ĐÔ°źÌìÌĂ

By Dave DeFusco

At the Katz School’s Graduate Symposium on Science, Technology and Health, Nikhil Deekonda, a student in the M.S. in Artificial Intelligence, unveiled the idea behind an AI-powered mobile app, called LungAware, designed to detect and classify lung cancer from CT scans.

“The goal of LungAware is simple but powerful—make early lung cancer detection fast, accessible and understandable,” said Deekonda. “We’re combining AI’s pattern recognition capabilities with a user-friendly mobile platform to support better health outcomes around the world.”

Lung cancer is among the leading causes of cancer-related deaths worldwide. Its prognosis improves dramatically with early detection, but accurate diagnosis often requires skilled radiologists and expensive infrastructure. In many clinical settings, especially rural or under-resourced ones, patients face delays, misdiagnoses or missed diagnoses altogether.

Deekonda saw an opportunity to address this with LungAware, a lightweight mobile application built around a sophisticated convolutional neural network. The app classifies CT scan images into one of three categories: malignant, benign or normal—and does so in seconds.

“The real innovation here isn’t just in the algorithm—it’s in the delivery,” said Deekonda. “Anyone with an Android phone and a CT scan can access AI-powered insights in real time.”

LungAware’s convolutional neural network—a machine learning model particularly well-suited for analyzing medical images—uses a series of layers to extract features, reduce noise and ultimately make a probabilistic prediction about whether a scan shows evidence of cancer.

What sets this app apart is its emphasis on interpretability. Using a method called Grad-CAM (Gradient-weighted Class Activation Mapping), the app highlights regions of the scan that influenced the AI’s decision.

“This isn’t a black box,” said Deekonda. “Users can actually see where the model is ‘looking,’ which gives clinicians confidence and helps bridge the gap between AI and human judgment.”

The app has a built-in safety check that makes sure the image being uploaded is actually a lung CT scan. This helps avoid mistakes and makes the app easier to use. To teach the app how to spot signs of lung cancer, he used a public collection of 1,190 CT scan images that had already been labeled by experts. He then trained the app to get better and better at recognizing patterns, so it can now tell with high accuracy whether a scan shows cancer or not.

Deekonda designed the app specifically for Android devices, recognizing their dominance in global smartphone markets, especially in developing countries. The intuitive interface allows users, clinical or otherwise, to upload a CT scan image, receive a classification and view the Grad-CAM overlay. This simplicity is intentional.

“We wanted this tool to be usable not just by radiologists, but by general practitioners, nurses and even patients themselves,” he said. “That’s how we close diagnostic gaps.”

Dr. Honggang Wang, professor and chair of the Department of Graduate Computer Science and Engineering, praised the project for its thoughtful integration of AI and healthcare delivery.

“Nikhil’s work reflects the kind of innovation we hope to inspire at the Katz School,” said Dr. Wang. “By focusing on real-world impact—on accessibility, transparency and speed—LungAware shows how artificial intelligence can serve the public good. It’s not just about building smart tools; it’s about solving urgent problems.”

Despite its promise, Deekonda acknowledges that LungAware is still in its early stages. Challenges remain, including the need for a more diverse dataset to improve generalizability, better computational efficiency for real-time use and compliance with healthcare regulations.

“We plan to integrate features like AI-generated medical reports, secure account access and expanded diagnostic capabilities beyond just lung cancer,” said Deekonda. “Eventually, LungAware could become a full respiratory screening suite, helping clinicians catch not just cancer but other serious conditions early.”