Project Details

Description

1) What is/are the main scientific or technological question(s) that you plan to address? Malaria is a very serious public health problem in many tropical and sub-tropical countries. To ameliorate this problem, it is crucial to implement accurate diagnosis techniques capable of detecting early on the specific parasites responsible for the infection. Currently, the standard diagnosis technique recommended by the World Health Organization, WHO, is the Romanowsky method, which is based on staining thin and thick blood smear samples and analyzing them with optical microscopy. This methodology is affordable, as it only requires an optical microscope equipped with a 100X magnification lens system, which is useful for developing countries where malaria is rampant. However, detecting the parasites, their morphology and their effects on blood cells, is a very slow process that is prone to errors. Indeed, the analysis of the samples requires multiple steps (see scheme below), in which the presence of parasites, the shapes of the blood cells, the colors and morphology of parasites, and many other subtle aspects, are assessed by a trained individual. Among the institutions that participate in this proposal, is the National Institute of Health in Colombia. Some of the scientists who participate in this proposal (in particular Dr. Liliana Jazmin Cortes Cortes) are precisely dedicated to training individuals around Colombia to be able to recognize the subtle aspect in the samples shown in the scheme below. We are acutely aware of the limitations of this approach: the quality of the images collected differ from area to area, and assessing whether or not a blood cell is oval (an indication of the disease, see below) is, in many cases, challenging. We wish to use modern computational techniques to make the diagnosis not only more accurate but also faster. This is the reason why we¿re submitting this proposal, as we¿re convinced that the deep-learning techniques developed by the Data NanoAnalytics, DNA, group at CNMS, will allow to accomplish our goals. To reach our goals, however, it¿s very clear that we have to proceed in a systematic way that is commensurate with the development of the disease. Malaria undergoes several stages of development, and the blood samples differ depending on the stage. Five parasites can cause the disease: Plasmodium (P.) vivax, P. falciparum, P. malariae, P. ovale y P. knowlesi. The first three are usually found in Colombia, whereas P. knowlesi is typically found in Asia. Each specie has a characteristic morphology and size, which depends too on the whether the parasite is sexual or asexual. Some other general morphological aspects to consider are: i) P. falciparum is the smallest parasite. ii) More than one parasite found in a blood cell indicates the presence of P. falciparum. iii) Asexual and sexual varieties are observed for all species, but P. falciparum presents only sexual ones. iv) Contact points of the parasite P. Vivax with blood cells produces granulations, (Shüffner granulations), that can be easily observed with optical microscope (Figure 1b). v) P. Vivax and P. ovale enhance and deform blood cells (Figure 1a), the later inducing oval shape. vi) There are color-based criteria to identify parasitic structures too. For instance, P. malariae has pigments with darker tones than the others. To demonstrate the application of deep-learning techniques, the first year of this proposal we will be focused on points iv) and v) above. In particular, Deep Convolutional Neural Networks, DCNN, and rotational invariant Variational Autoencoders, VAEs, will be used to look for Shüffner granulations and cells with oval shape in the blood samples. An example of granulations and cells with oval shape can be seen in Figure 1 below. In the second year we will focus on parasite morphology.
StatusFinished
Effective start/end date01/08/2131/07/22

Project funding

  • Internal
  • PONTIFICIA UNIVERSIDAD JAVERIANA