Research and Work
Experience
Deep Learning Automatic 3D Segmentation of Mandible from Cone Beam CT for Preoperative Planning
My role involved using Slicer software to combine hundreds of X-ray scans produced by cone beam computed tomography (CBCT) into a 3D model. Many of the patients scanned had metal implants, which interfered with the X-rays, and/or abnormalities due to previous injuries. As a result, I spent the majority of my time correcting the scans and refining the 3D models. These corrected models were then used to train an artificial intelligence algorithm, specifically deep learning, to “assist in future patient diagnosis, minimize operative and postoperative errors, reduce operating time, and improve overall surgical outcomes.”
During my undergraduate studies, I worked at NeuronOff as part of their R&D team for 2.5 years, contributing to several key projects. The majority of my time was dedicated to advancing the manufacturing process and leading a cancer treatment project. My work on this project centered on developing a modified version of the Injectrode—an injectable electrode—that could be used in combination with chemotherapy and radiation therapy to enhance their effects through hyperthermia. I also designed the manufacturing process and associated devices required for implementation, in addition to creating pitch decks for marketing and investors. Formal research has yet to be published due to animal testing limitations, so I’m unable to share further details.
Neuromodulation therapies are often limited by invasive surgeries, high device failure rates, and inconsistent therapeutic outcomes due to off-target effects or insufficient therapy at safe charge injection levels. To address these challenges, NeuronOff’s injectrodes is designed to reduce surgical invasiveness and device failure while providing a high surface area for safe charge injection. This study explores the mechanisms and limitations of increased charge injection through a high surface area Injectrode, formed by winding thin Pt-Ir wire into a flexible helical macroelectrode. In this study, I was responsible for the manufacturing of numerous injectrodes, which were used in testing the charge injection properties.
Currently, trial leads are used to assess a patient’s response to neuromodulation therapies, but these require invasive removal and replacement if permanent leads are approved. To evaluate the Injectrode’s potential as a non-invasive alternative to trial leads, we conducted tests by stimulating rodent sciatic nerves and measuring their movement response. My role involved the design and construction of the testing stand to measure this response without impeding the subjects' movement. Additionally, I manufactured many of the Injectrodes used and assisted in the animal testing.
