Everywhere where people look at photos, you can also use computers, says Bram van Ginneken (theme Rare Cancers) in the Volkskrant of 2 December 2017. Gradually computers can be trained to recognize diseases as cancer or tuberculosis with performance similar to doctors - or better. But these computer programs are not yet widely used.
The self-learning computers use so-called deep learning neural networks, and they have become extremely good in systematically plowing through large quantities of images in order to organize them, to classify them and to find deviations in the data. On the basis of large amounts of training data, the system can thus arrive at decisions or new insights. That is why networks that work with images are being used more and more, in various areas from exposing imitations of artworks to self-driving cars and face recognition for security purposes. And certainly also in health care, where doctors were always those who went looking for abnormalities, e.g., by the radiologist, under a microscope by a pathologist or with the naked eye by a dermatologist.
Read the complete Volkskrant article (in Dutch) >>