PROJECTS
Senior Design Project
Transrectal and Transperineal Ultrasound Mixed Reality Simulation
Teammates: Justina Chan, Alex Dluzneski, Youssef Elbanna, Marye Lee, Henry Nguyen, Brian Nazareth, Giancarlo Tejeda
I collaborated as a member of MediSims team on design and development of a mixed reality prostate biopsy simulation. We performed competitive analysis on existing simulators and observed clinical biopsy procedures to define needs. Cross-team collaboration took place between physical model team, hardware team, and software team. Under the software team, I co-developed a virtual environment of the internal human anatomy that included prostate, urethra, and other neighboring organs in Unity. We also implemented a scoring algorithm to keep track of participant's performance. As a team, we presented and ran a demonstration of the simulator prototype during the Design Competition at the 2019 BMES Annual Meeting in Philadelphia, PA.
Mention: BME STUDENTS WIN THE BMES STUDENT DESIGN COMPETITION
Data Science Project
Predicting Activity & Calories Burnt Using Variables from Apple Watch
Teammate: Geetika Singh
Apple Watch measures and collects heart beat (bpm), calories burned (kcal), steps walked (count) and distance travelled (km) but no data about when a particular observation was recorded, i.e. if the person was sleeping, exercising or was just involved in day-to-day activities. The goal of this project is to develop a classification model using Support Vector Machines (SVM) and Decision Trees (DT) to differentiate activities based on the available predictors. A secondary goal is to model calories burned using Apple Watch-collected data and user information as predictors and test the model using a Fitbit-generated data set to observe if any cross-platform competency exists.
Deep Learning Project
Gesture Recognition on Low Resolution American Sign Language (ASL) Images
American Sign Language is widely used among the Deaf Community in North America. Deep learning techniques can be used to help this community by creating a system that can in real time take an ASL sign and interpret it in real time, bridging the communication gap between ASL users and non-users. In this paper, some preliminary work is done to build a Convolutional Neural Network (CNN) model and test its performance on identifying hand gesture images of letters of the alphabet. The results show that the current model performs at over 85% accuracy
on images with various resolutions.