AI Advances in Cancer Care

One in five people is diagnosed with cancer, according to the World Health Organization.  

Two groups of students in Northwestern Engineering's Master of Science in Artificial Intelligence (MSAI) program recently completed projects aimed at helping cancer patients live longer, healthier lives.   

“The idea of developing a model that could potentially assist oncologists and speed up critical treatment planning resonated with me,” said Ashlesha Ahirwadi (MSAI ’25), who worked alongside Hanna Zelis (MSAI ’25) and Narasimha Karthik Jwalapuram (MSAI ’25) on one of the practicum projects. “At the same time, I knew we were stepping into one of the most technically challenging areas — 3D medical image analysis — where small gains in precision or sensitivity could make a big difference for patient care.”  

The two projects were done in partnership with Troy Teo, an instructor of radiation oncology at Northwestern's Feinberg School of Medicine. Both are examples of classroom knowledge turned into practical work, with the hope of helping some of the more than 20 million new cancer patients diagnosed globally each year.  

Currently, treatment planning tools are often slow, manual, and lack consistency. They require physicians to analyze hundreds of 3D CT scans per case, a labor-intensive process prone to human error. These mistakes can result in either missing parts of a tumor or exposing non-tumor cells to radiation. 

Ahirwadi and her teammates’ solution, dubbed PulmoScanAI, uses AI to automatically identify the bounds of lung tumors from those scans and provides fast, accurate, reliable information to improve radiologists’ efficiency and consistency in treatment planning.  

The new model came from the lessons students learned during their first two quarters in the MSAI program.  

“From a technical standpoint, I drew on deep learning theory, computer vision techniques, and optimization strategies we’d explored in class," Ahirwadi said. "From a systems perspective, I applied principles of problem framing, metric selection, and deployment planning.”  

Read how MSAI director Kristian Hammond thinks AI can improve medicine.  

Sree Nimmagadda (MSAI ’25) worked with Vrishani Shah (MSAI ’25) and Divyanka Thakur (MSAI ’25) on the second project. Their work was part of a larger pipeline aimed at automating the entire radiation oncology treatment planning process. The task was to predict the segmentation masks of surrounding organs that are at risk of exposure to radiation treatment doses. 

The ultimate goal was to optimize the correct angle to deliver treatment doses so healthy organs would not be affected. 

Nimmagadda said the biggest obstacle was organizing a massive amount of data. 

“We approached it with efficient data preprocessing,” she said. “Instead of loading all the data at once, we chunked our CT scans into cubes to quickly load them and stitched them back once loaded.” 

The students applied classroom lessons to their practicum projects, and now they expect to be able to take lessons learned from the practicum and apply them to the rest of their time in MSAI. Nimmagadda said she expects to apply what she learned to her professional life long after she graduates. 

“This project challenged us to apply AI in a workplace setting to deliver meaningful results,” she said. “I have become more organized in my workflow, better at communicating my work to people who may not have an AI background, and more confident in my ability to solve real-time problems.”  

Ahirwad agreed. 

“I came away with a deep appreciation for the role of collaboration between AI practitioners and domain experts,” she said. “The most effective solutions are built at that intersection.”