Brazilian scientists have developed an algorithm capable of identifying the protozoan 'Trypanosoma cruzi' in photos of blood samples obtained with mobile phone camerasWhen the Brazilian immunologist Helder Nakaya visited the Evandro Chagas Institute, in the city of Belém (state of Pará, Brazil), in 2017, there was a commotion because one of the best microscopists of the entity was going to retire at that time.In this way, much of the knowledge that allowed rapid and accurate identification of strains of the Leishmania protozoan would be lost.“I was intrigued by this waste of all that capacity accumulated over decades.It was then that we began to investigate and try to train the computer with the wisdom of the professional to act in the identification of microorganisms and at a low cost”, explains Nakaya.Five years later, a group of researchers, under the coordination of Nakaya and the scientist Mauro César Cafundó de Morais, published the results of a study showing that it is possible to use artificial intelligence to detect Trypanosoma cruzi, the protozoan that causes Alzheimer's disease. Chagas, in images of blood samples obtained with cell phones and analyzed in optical microscopes.The algorithm that the group developed is available in an article published in the scientific journal PeerJ.The research brought together professionals from various areas, from biology to mathematics and computing.“We achieved a good machine learning result.With the algorithm working well for Chagas disease, you will be able to adapt it to other imaging-dependent purposes, such as stool sample analysis, dermatology, and colposcopy,” says Nakaya, one of the principal investigators at the Center for Inflammatory Diseases Research. (CRID), a FAPESP Center for Research, Innovation and Dissemination (CEPID) based at the Ribeirão Preto School of Medicine of the University of São Paulo (FMRP-USP).Nakaya is also carrying out joint research with the Hospital Israelita Albert Einstein, the Pasteur-USP Scientific Platform and the Instituto Todos pela Saúde.Trained microscopists perform a form of diagnosis of Chagas disease: they detect the parasites in blood samples.This requires a professional microscope, which can be attached to a high-resolution camera, but that often makes this method expensive and difficult for low-income people to access.Chagas disease, classified by the World Health Organization (WHO) as one of the neglected tropical diseases (NTDs), is considered a chronic infectious condition, the prevention of which is related to the mode of transmission, that is, to the control of the insect known as vinchuca.This requires responses from health care networks.The infection caused by Trypanosoma cruzi, which is endemic in 21 countries of the American continent, affects approximately 6 million people, with an annual incidence of 30,000 new cases in the region, causing an average of 14,000 deaths per year.Likewise, it is estimated that around 70 million people live in areas exposed to kissing bugs and are at risk of contracting the disease.In Brazil, even though there is a downward trend in mortality rates, there has been an average of 4,000 thousand deaths per year caused by this disease during the last decade.The researchers developed a computer learning approach based on the so-called random forest, by creating an algorithm for the detection and counting of Trypanosoma cruzi trypomastigotes in images obtained with a phone camera cell phone.Trypomastigotes are the morphological form of the protozoan present in the extracellular phase and present in the blood of patients with acute disease.Micrographs of blood smear samples recorded in images with a resolution of 12 megapixels were analyzed.A set of morphometric parameters (shape and size), color and texture measurements were extracted from 1,314 parasites.In this stage, scientists João Santana Silva, Paola Minoprio and Ricardo Gazzinelli, specialists in parasites, helped "teach" the machine to recognize them, especially the Trypanosoma.USP researchers Roberto Marcondes César Jr. and Luciano da Fontoura Costa, experts in machine learning and image processing, also collaborated.Subsequently, the samples were divided into training and test sets, and then they were classified using the random forest algorithm.The results were precision and sensitivity values ​​considered high: they were located at 87.6% and 90.5% respectively.The area was analyzed with the Receiver Operating Characteristic curve (ROC curve), a graphical representation that illustrates the performance or throughput of a binary classifier system as its discrimination threshold varies.In this way, the group managed to automate the analysis of images acquired with a mobile device, thus obtaining an alternative with a view to reducing costs and increasing efficiency in the use of the optical microscope.“The idea is to generate images and analyze them in microscopes that can be sent to remote places in Brazil so that the application itself indicates whether or not it is Chagas disease.That is why it is important to also have a robust and low-cost microscope that can automatically collect the images”, adds Nakaya.According to the researcher, the proposal consists of leaving the algorithm open so that the scientific community contributes other data and other resources.One of the challenges now lies in obtaining low-cost microscopes, such as the paper-based equipment invented by scientists Manu Prakash and Jim Cybulski at Stanford University (United States), but which ended up not giving the expected results when applied with parasites.Copyright © 1997-2022 DataPress Multimedia - Director: Diego Rottman Reproduction permitted with permission requested