AI’s Role in Breast Cancer Detection
Artificial intelligence is increasingly influencing the landscape of cancer treatment, particularly in breast cancer detection.
At The Ohio State University Comprehensive Cancer Center, researchers are utilizing AI to identify patients who might develop lobular breast cancer, which can often be elusive in standard screenings.
Understanding Lobular Breast Cancer
As the most prevalent cancer among women, breast cancer is also the second leading cause of related fatalities. Within the U.S., lobular breast cancer is noted for being especially aggressive and can be tricky to diagnose, contributing to about 10% to 15% of all breast cancer cases.
This type of cancer doesn’t cluster into a distinct tumor; instead, it spreads as chains of cells, making it less visible on mammograms. Unfortunately, this means it often isn’t detected until it has advanced to other areas of the body. Even a decade after a diagnosis, lobular cancer has potential for recurrence.
Dr. Alia Roy from OSU emphasizes the urgent need for improved identification methods. “We really need better tools to predict which patients are at high risk,” she noted.
Adding to this complexity, nearly 40% of women over 40 have dense breast tissue, complicating detection further and elevating breast cancer risks.
Despite differences in how lobular and the more frequently encountered invasive ductal carcinoma develop and respond to treatment, current oncological guidelines don’t differentiate between the two. Dr. Roy points out that genomic tests often yield uncertain results in cases of lobular cancer, making it challenging for doctors to formulate effective treatment plans.
Advancements in Cancer Technology
Roy reiterated the challenges faced in diagnosing lobular breast cancer and identifying patients at high risk of recurrence. She shared with FOX News Digital that they’re applying AI technology to bolster early detection and identify those likely to experience a recurrence.
By harnessing AI models alongside digital pathology images, healthcare professionals aim to recognize biomarkers in patients considered high-risk. The intention is to develop a scoring system capable of predicting potential recurrence within the next decade.
This innovative AI tool is still under development, but clinical trials are in progress.
“Once fully developed, we hope this AI tool will assist in identifying at-risk patients across the board for lobular breast cancer,” Roy explained. She believes that understanding the likelihood of cancer returning within a specific timeframe can greatly impact monitoring strategies.
Considerations and Limitations
Dr. Harvey Castro, an ER physician from Texas who specializes in AI, spoke about the implications of the OSU findings, acknowledging it as a significant stride in AI application but also cautioning about ongoing hurdles. He pointed out that many AI systems perform well in controlled lab settings yet often struggle in real-world clinical environments.
A notable obstacle is the training of AI models on outdated data. “Medicine evolves quickly, and algorithms built on past images might miss current developments. This is referred to as temporal drift,” he stated.
Castro noted that dense breast tissue continues to be a major challenge for AI. The same density obscuring tumors from experts can confuse algorithms, especially among different demographic groups.
While AI is not expected to replace radiologists, it will reshape how they operate. It’s essential to ensure these tools are rigorously tested in diverse, real-world populations before being incorporated into standard practices.
