Binding Problem: Visual Perception
Our research has been focused on the field of visual perception, specifically on the issue of “Binding Problem”, i.e. the issue explaining how brain correctly integrate the visual features of objects such as color, shape, orientation and location.
Movement Imagery
ERD (Event-related desynchronization) corresponding to the limb movement imagery is one of the main approaches in movement imagery detection. In order to make ERD be more accurate and applicable, a simple approach is developed in BACE Lab namely weighted ERD or WERD. The approach has been successfully applied to a set of acquired EEG signals gathered by a 16-channel, 1kHz device in an isolated, dark room. Moreover, the proposed approach has led to similar successful results by applying to standard data-sets.
EEG-based Stuttering detection
Speech disorder or stuttering is associated with a mouth locking and a sudden interruption during the normal speech by repeating, and pulling up to the sounds, syllables, words and even phrases. Sometimes, it has destructive impact in stutterer’s social relationships. Also, it has factors such as isolation, depression, etc. So, Control and treatment of this disorder, especially in children is one of the goals of physicians and researchers.
Finding the relationship between speech disorders and brain signals and consequently, obtaining a pattern corresponding to the stutter is in interest of us in BACE Lab. In this direction we has done much effort in offline detection of stringing, however the way is open for online approaches.
EMG-based motion intention detection
Myoelectric control has a key role in human-machine interface applications such as orthosis control and teleoperation. Myoelectric signals are bio signals that are detectable from surface of the skin, and contain useful information about user’s moving intention. This project presents a methodology to estimate elbow joint angle from muscle’s data using neural network (NN). Proposed methodology can be expanded to estimate any joint angle by recording muscle activities concerning with the joint. Several data sets will be recorded and processed to train and test the NN. The trained network will use to predict elbow angles for individual data sets. We predict Results will show that trained network can estimate joint angle with an acceptable performance.
Mechanical Pulse Signal Analysis
Nowadays, non-invasive approaches for diagnosis of the illnesses have become a necessity for medical diagnosis. The goal of this study is to present a method for signal analysis corresponding to the mechanical pulse signals in order to perform medical diagnosis in regard of some diseases.
In the first step, the data relating to pulse pressure in the finger of left hand for 45 healthy persons and 45 CAD persons acquired by TFM was collected. Using AI methods, we could classify the two categories with the accuracy rate of 85%. Fluid modelling of blood flow and mechanical aspect of pulse generation is the next step of the research.
