Every year, thousands of women face the uncertainty of a breast lump diagnosis. Current standard tests often struggle to differentiate between common, harmless growths and those that require immediate, specialised treatment. We are developing a pioneering AI diagnostic platform designed to provide clinicians with unprecedented clarity. Our goal is simple: to save women from avoidable surgery, reduce the burden of invasive procedures, and ensure that every patient receives exactly the care they need.
Key Information
What is a Fibroepithelial Lesion?
Fibroepithelial lesions (FEL) are a category of breast masses that include two primary types:
- Fibroadenoma (FA): These are the most common benign breast tumours, typically found in younger women. They are harmless and can usually be managed with simple monitoring or minor procedures.
- Phyllodes Tumour (PT): These are much rarer tumours that can range from benign to potentially aggressive. Because they have the potential to recur or spread, international guidelines require more extensive surgical removal with wide margins.
Why do current Tests fall Short?
Currently, doctors use imaging and core needle biopsies to evaluate breast lumps. However, because these two types of tumours share very similar biological features, standard tests can lead to diagnostic ambiguity:
- Diagnostic Overlap: On a standard needle biopsy, the features of a harmless fibroadenoma and a phyllodes tumour can be virtually indistinguishable.
- Unnecessary Surgeries: Because clinicians must prioritise patient safety, nearly 50% of surgeries performed today for these lesions turn out to be medically unnecessary, as the growths are ultimately found to be benign.
- Upgrading Risks: Conversely, up to 23% of cases initially thought to be harmless on biopsy are "upgraded" to more aggressive tumours after surgery, sometimes requiring a second, more traumatic operation
A New Standard of Intelligence
We are developing a translational AI system that acts as a powerful digital consultant for medical teams. Our platform moves beyond simple visual assessment to identify deep biological signatures within standard clinical data.
- Multidimensional Analysis: The system is designed to evaluate a comprehensive range of patient information, providing a unified "Risk Score" to guide treatment decisions.
- Deep Learning Excellence: By training on one of the largest clinical datasets of its kind, spanning over 15 years of patient cases, our AI learns to recognise subtle markers of disease that the human eye might miss.
- Integrated Workflow: The platform is built to fit seamlessly into existing hospital systems, providing rapid, objective results that support radiologists and pathologists in real-time
The Future of Personalized Treatment
Our project is driven by a commitment to improving the lives of women through technology:
- Targeting 90%+ Accuracy: We are aiming for a diagnostic accuracy of at least 90%, the threshold required for doctors to feel confident in recommending monitoring over surgery for benign cases.
- Reducing Harm: By increasing diagnostic precision, we aim to significantly reduce the number of painful needle biopsies and unnecessary surgical procedures women must undergo.
- Efficiency for Healthcare: Shorter wait times and more accurate first-time diagnoses mean better use of hospital resources and faster peace of mind for patients