A team of researchers from Stanford University developed an artificial intelligence algorithm that diagnoses pneumonia from chest X-rays, the California-based university announced Nov. 15.
The research team created the deep learning algorithm using 112,120 chest X-rays labeled with up to 14 possible pathologies, including pneumonia. To compare the algorithm's accuracy, the researchers asked four Stanford radiologists to independently annotate a test set of 420 chest X-rays for possible indicators of the infection.
After just over a month of development, the researchers reported their algorithm was able to diagnose pneumonia from chest X-rays more accurately than the four radiologists. The level of accuracy was determined by comparing the algorithm and the four radiologists' diagnoses against a majority vote of radiologists, according to a research paper submitted to the scientific preprint website arXiv.org Nov. 14.
"The motivation behind this work is to have a deep learning model to aid in the interpretation task that could overcome the intrinsic limitations of human perception and bias, and reduce errors," explained Matthew Lungren, MD, an assistant professor of radiology at Stanford and co-author of the study.