AI Scoring Revolutionizes Mammography Turnaround Times Amid Radiology Staff Shortages
CHICAGO — The future of breast cancer detection may be getting a high-tech boost, thanks to artificial intelligence (AI). Findings presented on December 1 at the Radiological Society of North America (RSNA) 2024 conference suggest that AI scoring systems could significantly streamline mammography turnaround times, offering a potential solution to the ongoing challenge of staff shortages in breast imaging.
Dr. Gopal Vijayaraghavan, a radiologist from UMass Memorial Health in Worcester, Massachusetts, shared his team’s groundbreaking research during the conference. Their study explored how AI can prioritize mammography interpretation lists, ultimately reducing the time it takes to diagnose breast cancer. “It’s important in that a delay in breast cancer diagnosis can be very anxiety-provoking for the patient,” Vijayaraghavan explained. “We wanted to take measures in order to reduce that.”
The Growing Challenge of Staff Shortages in Radiology
Staff shortages in radiology have become a pressing issue, with increasing imaging volumes only adding to the strain. Many radiology practices are struggling to keep up, and the need for innovative solutions has never been greater. Researchers are now turning to AI as a potential game-changer in addressing these challenges.
Dr. Vijayaraghavan’s team conducted a year-long study between 2023 and 2024 to evaluate the effectiveness of AI in reducing turnaround times for breast cancer detection and diagnosis. The study utilized a commercial AI system called Transpara AI, developed by ScreenPoint Medical, and analyzed data from a staggering 46,782 breast cancer screening exams. Of these, 80% were digital breast tomosynthesis (DBT) exams, while the remaining 20% were digital mammography exams.
How the Study Was Conducted
The research team enlisted eight fellowship-trained, board-certified breast radiologists to interpret the exams. These radiologists, whose experience ranged from two to 30 years, were blinded to the AI system’s results to ensure unbiased analysis. The AI system assigned scores to each screening exam, ranging from 1 to 10, with higher scores indicating a greater risk of malignancy. Here’s how the scoring system worked:
- Scores of 10: Elevated risk of malignancy
- Scores of 8 and 9: Intermediate risk
- Scores below 7: Low risk
Out of the 192 cancers diagnosed during the study, 182 cases (94.8%) received scores of 8 or higher, accounting for 45.6% of all screens. Meanwhile, scores of 7 or higher identified 191 of the 192 cancers, covering 58% of the screens. Notably, the AI system did not diagnose any cancers for scores of 5 or below.
AI’s Strengths and Limitations
While the AI system showed remarkable accuracy in identifying high-risk cases, it also demonstrated a tendency to assign higher scores than necessary. For instance, the system assigned a score of 10 to 7,302 screens, but radiologists recalled only 1,424 of these cases (19.5%). This discrepancy highlights the need for human oversight in interpreting AI-generated scores.
Despite this limitation, Dr. Vijayaraghavan emphasized the value of AI in clinical decision-making. “Implementing the AI could prioritize cases that are most likely to have an abnormality that could lead to a breast cancer diagnosis,” he said. “Not that this is going to reduce the workload, but then you allot priorities to cases that are more likely to have a breast cancer.”
Real-World Impact and Future Directions
One of the most significant outcomes of the study was the ability to use AI scores to optimize workflows. By setting a threshold score of 7 or higher, the team was able to prioritize cases more effectively. This change also allowed the institution to subcategorize BI-RADS 4 lesions, expediting the time from diagnosis to biopsy.
Looking ahead, Dr. Vijayaraghavan and his team plan to conduct follow-up studies to evaluate cancer detection rates and recall rates after one full year of AI implementation. These future studies will provide valuable insights into the long-term impact of AI on breast cancer screening and diagnosis.
Why This Matters
The integration of AI into radiology practices could be a game-changer, especially in the face of ongoing staff shortages. By streamlining workflows and prioritizing high-risk cases, AI has the potential to improve patient outcomes and reduce the anxiety associated with delayed diagnoses. While it’s not a replacement for human expertise, AI serves as a powerful tool to support radiologists in their critical work.
As the field of radiology continues to evolve, the findings presented at RSNA 2024 underscore the importance of embracing innovative technologies like AI. With further research and refinement, AI could become an indispensable part of breast cancer detection and diagnosis, ultimately saving lives and improving the quality of care for patients worldwide.
Originally Written by: AuntMinnie Staff
