Machine learning boosts cancer analysis accuracy, #healthtechrevolution.

A computer system called Histomorphological Phenotype Learning (HPL) developed by teams from Glasgow and New York universities in Britain and the UK has shown outstanding accuracy in spotting cancer signs in tissue samples. This system has the potential to improve the speed and accuracy of cancer diagnoses and predict patient outcomes more reliably. The system uses self-supervised deep learning to analyze high-resolution images of tissue samples and identify recurring visual elements that correspond to different cancer phenotypes. The algorithm was able to distinguish between features of squamous cell lung cancer with 99% accuracy and make predictions about cancer recurrence with 72% accuracy, surpassing human pathologists’ accuracy of 64%. The researchers believe that as more data is added, the system will become even more accurate in identifying patterns in cancer datasets. The ultimate goal is to provide doctors and patients with a tool that can improve their understanding of prognosis and treatment. The study was published in Nature Communications and received funding from various research councils and institutes. Researchers from New York University, University College London, and the Karolinska Institute also contributed to the paper.

Source link

Source link: http://www.labnews.co.uk/article/2095661/machine-learning-offers-means-to-raise-accuracy-of-cancer-analysis