Cognitive Sciences and Psychology
Title : | Visually Evoked EEG Classification and Perceived Image Regeneration using Deep Learning for Brain Fingerprinting |
Area of research : | Cognitive Sciences and Psychology |
Focus area : | Neuroscience, Forensic Science |
Principal Investigator : | Dr. Shamama Anwar, Birla Institute Of Technology, Mesra, Jharkhand |
Timeline Start Year : | 2023 |
Timeline End Year : | 2026 |
Contact info : | shamama@bitmesra.ac.in |
Details
Executive Summary : | Forensic Science is an important constituent of the criminal justice system. Forensics experts scrutinize and analyze evidence from crime scenes and other locations in terms of developing some important cues towards objective that can aid in the investigation and criminal prosecution process. Brain fingerprinting is an impartial, scientific method for detecting hidden information stored in the brain by non-invasively measuring electroencephalographic (EEG) brain responses, or brainwaves, with sensors placed on the scalp. The technique entails displaying phrases, expressions, or images enclosing pertinent description about a crime scenario on a computer screen in succession with other, irrelevant stimuli. The brain's responses to stimuli are analysed. When the brain interprets information about particular aspects, the computer tries to analyse the brain responses that can discern distinctive feature of brainwave patterns. When a person recognises something as significant in the current context, the response is characterized by a specific brainwave pattern. In a brain fingerprinting test, words or pictures relevant to a crime scene are presented on a computer screen, in a series with other, irrelevant words or pictures. The brainwave responses of subjects to these stimuli are measured non-invasively with a headset equipped with EEG sensors. This concept can further be enhanced by allowing the generation of these perceived images based on the EEG acquired while viewing the image. Such a visual image reconstruction from the brain has been a hot and explosive topic in recent years, and the neuroscience research has found substantiation based on the possibility of decoding neuroimaging data, i.e., how the human brain functions. Simultaneously, the recent renaissance of deep learning combined with a strong interest in the scientific and research community on generative methods has made it possible to learn data distribution from noise to produce realistic image. The major challenge is to regenerate images with improved quality from the input data. Therefore, it is essential to transmit input data with crucial cues and information about the visual content of images. This technology has the potential to open new opportunities in marketing, Brain-Computer Interface (BCI), healthcare and legal and forensic sector as well. In addition to accurately recalling many details of the crimes people witnessed, eyewitnesses must frequently recall the faces and other important attributes of the perpetrators of those crimes. Eyewitnesses are frequently asked to describe the perpetrator to law enforcement and then to make identifications based on photographs. Hence inline with this, the proposal aims at developing an EEG classification and Image regeneration technique that aims to classify the EEG signal based on the object a user is viewing on a computer screen and further attempts to regenerate a similar image based on the cue received from the brain EEG. |
Co-PI: | Dr. Vandana Bhattacharjee, Birla Institute Of Technology, Mesra, Jharkhand-835215 |
Total Budget (INR): | 24,15,705 |
Organizations involved