Improving the accuracy of breast-conserving surgery using real-time imaging

The challenge:

Every year, over 20,000 people in Australia are diagnosed with breast cancer. Breast-conserving surgery, which aims to remove the tumour while preserving as much healthy breast tissue as possible, is a common surgical procedure in Australia. The only way surgeons can accurately determine if all the cancer has been removed is under a microscope after the operation.

Around 15-30% of people undergoing breast-conserving surgery require a second operation because cancer cells are found at the edges (margins) of the removed breast tissue. This can be costly for the healthcare system and can also place a physical, emotional and financial burden on individuals, while delaying post-surgery treatment, which may negatively affect long-term survival. There is a need for a faster and more accurate way to identify cancer at the tissue margins during surgery.

Project description:

In this NBCF-funded project, Professor Brendan Kennedy and his team at The University of Western Australia and the Harry Perkins Institute of Medical Research will bring new imaging technologies closer to use in the operating theatre, enabling surgeons to assess the margins of removed breast tissue in real time.

They will be using optical coherence tomography (OCT), a high-resolution imaging technique that can rapidly create 3D images of fresh tissue without dyes, as an alternative to support the standard testing done under a microscope (histology) after breast-conserving surgery. While OCT has the potential to show cancer cells remaining at the cut surface during surgery, the scans can be difficult to interpret as they are grainy and don’t ‘match’ with what clinicians can see under the microscope.

Machine learning, a type of artificial intelligence, will be used to solve this problem by accurately matching OCT scans with standard histology images of the same breast tissue to confirm what features of the OCT scan mean in biological terms. These machine learning models will be trained to convert the OCT scans into ‘virtual histology’ (computer generated images that resemble what clinicians see under the microscope), which will be tested for how accurately it can identify cancer at the margins of breast tissue.

Potential impact:

Professor Kennedy’s project could one day enable surgeons to assess breast tissue margins during the initial breast-conserving surgery. Surgeons could gain an accurate and reliable margin assessment tool that helps them visualise any remaining cancer cells in real-time. This could help people diagnosed with breast cancer avoid a second operation, start post-surgery treatment sooner, and reduce the physical, emotional and financial burden of care, improving outcomes and reducing healthcare costs. These benefits may be particularly important for people living in rural and remote areas who often need to travel long distances for surgery.

Grant code: 2025/RPG0114
Active years: 2026-2029
Scientific project title: Virtual Histology for Intraoperative Tumour Margin Assessment