An artificial intelligence tool piloted by the NHS successfully identified tiny signs of breast cancer in 11 women that had been missed by human clinicians. The programme, named Mia, analysed the mammograms of over 10,000 women alongside radiologists in NHS Grampian, Aberdeen.
While most scans were cancer-free, Mia flagged all cases where cancer was present, including an extra 11 that had not been spotted by doctors. One patient, Barbara, had a 6mm tumour detected early, requiring only a minor operation and five days of radiotherapy. She noted her treatment was far less invasive than that of her sister and mother, who had also battled the disease. Without the AI tool, her cancer might not have been discovered until her next routine mammogram three years later.
The developer, Kheiron Medical, claims that tools like Mia could reduce waiting times for results from 14 days to three. Currently, two radiologists review each scan; the hope is that one could eventually be replaced by the AI, halving the workload. Dr Gerald Lip, clinical director of breast screening in the north east of Scotland, said only 81 of the 10,889 participants declined AI review of their scans.
Mia was trained on millions of mammograms from diverse women worldwide. However, its machine learning feature was disabled due to health regulations, preventing it from learning on the job. It also had no access to patient history, meaning it could flag previously harmless cysts. The University of Aberdeen validated the research, but results have not yet been peer reviewed.
Despite limitations, Dr Lip considers Mia a friend and augmentation to his practice, freeing him for more patient interaction. The Royal College of Radiologists noted the encouraging potential of AI in diagnostics.
