An artificially intelligent computer system is better at identifying brain tumours from scans than a human doctor, a study has found.
Researchers used an advanced form of imaging combined with an artificial intelligence (AI) algorithm to see how successful it would be at correctly identifying tumours in scans. The AI diagnosis was 94.6 per cent accurate, compared with 93.6 per cent for a human pathologist. It was also far quicker.
The study, published in Nature Medicine, used a technique called stimulated Raman histology, using laser imaging to highlight features of the brain which do not usually show up in standard images taken from scans. It produces microscopic images in high resolution.
The AI algorithm was trained using 2.5 million samples taken from 415 patients,
