signal, which are log energy entropy, SURE entropy, and cross correntropy. Later in this paper, the feature extraction section provides a detailed definition of each feature. Our proposed approach was evaluated by carrying out testbed experiment and by using the dataset I of the public BCI competition III data. A classification accuracy of 95% was achieved during the testbed evaluation. A classification accuracy of 93.02% was achieved using the dataset. Our proposed algorithm was implemented in MATLAB.