Breast Cancer Screenings that Use AI May Reduce Unnecessary Testing

In a collaborative effort between researchers at Washington University School of Medicine in St. Louis and Whiterabbit.ai, a Silicon Valley-based tech startup, a recent study suggests that integrating artificial intelligence (AI) into the analysis of mammograms could enhance breast cancer screening processes.

The study, published in the journal Radiology: Artificial Intelligence, proposes that AI could mitigate false positives while ensuring cancer cases aren't overlooked.

The research team trained the AI model using a substantial dataset of 123,248 2D digital mammograms, including 6,161 showing cancer. After training, the model was validated and tested on three independent sets of mammograms from institutions in the U.S. and the UK.

The study involved comparing the outcomes of conventional diagnostic procedures with those simulated with AI assistance. For instance, in the largest dataset of 11,592 mammograms, AI identified 34.9% as negative. If these negative mammograms had been removed from radiologists' workload, the simulation suggested a substantial reduction in callbacks for diagnostic exams and biopsies without compromising cancer detection rates.

Ultimately, the study highlights the potential of AI to enhance breast cancer screening by improving workflow efficiency, minimizing unnecessary interventions, and maintaining accurate identification of cancer cases.

This week Feel For Your Life launched a NEW AI feature, BreastFriendAI. It’s a digital chat device that creates a customized advocacy plan by preparing you for screenings, explaining your pathology reports, and assisting you through follow-ups, treatments and second opinions. Check it out right here.