During the past decade, outbreaks of visceral leishmaniasis (VL) due to Leishmania chagasi have occurred in over a half dozen major Brazilian cities. The emergence of urban leishmaniasis appears to have resulted from massive population movements from rural areas to crowded and precarious housing settlements. In Teresina, Piauí, nearly 1000 cases of VL were reported during 1993 and 1994. Over 90% of cases required hospitalization, and 5% were fatal despite treatment. Measures to control the sand fly vector, Lutzomyia longipalpis, and the principal canine reservoir, the domestic dog, have not eliminated transmission of the parasite. Improved methods for identifying areas at risk for transmission are necessary for accurate application of interventions for control. The purpose of this project is to apply remote sensing and geographic information system (GIS) technologies to characterize the landscape elements associated with transmission of the parasite and the presence of the vector and canine reservoir.
The place of residence of each of the 951 persons with symptomatic VL in 1993-4 was determined by reviewing records from all hospitals in Teresina and the Fundaçaõ Nacional da Saúde (FNS), the national health service that distributes antileishmanial drugs. FNS also provided the results of serological tests for antibodies to L. chagasi taken from the 57,062 dogs that were screened during 1993-4 in 74 Teresina districts. Remotely sensed data of the 16x22-km study area included a Landsat Thematic Mapper (TM) scene (30-m resolution) acquired 18 October 1995 and a SPOT panchromatic scene (10-m resolution) acquired 15 September 1996. Both scenes were registered to a UTM coordinate system. A classification algorithm was used to assign the TM pixels to 30 clusters based on their spectral reflectance characteristics. These clusters were displayed on a computer and further grouped into 16 landcover classes by comparison with color aerial photographs taken in October 1992 and April 1996, as well as ground-level photographs of 75 reference points, the coordinates of which were determined with a hand-held Magellan GPS receiver. In another procedure, a kernel-based spatial reclassification (SPARK) algorithm was used to classify urban features based on texture, resulting in a 16-class land use map. The boundaries of Teresina's 74 districts were digitized in ARC/Info, and subsequently used to extract the proportions of landcover and land use classes by district. As a third product, the TM and SPOT data were merged into a single image to provide both multispectral information and high spatial resolution. All image processing and GIS tasks were conducted at NASA's CHAART.
Rates of human VL and canine seropositivity were calculated for each of the districts and statistically compared to the landcover and land use proportional data. In general, high rates of human disease and canine seropositivity were found in districts with extensive forested land or medium- and low-density housing, while low rates were found in districts in which densely settled residential and commercial areas were prominent and trees were sparse. Studies planned for this year include further spatial analysis of the data collected at the district level, determination of the geographic coordinates of individual cases to permit analysis at the peridomestic level, and incorporation of the merged TM/SPOT data into a case-control study of risk factors for human VL.
Back to
Leishmaniasis in Brazil Project Introduction
Last updated: October 2001