Cerebral Palsy (CP), which occurs in about 2 out of 1000 births, is a group of disorders that affect movement and posture. CP is the result of a lesion/injury in the developing brain, usually before or during birth. CP manifests itself early in life, during infancy or preschool years, with delayed or abnormal motor progress. Currently, there is no treatment for CP beyond some forms of physical therapy, mostly focused on children. In recent years, Brain Computer Interfaces (BCI) have been used successfully to enable persons who are completely immobile to use their electrical brain signals (EEG) for communication and control of objects in their environment. However, this approach has had limited success in persons with CP because most CP individuals are subject to unpredictable body movements (spasms), and the EEG signals capture such movements, contaminating the signals for a particular application. Thus the goal of this project is to explore an alternative platform for EEG-based BCI for cerebral palsy by using a two-pronged approach. First, new techniques will be used to isolate and analyze signals of interest, i.e., to identify the useful signal out of its noisy background. Second, the approach will focus on an individual person with CP (as opposed to a group). A BCI model from a healthy subject will be transferred and adapted to the CP subject. The healthy subject, who except for the fact that he does not suffer from CP, will otherwise be similar in gender, age, cultural background, educational level, and intellectual abilities.
The overall goal of this exploratory research project is to develop a Brain Computer Interface (BCI) for an individual suffering from Cerebral Palsy (CP), i.e., a project focused on one individual considering the particular characteristics of that individual's physical and intellectual abilities. The final objective is to enable this individual to perform some simple actions (e.g., hold a cup, hold an apple) that healthy individuals can perform on a routine basis. The BCI approach is based on capturing brain activity in the form of electroencephalogram (EEG) signals. BCI for CP subjects is particularly difficult due to the involuntary body movements (spasms) that afflict CP subjects, which produce unwanted/unneeded brain signals. To achieve the stated objective, the Research Plan is organized under 4 tasks: 1) compare EEG signals obtained from the CP subject with those from a healthy subject with similar physical and intellectual abilities; 2) compare, for each of the two subjects, the EEG signals obtained in a neutral state to those associated to a cognitive (visual and motor imagining) task; 3) develop a novel machine learning approach based on fuzzy sets/fuzzy logic and transfer learning to obtain a model for the CP subject adapted from the model for the healthy subject; and 4) leverage the previous three tasks to propose a novel, general procedure for BCI for disabled individuals.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
