Remote Sensing and Geographic Information Systems for the Study of Schistosomiasis in the State of Minas Gerais, Brazil Corina C. Freitas, Ricardo J. P. S. Guimarães, Luciano V. Dutra, Flavia T. Martins, Erica J. C. Gouvêa Image Processing Division National Institute for Space Research - INPE São José dos Campos, Brazil {corina, ricardo, dutra, flavinha, erica}@dpi.inpe.br Ricardo A. T. Santos 1 , Ana C.M. Moura 2 , Sandra C Drummond 3 ,Ronaldo S. Amaral 4 , Omar S. Carvalho 5 1 Centro Tecnológico da Aeronáutica – CTA 2 Universidade Federal de Minas Gerais - UFMG 3 Secretaria de Estado de Saúde de Minas Gerais 4 Secretaria de Vigilância em Saúde/MS 5 Centro de Pesquisas René Rachou/FioCruz-MG Abstract—This article uses remote sensing and geographical information system to establish a statistical model for estimating schistosomiasis prevalence in the state of Minas Gerais, Brazil. Remote sensing data were derived from MODIS and SRTM. The final regression model includes the Digital Elevation Model and winter Normalized Difference Vegetation Index variables. A risk map for the entire state of Minas Gerais is built, based on these variables. Keywords-geographical information systems (GIS), remote sensing, risk models, Schistosomiasis. I. INTRODUCTION The schistosomiasis mansoni is an endemic disease transmitted by snails of Biomphalaria gender. The disease is present in several countries, especially in the developing ones, such as some American and African countries [1]. The extensive distribution of the hosts in Minas Gerais, Brazil gives to schistosomiasis an expansive characteristic, even for those areas considered indene [2-4]. In the endemic areas, the high concentration of the hosts, associated to other risk factors, facilitates the existence of communities with high prevalence of schistosomiasis. The distribution of the schistosomiasis in the state of Minas Gerais is not regular, with high prevalence areas followed by areas where the transmission is low or almost null. Schistosomiasis is a disease determined, in space and time, from environmental factors such as vegetation, temperature, land use, water collections, etc. Characterization of environmental factors involves the use of satellite imagery and spatial analysis. With the use of geographic information systems (GIS) and remote sensing (RS) data it is possible to take into account the heterogeneity of the diseases spatial distribution, which are related to environmental parameters. The link between disease, environmental conditions, geographical information systems, and remote sensing is an increasing research area and it plays a potential and a valuable rule for schistosomiasis studies [5-7]. The objective of this paper is to establish, using a geographic information system, statistical models as a function of remote sensing data, aiming at mapping schistosomiasis risk areas in the state of Minas Gerais, Brazil. II. METHODOLOGY A. Variables Acquisition The remote sensing data used were derived from MODIS (Moderate Resolution Imaging Spectroradiometer) and SRTM (Shuttle Radar Topography Mission). MODIS MOD13Q1 images (h14v10, h14v11, h13v10 and h13v11) from two dates (one in summer and another in winter) were taken, and for each date nine variables were used: Blue, Red, Near Infrared (NIR), Medium Infrared (MIR), Enhanced Vegetation Index (EVI), Normalized Difference Vegetation Index (NDVI), and vegetation (VEG), soil (SOIL) and shadow (SHD) indices derived from the mixture model [8]. Two variables from SRTM elevation data were also used: the digital elevation model (DEM) itself and the declivity (DEC), derived from the DEM. The schistosomiasis prevalence data, gathered from Brazilian Health National Foundation and from Health Secretariat of Minas Gerais State Annual Reports, consisted of historical data from 189 municipalities. These data were randomly divided in two sets: one with 142 cases for variables selection and model definition, and another with 47 cases for model validation. The spatial distribution of these sets can be observed in Fig. 1. B. Variables Selection and Model Building The MODIS and SRTM products were used as input variables to establish the multiple regression model for prevalence risk. Mean values for each municipality were taken to relate the variables. The relations among the dependent (prevalence, denoted by Pv) and the twenty independent variables (nine from MODIS in summer time, nine from MODIS in winter time and two from SRTM) were analyzed in terms of correlation, 0-7803-9510-7/06/$20.00 © 2006 IEEE 2436