Point-of-care infrared thermal imaging for differentiating venomous snakebites from non-venomous and dry bites

Abstract

Background: Local envenomation following snakebites is accompanied by thermal changes, which could be visualized using infrared imaging. We explored whether infrared thermal imaging could be used to differentiate venomous snakebites from non-venomous and dry bites.

Methods: We prospectively enrolled adult patients with a history of snakebite in the past 24 hours presenting to the emergency of a teaching hospital in southern India. A standardized clinical evaluation for symptoms and signs of envenomation including 20-minute whole-blood clotting test and prothrombin time was performed to assess envenomation status. Infrared thermal imaging was done at enrolment, 6 hours, and 24 hours later using a smartphone-based device under ambient conditions. Processed infrared thermal images were independently interpreted twice by a reference rater and once by three novice raters.

Findings: We studied 89 patients; 60 (67%) of them were male. Median (IQR) time from bite to enrolment was 11 (6.5-15) hours; 21 (24%) patients were enrolled within 6 hours of snakebite. In all, 48 patients had local envenomation with/without systemic envenomation, and 35 patients were classified as non-venomous/dry bites. Envenomation status was unclear in six patients. At enrolment, area of increased temperature around the bite site (Hot spot) was evident on infrared thermal imaging in 45 of the 48 patients with envenomation, while hot spot was evident in only 6 of the 35 patients without envenomation. Presence of hot spot on baseline infrared thermal images had a sensitivity of 93.7% (95% CI 82.8% to 98.7%) and a specificity of 82.9% (66.3% to 93.4%) to differentiate envenomed patients from those without envenomation. Interrater agreement for identifying hot spots was more than substantial (Kappa statistic >0.85), and intrarater agreement was almost perfect (Kappa = 0.93). Paradoxical thermal changes were observed in 14 patients.

Conclusions: Point-of-care infrared thermal imaging could be useful in the early identification of non-venomous and dry snakebites.

Fig 1. Study catchment area highlighted on political map of India.

It includes union territory of Puducherry, where the study hospital is located, and four adjoining districts in the state of Tamil Nadu. Courtesy: https://www.datawrapper.de/.

Fig 2. Infrared thermal imaging.

Visible light image showing minimal swelling of the right foot in a patient bitten by a Russell’s viper on the right third toe (Panel A). Corresponding unprocessed thermal image is shown in Panel B. Images acquired about 22 hours following the snakebite are presented here to demonstrate typical changes. Image processing was done by adjusting the lower limit of temperature range to make thermal difference between the limbs readily apparent (Panel C). Measurement of ΔTmax using box measurement tool is depicted in Panel D.

Fig 3. Flowchart depicting classification of study patients based on local swelling and coagulopathy.

Fig 4. Dotplot of ΔTmax presented by hot spot status.

Dotted lines across data points represent the median and error bars represent interquartile range.