Comparative Evaluation for PCA and ICA on Tongue-Machine Interface Using Glossokinetic Potential Responses

Abstract

The tongue-machine interface (TMI) between the paralyzed person and computer makes it possible tomanage assistive technologies. Severely disabled individuals caused by traumatic brain and spinal cordinjuries need continuous help to carry out everyday routines. The cranial nerve is arisen directly fromthe brain to connect the tongue that is one of the last affected organs in neuromuscular disorders.Besides, the tongue has highly capable of mobility located in the oral cavity, which also providescosmetic advantages. These crucial skills make the tongue to be an odd organ employed in the humanmachine interfaces. In this study, it was aimed to investigate 1-D extraction and develop a novel tonguemachine interface using the glossokinetic potential responses (GKPs). This rarely used bio-signs areoccurred by contacting the buccal walls with the tip of the tongue in the oral cavity. Our study, named asGKP-based TMI measuring the glossokinetic potential responses over the scalp, may serve paralyzedpersons an unobtrusive, natural, and reliable communication channel. In this work, 8 males and 2females, aged between 22-34 naive healthy subjects, have participated. Linear discriminant analysis andsupport vector machine were implemented with mean-absolute value and power spectral density featureextraction process. Moreover, independent component analysis (ICA) and principal component analysis(PCA) were used to evaluate the reduced dimension of the data set for GKPs in machine learningalgorithms. Furthermore, the highest result was obtained at 97.03%.