Vishal Balaji Sivaraman

About

My passion for artificial intelligence, particularly in healthcare, began during high school after a near fatal experience caused by a retinal tear from a cricket ball strike. This incident revealed the transformative power of combining healthcare with AI, setting me on a path to develop innovative AI driven solutions. During my recovery, I witnessed how AI significantly aided in my healing, sparking my desire to create AI applications that assist doctors and improve patient care. During my undergraduate studies in India, I developed two AI frameworks: Emotion Detection for Individuals with Neurological Disorders Using Speech (EDNUS) and an AI based speech assessment tool. EDNUS assists clinicians in diagnosing mental health conditions through speech signals, while the speech assessment tool identifies and validates speech disfluencies. Both projects resulted in publications in reputable journals, enhancing my research profile and further fueling my passion for integrating AI in healthcare. Continuing my academic journey, I pursued a master’s degree in Electrical and Computer Engineering at the University of Florida, where I met Dr. Wei Shao, an Associate Director of the Intelligent Critical Care Center and Assistant Professor in the Department of Medicine. He also serves as the Principal Investigator of the MIRTH AI lab, which leads the development of AI solutions including custom large language models for medical diagnosis and prognosis. Under his mentorship, I reconnected with my passion for AI in healthcare, working as a full time graduate research assistant at the start of my master’s. For nearly 2.5 years, I worked on two research projects under Dr. Shao’s guidance: RetinaRegNet and multiclass aorta and aortic branch segmentation using CIS UNet. RetinaRegNet is a dual stage zero shot retinal image registration framework that registers two dimensional retinal scans across modalities without model finetuning. This framework produced exceptional results across three challenging datasets: high resolution Retcam fundus scans, ultra widefield fundus scans, and low contrast laser speckle flowgraphy scans captured from patients diagnosed with uveal melanoma, a rare eye cancer. The research findings were published in the journal Computers in Biology and Medicine (IF=7.0) , ranked in the top nine percent for Computer Science Applications. This project also served as the foundation for my master’s thesis, which I successfully defended. Simultaneously, I contributed to another project focused on developing a robust three dimensional segmentation model based on the SwinUnetr model, titled Context Infused Swin UNet. This model accurately segments the aorta and its branches from high resolution CTA scans, enabling clinicians to plan and perform endovascular surgeries. The model outperforms the SwinUnetr by balancing segmentation accuracy and computational efficiency through the novel Context Aware Shifted Window Self Attention bottleneck block. The research findings were published in the journal Computerized Medical Imaging and Graphics (IF=5.4). Additionally, I contributed to data curation and preprocessing by manually annotating a databank of approximately 100 three dimensional CTA volumes, each containing between 578 and 800 slices, for the AortaSeg24 Challenge hosted by our lab, with participation from 121 teams worldwide. The findings from this challenge were submitted as a research paper to the journal Medical Image Analysis (IF=10.7), currently under review, with my contributions recognized through first co authorship. My academic pursuits have been complemented by corporate experience. As an Associate Systems Engineer Intern at Northrop Grumman, I contributed to the Akida Brainchip initiative, developing autonomous dynamic scripts to train and test advanced AI models on a low power neuromorphic edge AI processor. This experience combined cognitive computing with neuromorphic processing, deepening my technical expertise and understanding of practical AI applications in healthcare. Outside academia and research, my interests in photography and fitness help me maintain balance. Photography allows me to capture the scenic beauty of Florida, while swimming and gym workouts keep me physically and mentally sharp, supporting my intellectual growth. Looking ahead, I am eager to continue my academic journey by pursuing a Ph.D. in Electrical and Computer Engineering with a focus on developing dynamic, generalizable zero shot registration models aimed at transforming healthcare. My goal is to assist clinicians in real time with early disease diagnosis and prognosis, enabling the creation of actionable, personalized treatment plans that enhance patient outcomes.