A new research paper authored by Marthe Larsen from the Cancer Registry of Norway, and co-authored by Karl Øyvind Mikalsen from SPKI and several others, has shed light on the potential of artificial intelligence (AI) in prior screening mammography. The study analyzed data from BreastScreen Norway spanning from January 2004 to December 2019, evaluating AI risk scores assigned to screening mammograms in women later diagnosed with breast cancer.
The retrospective study encompassed 2787 prior screening examinations from 1602 women, with an average age of 59 years. Among these cases, 1016 were screen-detected and 586 were interval cancers. Remarkably, the study revealed that 38.3% of screen-detected cancers and 39.4% of interval cancers were assigned the highest AI risk score, denoted as 10, on the mammograms preceding their diagnosis.
Of particular note, for screen-detected cancers, where AI scores were available two screening rounds (4 years) prior to diagnosis, an impressive 23.0% exhibited a score of 10. Moreover, the research demonstrated a significant association between mammographic features and AI scores for invasive screen-detected cancers (P < .001). Notably, density with calcifications was recorded for 13.6% of screen-detected cases with a score of 10, in contrast to 4.6% for those with a score ranging from 1 to 7.
The study's findings hold great promise for the early detection of breast cancer, potentially revolutionizing mammography screening practices. This research underscores the transformative potential of AI in enhancing the effectiveness of breast cancer screening, offering a ray of hope for countless women worldwide.
For more detailed information, the full research paper is available through the journal Radiology.
This news article has been written with input from chatGPT.
