AI digital diagnostic system could detect cervical cancer
Using AI and mobile digital microscopy, researchers may be able to screen and detect for cervical abnormalities as precursors to cervical cancer that can be carried out on portable scanners.
A study led by researchers at Karolinska Institutet (Solna, Sweden) trained an AI system to recognize these abnormalities which, if detected early, could lead to successful treatment of cervical cancer. Cervical smear samples were taken from 740 women at a rural clinic in Kenya. These samples were then made digital via a portable scanner and uploaded using mobile networks to a cloud-based deep-learning system (DLS). Using just under half of the smears, the system was trained to recognize various precancerous lesions. The remaining smears were used to evaluate the system’s accuracy.