Artificial Intelligence's potential role in enhancing the early discovery of interval breast cancers.
Innovative Study Reveals AI's Potential in Identifying Hidden Breast Cancer Cases
Researchers from the our website, Jonsson Comprehensive Cancer Center have recently spearheaded a study that suggests Artificial Intelligence (AI) could revolutionize breast cancer screening. This groundbreaking discovery could spot interval breast cancers - those that grow between regular check-ups - before they escalate, leading to earlier treatment, improved screening practices, and better patient outcomes.
Published in the prestigious Journal of the National Cancer Institute, the study demonstrates AI's ability to detect "visible-by-mammography" types of interval cancers more promptly. These include tumors that may appear on mammograms but go undetected by human radiologists due to their subtlety, or faint signs that are easy to miss.
Researchers project that incorporating AI into the screening process could potentially reduce the number of interval breast cancers by approximately 30%. Dr. Tiffany Yu, the study's lead author and an assistant professor at the David Geffen School of Medicine at UCLA, elaborates, "Identifying these cancers earlier gives them a better chance of being treatable. Catching cancer early could mean less aggressive treatment and improved chances for a better outcome."
While similar research has been carried out in Europe, this is one of the first studies in the United States to explore the use of AI for early detection of interval breast cancers in the U.S. The research highlights differences in U.S. and European screening practices. Primarily, most U.S. mammograms are performed using digital breast tomosynthesis (DBT), often known as 3D mammography, and patients are typically screened every year. Alternatively, European programs usually opt for digital mammography (DM), or 2D mammography, and screen patients every two to three years.
The retrospective study scrutinized data from nearly 185,000 past mammograms between 2010-2019, combining both DM and DBT. Researchers examined 148 cases where women were diagnosed with interval breast cancer.
Subsequently, radiologists evaluated these cases to ascertain why the cancer wasn't detected earlier. The study adapted a European classification system to categorize the interval cancers. Categories include: Missed reading error, minimal signs-actionable, minimal signs-non-actionable, true interval cancer, occult (completely invisible on mammogram), and missed due to a technical error.
Researchers then implemented a commercially available AI software called Transpara to the initial screening mammograms before the cancer diagnosis to determine if it could detect subtle signs of cancer that might have been overlooked by radiologists during initial screenings, or at least mark them as concerning.
Key Insights:
- The AI flagged approximately 76% of mammograms that were initially read as normal but were later connected to an interval breast cancer.
- AI identified 90% of missed reading error cases, where the cancer was visible on the mammogram but missed or misinterpreted by the radiologist.
- It detected about 89% of minimal-signs-actionable cancers, showing very subtle signs, and 72% of minimal-signs-non-actionable cases that were likely too subtle to trigger action.
- For cancers that were occult or invisible on the mammogram, the AI flagged 69% of cases.
- It was somewhat less effective at identifying true interval cancers, those that weren't present at the time of screening but developed later, flagging approximately 50% of these cases.
Dr. Hannah Milch, assistant professor of Radiology at the David Geffen School of Medicine and senior author of the study, notes, "While our findings are exciting, there are a number of issues that need further exploration in real-world settings. For instance, the AI tool flagged 69% of the screening mammograms with occult cancers. However, it accurately marked the actual cancer only 22% of the time."
Ongoing large-scale prospective studies are essential to understand how radiologists would incorporate AI into their practices and address crucial questions, such as managing cases where the AI flags areas as suspicious that aren't visible to the human eye, especially when the AI isn't always accurate at pinpointing the exact location of cancer.
Dr. Yu concludes, "Although AI isn't flawless and should not be used alone, our results support the idea that AI could serve as a valuable second pair of eyes, particularly for the hardest-to-catch cancers. It's all about providing radiologists with better tools and giving patients the best chance at catching cancer early, potentially leading to more lives saved."
Other contributors, all from UCLA, include Dr. Anne Hoyt, Dr. Melissa Joines, Dr. Cheryce Fischer, Dr. Nazanin Yaghmai, Dr. James Chalfant, Dr. Lucy Chow, Dr. Shabnam Mortazavi, Christopher Sears, Dr. James Sayre, Dr. Joann Elmore, and Dr. William Hsu.
The research was supported by the National Institutes of Health, the National Cancer Institute, the Agency for Healthcare Research and Quality, and Early Diagnostics Inc.
- The Jonsson Comprehensive Cancer Center's study reveals that AI has the potential to significantly impact healthcare and health-and-wellness sectors, particularly in the earlier detection of breast cancer by identifying subtle signs that human radiologists might overlook.
- The use of AI in the screening process, as suggested by the study, could potentially reduce the occurrence of interval breast cancers by approximately 30%, enabling earlier treatment and better patient outcomes.
- Ongoing studies are essential to understand the practical implications of incorporating AI into medical-conditions diagnosis practices, such as dealing with false positive results, addressing issues like managing cases where AI flags areas as suspicious that aren't visible to the human eye, and enhancing the reliability of AI for improved healthcare services.
