Projects and Research Areas

 

Some of the current projects at the BSIA lab are as follows:

Developing clinical decision-making systems

In this line or research, we focus on developing signal processing techniques to analyze physiological information and identify relationships with sources of relevant clinical data. At BSIA lab, we have established active collaborations with national and international clinicians to provide reliable information for effective decision making in some of the world’s most pressing diseases as listed below.

Atrial fibrillation (AF)

AF is the most common sustained arrhythmia and is treated using Radiofrequency (RF) ablation to create multiple RF lesions, which form lines of electrical block that disrupt arrhythmic wavefronts. However, there are several open questions and technology limitations to provide a successful and safe AF ablation. This project investigates merging anatomy and electrophysiology behavior of the atrium to better understand AF phenomenons and refine existing AF ablation techniques. This combined strategy uses real-time and adaptive signal and image processing techniques to characterize the electrical properties of AF and infer anatomical properties of the cardiac substrate.

Infantile spasms (ISS)

ISS is a devastating epileptic syndrome that affects children under the age of 1 year. The diagnosis of infantile spasms is based on the semiology of the seizure and the EEG background characterized by hypsarrhythmia. Since EEG is a key factor in the diagnosis of ISS, misinterpretation could result in serious consequences including inappropriate treatment. Quantitative EEG would enhance our ability to correctly identify infants with ISS and treat them appropriately. Quick and effective treatment of infantile spasms is thought to alter the neurodevelopmental outcome for some children. In this project, our focus is to develop novel algorithms to characterize the relevant electrical abnormality in EEG of children with ISS.

Parkinson's Disease (PD)

PD is a chronic, progressive disorder that leads to observable movement abnormalities including tremor, bradykinesia/akinesia (reduced speed and quantity of spontaneous movement), and gait/balance impairment leading to falls. Levodopa improves these motor impairments, but in advanced disease levodopa can also cause excess involuntary movements (levodopa-induced dyskinesias) that are dose-dependent and become progressively more severe with disease progression. In the majority of patients, fluctuations in response to medications emerge such that patients cycle between “OFF” state (in which medications have worn off, resulting in slow, difficult voluntary movements, re-emergence of tremor, and dystonic dyskinesias characterized by sustained twisted postures of the limbs or body), and “ON” state (in which medications are working, voluntary movements and tremor are improved, but can be accompanied by choreiform dyskinesias, which are jerky involuntary movements of different body parts). In this project, we are developing novel signal processing algorithms to automatically analyze ambulatory gate data and identify the fluctuations in response to medications for PD patients.

Developing Data-driven Signal Analysis Framework for Enhanced Non-stationary Data Analytics

The focus of this research program is to develop new algorithms for effective analysis of non-stationary signals as demanded in many real-world problems. Time-frequency (TF) representation has found wide use in many challenging signal processing tasks including classification, interference rejection and retrieval. Furthermore, TF analysis offers a framework through which we can understand the underlying processes of complex, nonlinear and non-stationary systems. Developing effective feature extraction tools for modeling the TF representation is important for reducing dimensionality and redundancy, and obtaining the essential TF structure of the observed data that is necessary for understanding the data generation mechanism. The goal of this research program is to develop data-driven feature extraction method that will provide increased flexibility and the ability to adapt to the dynamic and time varying nature of real and multi-channel signals.