Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 May 19;23(1):13.
doi: 10.1186/s12942-024-00371-w.

Using image segmentation models to analyse high-resolution earth observation data: new tools to monitor disease risks in changing environments

Affiliations

Using image segmentation models to analyse high-resolution earth observation data: new tools to monitor disease risks in changing environments

Fedra Trujillano et al. Int J Health Geogr. .

Abstract

Background: In the near future, the incidence of mosquito-borne diseases may expand to new sites due to changes in temperature and rainfall patterns caused by climate change. Therefore, there is a need to use recent technological advances to improve vector surveillance methodologies. Unoccupied Aerial Vehicles (UAVs), often called drones, have been used to collect high-resolution imagery to map detailed information on mosquito habitats and direct control measures to specific areas. Supervised classification approaches have been largely used to automatically detect vector habitats. However, manual data labelling for model training limits their use for rapid responses. Open-source foundation models such as the Meta AI Segment Anything Model (SAM) can facilitate the manual digitalization of high-resolution images. This pre-trained model can assist in extracting features of interest in a diverse range of images. Here, we evaluated the performance of SAM through the Samgeo package, a Python-based wrapper for geospatial data, as it has not been applied to analyse remote sensing images for epidemiological studies.

Results: We tested the identification of two land cover classes of interest: water bodies and human settlements, using different UAV acquired imagery across five malaria-endemic areas in Africa, South America, and Southeast Asia. We employed manually placed point prompts and text prompts associated with specific classes of interest to guide the image segmentation and assessed the performance in the different geographic contexts. An average Dice coefficient value of 0.67 was obtained for buildings segmentation and 0.73 for water bodies using point prompts. Regarding the use of text prompts, the highest Dice coefficient value reached 0.72 for buildings and 0.70 for water bodies. Nevertheless, the performance was closely dependent on each object, landscape characteristics and selected words, resulting in varying performance.

Conclusions: Recent models such as SAM can potentially assist manual digitalization of imagery by vector control programs, quickly identifying key features when surveying an area of interest. However, accurate segmentation still requires user-provided manual prompts and corrections to obtain precise segmentation. Further evaluations are necessary, especially for applications in rural areas.

Keywords: Drone; Mosquito-borne diseases; Remote sensing.; Segment anything model; UAV.

PubMed Disclaimer

Conflict of interest statement

The authors declare no competing interests.

Figures

Fig. 1
Fig. 1
Geographic location of the study sites in Peru, Burkina Faso, Ivory Coast, Malaysia, and Tanzania
Fig. 2
Fig. 2
Workflow for gold standard generation
Fig. 3
Fig. 3
Text prompt segmentation pipeline
Fig. 4
Fig. 4
Iquitos, Peru. (a) Grounding Dino bounding boxes (b) SAM output segmentation
Fig. 5
Fig. 5
(a) Iquitos, Peru and (d) Malaysia show point prompts in magenta used for water and buildings segmentation. (b) and (e) show the output of SAM in red polygons. (c) and (f) show the gold standard in blue polygons
Fig. 6
Fig. 6
Dice coefficient histograms per class
Fig. 7
Fig. 7
Boxplot of Dice coefficients for each class by site
Fig. 8
Fig. 8
Dice coefficient for each site according to the water text prompts used with 0.3 box threshold
Fig. 9
Fig. 9
Dice coefficient for each site according to the building text prompts used with 0.3 box threshold
Fig. 10
Fig. 10
Results using a 0.3 Box threshold. The image on the left corresponds to the buildings gold standard, the middle image shows the Grounding Dino bounding boxes in red and the right figure shows in blue SAM’s output. The sample from Kudat does not contain houses
Fig. 11
Fig. 11
Results using a 0.3 Box threshold. The image on the left corresponds to the water body gold standard, the middle image shows the Grounding Dino bounding boxes in red and the right figure shows in blue SAM’s output

Similar articles

References

    1. de Souza WM, Weaver SC. Effects of climate change and human activities on vector-borne diseases. Nat Rev Microbiol 2024. - PubMed
    1. Gibb R, Colon-Gonzalez FJ, Lan PT, Huong PT, Nam VS, Duoc VT, Hung DT, Dong NT, Chien VC, Trang LTT, et al. Interactions between climate change, urban infrastructure and mobility are driving dengue emergence in Vietnam. Nat Commun. 2023;14(1):8179. doi: 10.1038/s41467-023-43954-0. - DOI - PMC - PubMed
    1. Mondal N. The resurgence of dengue epidemic and climate change in India. Lancet. 2023;401(10378):727–8. doi: 10.1016/S0140-6736(23)00226-X. - DOI - PubMed
    1. Xu Y, Zhou J, Liu T, Liu P, Wu Y, Lai Z, Gu J, Chen X. Assessing the risk of spread of Zika virus under current and future climate scenarios. Biosaf Health. 2022;4(3):193–204. doi: 10.1016/j.bsheal.2022.03.012. - DOI
    1. Marques R, Kruger RF, Cunha SK, Silveira AS, Alves D, Rodrigues GD, Peterson AT, Jimenez-Garcia D. Climate change impacts on Anopheles (K.) cruzii in urban areas of Atlantic Forest of Brazil: challenges for malaria diseases. Acta Trop. 2021;224:106123. doi: 10.1016/j.actatropica.2021.106123. - DOI - PubMed

Publication types