Ultrasound Echotexture Analysis as a Diagnostic Adjunct for Naja atra Envenomation Versus Cellulitis in Acute Limb Swelling
Open Access
- 30 October 2026
- journal article
- research article
- Published by MDPI AG in Diagnostics
- Vol. 16 (21) , 3594
Abstract
Objectives: Naja atra (N. atra) envenomation and cellulitis frequently present with overlapping clinical features of acutely swollen extremities, creating a diagnostic dilemma during the early clinical course of acute limb swelling. This study investigated whether ultrasound imaging with echotexture analysis can effectively differentiate these two conditions. Methods: This retrospective cross-sectional study conducted at Taichung Veterans General Hospital included 115 ultrasound images from 84 patients categorized into three groups: confirmed N. atra envenomation (n = 45), confirmed non-snakebite cellulitis (n = 40), and controls (n = 30). Gray-level co-occurrence matrix (GLCM) features including contrast, correlation, energy, and homogeneity, along with the grayscale value and the power Doppler vascularity-to-ROI ratio, were quantitatively computed from ultrasound images. Group comparisons used generalized estimating equations to account for multiple images from the same patient, with Bonferroni correction for pairwise comparisons. Diagnostic accuracy was evaluated using receiver operating characteristic (ROC) curve analysis. Results: Significant differences among the three groups were found for contrast (p < 0.001), correlation (p < 0.001), homogeneity (p = 0.005), and the vascularity-to-ROI ratio (p < 0.001). In pairwise comparisons between N. atra envenomation and cellulitis, contrast was significantly higher (0.273 ± 0.068 versus 0.230 ± 0.040, p = 0.001) and homogeneity was lower (0.878 ± 0.020 versus 0.890 ± 0.017, p = 0.004) in the N. atra group. ROC curve analysis for discriminating N. atra envenomation from cellulitis yielded an Area Under the Curve of 0.707 (95% confidence interval, 0.591–0.824) for contrast and 0.684 (95% confidence interval, 0.568–0.800) for homogeneity. At an optimal cutoff of ≥0.243, contrast achieved 73.3% sensitivity and 70.0% specificity. Conclusions: Echotexture analysis, specifically contrast and homogeneity, may differentiate N. atra envenomation from cellulitis. These parameters may reflect the heterogeneous tissue destruction induced by venom compared to the diffuse inflammation seen in cellulitis. These proof-of-concept findings require validation in larger prospective cohorts before clinical application.Keywords
Funding Information
- Taichung Veterans General Hospital (TCVGH-1150802D)
- Taichung Veterans General Hospital (TCVGH-1155101C)
- Taichung Veterans General Hospital (TCVGH-1155101D)
- National Science and Technology Council (NSTC 114-2314-B-075A-010)
- Taiwan Food and Drug Administration (MOHW115-FDA-D-212-134451)
This publication has 10 references indexed in Scilit:
- Value of multimodal ultrasound in the assessment of snakebiteBMC Medical Imaging, 2025
- Clinical predictors of early surgical intervention in patients with venomous snakebitesEuropean Journal of Medical Research, 2023
- Echotexture Analysis of L4 Supraspinous Enthesis in Ankylosing SpondylitisJournal of Medical Ultrasound, 2021
- Texture Analysis of Supraspinatus Ultrasound Image for Computer Aided Diagnostic SystemHealthcare Informatics Research, 2016
- Sonographic signs of snakebiteClinical Toxicology, 2014
- Snake biteThe Lancet, 2010
- ABSCESS: Applied Bedside Sonography for Convenient Evaluation of Superficial Soft Tissue InfectionsAcademic Emergency Medicine, 2005
- Taiwan’s venomous snakebite: epidemiological, evolution and geographic differencesTransactions of the Royal Society of Tropical Medicine and Hygiene, 2004
- Nonparametric Analysis of Clustered ROC Curve Data1997
- Textural Features for Image ClassificationIEEE Transactions on Systems, Man, and Cybernetics, 1973