Mammography Study Suggests DBT-Based AI May Help Reduce Disparities with Breast Cancer Screening
New research suggests that AI-powered assessment of digital breast tomosynthesis (DBT) for short-term breast cancer risk may help address racial disparities with detection and shortcomings of traditional mammography in women with dense breasts.
Can artificial intelligence (AI) play a role in mitigating disparities with breast cancer screening?
For the multicenter study, presented at the 2024 San Antonio Breast Cancer Symposium (SABCS), researchers evaluated the ProFound AI Breast Health Suite (iCAD) for assessing short-term breast cancer risk based on digital breast tomosynthesis (DBT) images for 3,558 women. The cohort included 394 cases of breast cancer detection, according to the study. For the ProFound AI Breast Health Suite, one-year AI risk scores < 34 percent denote average risk, scores between 34 percent and 60 percent reveal intermediate risk and scores > 60 percent are considered high-risk scores.
Here one can see a non-spiculated mass on the upper left mammography image for a 62-year-old Black woman who had scattered fibroglandular densities at initial screening and a grade 3 invasive ductal carcinoma detect in subsequent mammography screening (upper right) with the AI detection score (below) increasing from 37 to 85. (Images courtesy of 2024 San Antonio Breast Cancer Symposium (SABCS) and iCad.)
Overall, women with intermediate risk scores comprised 13.1 percent of the cohort, high-risk scores accounted for 18.2 percent of the study participants and the remaining 68.7 percent of the cohort had average risk scores.
While acknowledging differences in the racial composition of the adjusted high-risk screening population, the study authors noted similarities in breast cancer prevalence for those who had high AI one-year risk scores. In this group, the researchers pointed out a 1 in 38 prevalence of breast cancer in White women, a 1 in 41 prevalence in Black women, a 1 in 34 prevalence for Asian women and a 1 in 55 prevalence for women of other races.
“The case scores did not vary significantly across racial subgroups in our dataset, suggesting that the accuracy of the AI software was consistent across races,” noted Chirag Parghi, M.D., M.B.A., the chief medical officer for Solis Mammography.
The study authors also noted similarities in AI high-risk scoring between women with dense breasts and those with non-dense breasts. For women with dense breasts and high AI risk scores, the breast cancer prevalence was 1 in 36 in comparison to a 1 in 41 prevalence for women with non-dense breasts and high AI risk scores, according to the researchers.
“Based on our results, image-based risk can potentially offset known gaps in breast cancer detection by traditional mammography in patients with dense tissue and address known disparities across races,” emphasized Parghi and colleagues.
Reference
1. Parghi C, Pantleo J, Shisler J, et al. Image-derived short term breast cancer risk score in the analysis of breast cancer prevalence in screening population by race and breast density. Poster presented at the 2024 San Antonio Breast Cancer Symposium (SABCS) December 10-13, San Antonio, Texas.
Recent Videos
Related Content
Can Abbreviated Breast MRI Have an Impact in Assessing Post-Neoadjuvant Chemotherapy Response?
April 24th 2025Article
New research presented at the Society for Breast Imaging (SBI) conference suggests that abbreviated MRI is comparable to full MRI in assessing pathologic complete response to neoadjuvant chemotherapy for breast cancer.
Clarius Mobile Health Unveils Anterior Knee Feature for Handheld Ultrasound
April 23rd 2025Article
The T-Mode Anterior Knee feature reportedly offers a combination of automated segmentation and real-time conversion of grayscale ultrasound images into color-coded visuals that bolster understanding for novice ultrasound users.
What is the Best Use of AI in CT Lung Cancer Screening?
April 18th 2025Article
In comparison to radiologist assessment, the use of AI to pre-screen patients with low-dose CT lung cancer screening provided a 12 percent reduction in mean interpretation time with a slight increase in specificity and a slight decrease in the recall rate, according to new research.
