Hologic's AI Technology Enhance Mammography's Capability in Detecting Challenging Cancers

Recent studies highlight the potential of Hologic's AI-enabled mammography technology to improve the detection of difficult-to-identify breast cancers. Findings were presented at the Society of Breast Imaging Symposium in Seattle, where the focus was on the use of Artificial Intelligence (AI) in identifying invasive lobular cancer (ILC), a subtype known for its diagnostic challenges.

According to Mark Horvath, President of Breast & Skeletal Health Solutions at Hologic, "Invasive lobular cancers are more challenging to detect on a mammogram because of their unique characteristics. In the study, AI maintained high sensitivity for flagging these cancers, including some that had been interpreted as negative at a prior screening."

A retrospective study conducted by Massachusetts General Hospital (MGH) reviewed invasive lobular cancer cases from a 10-year period. The study categorized 239 cases into two groups: those successfully identified by radiologists during routine screenings (195 cases) and those initially noted as negative but diagnosed within a year (44 cases). The subsequent analysis using Hologic’s Genius AI® Detection solution demonstrated that the technology successfully identified nearly 90% of ILC cases. Notably, the AI also detected 43% of cases previously assessed as negative.

Breast cancer remains a global health challenge, with forecasts suggesting a rise to 3.2 million new cases by 2050. Invasive lobular cancer accounts for approximately 10–15% of all breast cancer types, whereas ductal carcinoma, originating from milk ducts, is more prevalent. ILC’s biology poses unique challenges for mammographic detection, often leading to later diagnosis.

Additional sessions at the symposium included discussions on the integration of AI in breast imaging, specifically with Hologic’s 3DQuorum technology. This technology aims to streamline the imaging process by reducing the number of slices radiologists must analyze, maintaining sensitivity and accuracy.

Despite the encouraging results, some study limitations were noted. The MGH study did not evaluate real-time clinical outcomes such as false-positive rates, recall rates, or biopsy results. It was conducted within a single institution, and AI performance was assessed retrospectively without actively influencing the decision-making process of radiologists during screenings.