A new machine learning-based tool is 98 percent effective in identifying early signs of breast cancer, as reported in a study.
Researchers at the University of Edinburgh developed this fast, non-invasive method that integrates laser analysis with machine learning. They claim it is the first technique capable of detecting patients in the very early stages of breast cancer, and it may lead to a screening test for various cancer types.
This technique can detect subtle changes that occur in the bloodstream during the initial phases of the disease—referred to as stage 1a—which current testing methods often miss.
Standard breast cancer tests, including physical examinations, X-rays, ultrasound scans, or tissue sample analyses (biopsies), typically screen individuals based on age or risk factors.
The pilot study, published in the Journal of Biophotonics, analysed 12 blood samples from breast cancer patients alongside 12 samples from healthy individuals. In this study, the researchers refined a laser analysis method known as Raman spectroscopy and combined it with machine learning.
They achieved a 98 per cent effectiveness rate in identifying breast cancer at stage 1a.
The process involves directing a laser beam into blood plasma obtained from patients. A spectrometer device assesses the properties of the light after its interaction with the blood. The device then reveals minute alterations in the chemical composition of cells and tissues, indicating early signs of the disease.
By employing a machine learning algorithm, doctors can interpret these results. This innovative approach also allowed the team to differentiate between the four primary breast cancer subtypes with over 90 per cent accuracy, enabling patients to receive more tailored and effective treatments.