[REVISION HISTORY]

Advancements and limitations in AI for breast cancer and med

Updated 9 times since CLSTR started tracking revisions of this situation.

What changed

2026-09-05 15:03 UTC → 2026-09-10 23:01 UTC · added ~~removed~~

~~Breast cancer detection~~ Advancements and limitations in AI ~~research advancements~~ for breast cancer and med

Developments in breast cancer detection involve both large-scale public health screening programs and advancements in artificial intelligence for medical imaging. In Croatia, the National Program for Early Detection of Breast Cancer has provided free annual screenings for women aged 49 to 70 since 2006. ~~The program conducts approximately 150,000 screenings annually with a 70% response rate, having identified over 8,300 carcinomas and contributing to decreased mortality.~~ Technological research continues to focus on improving diagnostic accuracy and radiologist confidence. Research utilizing the MedSAM ~~pre-trained~~ model and the CSAW-CC dataset aims to create explainable segmentation ~~tools. One study achieved a Dice coefficient of 0.72 and an IoU of 0.59, with surveyed radiologists expressing satisfaction with the predictions. A study published in the journal Radiology~~ tools, while studies involving Greek researchers and Swedish mammogram data ~~has further highlighted the predictive power of AI. Evaluating three commercial systems, researchers found that~~ have shown AI ~~could~~ can identify signs of cancer up to ten years before a clinical diagnosis. Recent research has also expanded the ~~diagnostic~~ utility of mammography to include cardiovascular health. A retrospective study presented at the European Society of Cardiology Congress in Munich by researchers from Chaim Sheba Medical Center and Tel Aviv University analyzed 97,364 mammograms from 29,921 women. Their health, with deep learning ~~model demonstrated significant~~ models demonstrating high accuracy in identifying ~~cardiovascular risks: 86%~~ risks for stroke, ~~79% for~~ hypertension, and ~~78% for~~ coronary heart disease. These results remained consistent regardless of age or existing cancer diagnosis. Furthermore, separate findings indicated that AI could potentially identify subtle signs of breast cancer up to six years before a clinical diagnosis, suggesting mammography could serve a dual purpose using existing screening infrastructure. Newer developments highlight both the practical ~~application~~ applications and limitations of medical AI. A study from the Technical University of Munich found that while AI assistance helped radiologists interpret chest X-rays faster and increased confidence, it did not consistently improve overall diagnostic accuracy and occasionally increased false positives. Conversely, the company Vara received Class IIb CE certification for an autonomous triage system that can independently report certain mammograms as normal, aiming to reduce workload in screening programs. In personalized medicine, researchers at NYU Langone Health developed NYU-DRP, a deep-learning tool that predicts a woman’s five-year breast cancer risk. By analyzing longitudinal 3D mammograms over multiple years, the tool identified high-risk individuals with 72 percent accuracy, outperforming single 3D mammogram analysis and AI-assisted 2D testing.

Versions

Only revisions since CLSTR began indexing content versions appear here. Select a version to see what changed compared to the one before it.