AI a proven success in clinical breast cancer diagnosis

A new artificial intelligence (AI) tool has been proven to identify breast cancer cases humans miss, acting as an effective safety net for cancer diagnosis.

AI can recognise patterns in large datasets far quicker than humans and often recognise patterns not apparent to us. This can make AI a useful ‘safety net’ in diagnoses.

The NHS Breast Screening Programme began in 1988 and is available to women every 3 years between 50 and 70 years of age. Every year, 55,000 people in the UK are diagnosed with breast cancer. Roughly 11,000 people die from breast cancer each year in the UK. However, the screening programme is estimated to prevent 1700 deaths per year. AI integration with traditional screening methods may prevent further deaths.

In a 2023 study led by Kheiron Medical Technologies and Imperial College London, an AI tool called Mia was used to double-check negative results from traditional screening. Mia was trained on over 3 million images and is theoretically much more ‘experienced’ than any radiologist could be.

Typical population screening works by taking mammograms (X-rays of each breast) which are then analysed by two radiologists. Patients can be recalled to an assessment clinic if results are inconclusive or potentially cancerous. This happens when both the radiologists agree on this. If they both agree that the results are negative (non-cancerous), there is no further assessment of the patient.

In the Kheiron study, Mia was used as a 3rd reader of the scans when radiologists dismissed the case. It improved cancer detection rates by up to 13%.

The 2-year-long study involved over 25,000 women at 4 screening sites in Hungary. This demonstrates the large-scale clinical application of a commercially available AI system. Kheiron are delighted with the results and are looking to use Mia in a US and UK rollout soon. Leeds Teaching Hospitals NHS Trust announced they were beginning a trial with Mia in March 2023.

Developments in AI could be a huge relief to the NHS. The UK currently has a 29% shortfall of clinical radiologists and a 17% shortfall of clinical radiologists. An effective integration of AI tools into existing cancer screening programmes may ease the delays caused by ongoing labour shortages. A different AI tool, used at Addenbrooke’s Hospital in Cambridge, has cut waiting times for breast cancer radiotherapy patients.

As well as numerous UK biotech companies producing excellent AI tools, AI also has political backing. Prime Minister Rishi Sunak is keen for the UK to play a leading role in AI. In March this year, the UK government announced £16 million of funding to develop AI’s use in medical and genetic diagnostics. Sunak also held the first global AI summit last month in Bletchley Park, focusing on the safety, regulation, and use of AI.

Latest