(364ae) Innovative Approaches to Medical Challenges Using Computational Chemical Engineering Principles

My research interests lie in the field of computer aided engineering with medical applications. I focus on applying Computational Fluid Dynamics (CFD) and Machine Learning (ML) to medically relevant problems.

Research Experience

I have applied CFD to model hemodynamics in the coronary arteries. This study revealed detailed flow fields in arteries, leading to a better understanding mechanisms of coronary artery disease. Coronary artery disease affects 1 in 20 American adults aged 20 years or older and is the leading cause of death in the country. The goal of this project was to create a non-invasive tool to inform clinicians whether surgical intervention is needed based upon data collected from ultrasound scans. By creating a non-invasive method, patients are spared from costly, uncomfortable, and unnecessary invasive surgeries that probe fluid flow directly. Special attention was paid to develop computationally efficient implementations, which are of direct clinical interest. To make the method clinically relevant, results are needed within 10-15 minutes using a desktop computer. The time-sensitivity of the problem led to our work on developing deep learning techniques using graph convolutional networks, to create surrogate models for a simple disturbed flow system. My colleagues and I were able to reduce the batch time for simulations in these systems by a factor of 20. Work is still ongoing to apply these and related methods to real patient arteries.

My experiences with coronary arteries led into new work focusing on the mass transport on the surface of the eye. Dry eye syndrome (DES) is a disease in which patients consistently experience a dry, itchy, or burning sensation in their eyes. DES affects nearly 20 million people in the US. Surprisingly, most DES cases are not caused by a lack of tears. Instead, DES is most commonly caused by disfunction of a thin layer of lipids at the tear-air interface called the tear film lipid layer (TFLL). To better understand the mechanisms underlying DES and treatment with new classes of drugs, my colleague ran a clinical trial to collect videos of TFLL development and stability. My work here has been focused on creating data processing and analysis software to automate the workflow in analyzing video data from eye emissivity studies. In Python, I was able to use artificial intelligence through frame differencing and thresholding in open cv, along with tkinter to create a simple graphical application that accelerated data processing by over 50-fold. We are currently automating this process using convolutional u-nets to perform image segmentation. These data will be used in conjunction with CFD simulations to examine the transport of lipids across the surface of the eye following a blink. These studies will aid in understanding of the etiology of DES and help clinicians quickly diagnose and treat DES.