Related Experiment Video
Updated: Aug 6, 2026
Quantitative Visualization and Detection of Skin Cancer Using Dynamic Thermal Imaging
Published on: May 5, 2011
Handheld infrared thermography for triage of LVAD driveline infection: A prospective pilot diagnostic accuracy study
Theofanis Mavrepis1, Stijn Legtenberg1, Peter van der Meer1
1University of Groningen, Departement of Cardiology, University Medical Center Groningen, Groningen, The Netherlands.
Background:
Existing approaches to diagnose driveline infections (DLI) in patients with durable left ventricular assist devices (LVADs) differ in speed, cost, and specificity. Handheld infrared thermography may support DLI through rapid, easy, non-contact triage by detecting local temperature elevation.
Methods:
In this prospective, single-center diagnostic accuracy study, adult LVAD patients underwent handheld thermal imaging of the driveline exit site. Performance of thermography in predicting clinically diagnosed DLI was assessed using receiver operating characteristic analysis (AUC with 95% confidence intervals [CI], DeLong). The threshold derived using Youden's J was considered exploratory and intended for internal within-study performance estimation only; sensitivity, specificity, PPV, and NPV were calculated with exact (Clopper-Pearson) 95% CIs. Associations were assessed using logistic regression and Firth's penalized logistic regression; correlation with high-sensitivity C-reactive protein was assessed using Spearman's ρ.
Results:
Fifty patient episodes were analyzed, of which 6 (12%) had DLI. Median peri-driveline temperature was higher in patients with DLI than without DLI (36.1°C vs 34.4°C; p = 0.005). Peri-driveline temperature demonstrated good discrimination for DLI (AUC 0.86, 95% CI 0.72-1.00). The threshold of 34.7°C yielded sensitivity 100% (95% CI 54-100), specificity 61% (95% CI 45-76), PPV 26% (95% CI 10-48), and NPV 100% (95% CI 87-100). Each 1°C increase in temperature was associated with higher odds of DLI in Firth regression (OR 3.7, 95% CI 1.4-14.5; p = 0.006). Temperature and hs-CRP were weakly correlated (ρ = 0.11; p = 0.45). Sensitivity analysis, including a second infection episode, produced similar discrimination.
Conclusions:
Handheld infrared thermography showed good discrimination for DLI, and no false negatives were observed at a low threshold, albeit with limited specificity. These findings support further evaluation as a triage adjunct rather than a stand-alone diagnostic test.
