Optimization of automated segmentation of monkeypox virus-induced lung lesions from normal lung CT images using hard C-means algorithm

Marcelo A. Castro; David Thomasson; Nilo A. Avila; Jennifer Hufton; Justin Senseney; Reed F. Johnson; Julie Dyall

doi:10.1117/12.2006072

Translator Disclaimer

29 March 2013 Optimization of automated segmentation of monkeypox virus-induced lung lesions from normal lung CT images using hard C-means algorithm

Marcelo A. Castro, David Thomasson, Nilo A. Avila, Jennifer Hufton, Justin Senseney, Reed F. Johnson, Julie Dyall

Author Affiliations +

Proceedings Volume 8672, Medical Imaging 2013: Biomedical Applications in Molecular, Structural, and Functional Imaging; 867222 (2013) https://doi.org/10.1117/12.2006072
Event: SPIE Medical Imaging, 2013, Lake Buena Vista (Orlando Area), Florida, United States

Abstract

Monkeypox virus is an emerging zoonotic pathogen that results in up to 10% mortality in humans. Knowledge of clinical manifestations and temporal progression of monkeypox disease is limited to data collected from rare outbreaks in remote regions of Central and West Africa. Clinical observations show that monkeypox infection resembles variola infection. Given the limited capability to study monkeypox disease in humans, characterization of the disease in animal models is required. A previous work focused on the identification of inflammatory patterns using PET/CT image modality in two non-human primates previously inoculated with the virus. In this work we extended techniques used in computer-aided detection of lung tumors to identify inflammatory lesions from monkeypox virus infection and their progression using CT images. Accurate estimation of partial volumes of lung lesions via segmentation is difficult because of poor discrimination between blood vessels, diseased regions, and outer structures. We used hard C-means algorithm in conjunction with landmark based registration to estimate the extent of monkeypox virus induced disease before inoculation and after disease progression. Automated estimation is in close agreement with manual segmentation.

Citation Download Citation

Marcelo A. Castro, David Thomasson, Nilo A. Avila, Jennifer Hufton, Justin Senseney, Reed F. Johnson, and Julie Dyall "Optimization of automated segmentation of monkeypox virus-induced lung lesions from normal lung CT images using hard C-means algorithm", Proc. SPIE 8672, Medical Imaging 2013: Biomedical Applications in Molecular, Structural, and Functional Imaging, 867222 (29 March 2013); https://doi.org/10.1117/12.2006072

ACCESS THE FULL ARTICLE