Termografia em lesões por pressão

Abstract

Pressure injuries (PI) represent a major public health problem, especially among bedridden patients, older adults, and individuals hospitalized in intensive care units or general wards. Currently, their assessment is primarily based on visual inspection and risk assessment tools, such as the Braden Scale, which is widely used in Brazil. However, these methods may present limitations because they rely on professional interpretation, making the monitoring of lesion progression and healing less precise. Despite the large number of studies on these tools, there is still no consistent evidence demonstrating their effectiveness in reducing the incidence of pressure injuries. Furthermore, international guidelines indicate that each scale has limitations and does not assess all risk factors involved in the development of these lesions. This study aimed to analyze the thermal behavior of pressure injuries using thermographic images. Individuals of both sexes, aged between 20 and 90 years, admitted to the intensive care unit of a reference hospital in Curitiba, Paraná, Brazil, were included in the study. Patients unable to undergo repositioning were excluded to standardize thermographic image acquisition. Initially, thermograms from five participants with pressure injuries at different stages were analyzed, comparing injured areas with healthy tissue regions. Higher temperatures were associated with regions of increased vascularization, whereas necrotic areas exhibited lower temperatures. Subsequently, the Pennes bioheat equation was applied to estimate blood perfusion under three conditions: no injury, intermediate-stage injury, and advanced-stage injury. The results indicated near-zero perfusion in the no-injury condition, positive perfusion in the intermediate stage (𝜔ˆ = +0.01023), and negative perfusion in the advanced stage (𝜔ˆ = −0.000272), suggesting a progressive reduction in blood flow as the injury advanced. In the final stage, thermograms from 30 patients were analyzed, totaling 43 contour curves initially and an additional 28 curves for external validation. Regions of interest were segmented using the K-means algorithm with a thermal threshold of 34.9 °C. Injury risk was classified using the Naive Bayes method, enabling the development of a thermographic color scale for pressure injury risk assessment based on Bayesian classification. The proposed model achieved an accuracy of 90.0%, sensitivity of 89.47%, specificity of 90.90%, and a Cohen’s kappa coefficient of 0.81. It is concluded that infrared thermography, integrated with mathematical and probabilistic models, constitutes a promising tool for multidisciplinary healthcare teams, enabling the objective and early identification of tissue alterations with high prognostic potential.