Pre-trained convolutional neural networks as feature extractors toward improved malaria parasite detection in thin blood smear images

Thoma GR, Long LR, Antani S
September 2006 Technical Report to the LHNCBC Board of Scientific Counselors.

NLM is recognized worldwide for the quality and value of its information services to biomedical research and practice. We have the opportunity to advance the value of these and future services, possibly augmented by biomedical images, by conducting the required research and development toward: (1) integrating existing technologies to make images more readily available, along with associated descriptive information, within integrated multimedia management systems, and to allow the collection of interpretive information (including graphical information) from these images; and (2) creating advanced methods for indexing, classifying, and retrieving images, based not only on what has been recorded about the images, but on the image contents themselves.