abstract

• Objective: Diagnosis of acute appendicitis may be challenging. The current imaging methods have modest sensitivities and specificities. This study aimed to evaluate skin infrared thermal (IRT) imaging as a diagnostic adjunct for acute appendicitis in adults. Methods: We conducted a prospective observational study in 51 patients with appendicitis, 20. with abdominal pain (no appendicitis) and 51 healthy controls. Temperature of the skin overlying abdominal lower quadrants, right (RLQ) and left (LLQ), were recorded and compared between groups. Using a machine learning algorithm, we identified which parameters discriminated between groups and compared the diagnostic accuracy of IRT to the final diagnosis. Results: RLQ temperature >34 °C discriminate healthy controls and patients with acute abdomen; a delta >0.35 °C in RLQ vs LLQ discriminate patients with and without appendicitis. Compared to the ultrasound-based diagnosis, IRT has a sensitivity of 91%25 and a specificity of 56%25 to diagnose appendicitis. Conclusions: IRT could be an adjunct tool in the diagnosis of appendicitis in settings with limited resources. Further research is needed to confirm and translate our results into clinical practice. © 2019 Elsevier B.V.

authors

• González Contreras Francisco Javier
• Guevara-Torres L.
• Kolosovas Machuca Eleazar Samuel
• Martínez-Jiménez M.A.
• Ortiz-Dosal A.
• Ramírez-GarcíaLuna J.L.
• Vera-Bañuelos L.R.

publication date

• 2020-01-01

funding provided via

• Consejo Nacional de Ciencia y Tecnología, CONACYT  Grant
• Fonds de Recherche en Santé Respiratoire, FRSR  Grant

published in

• Infrared Physics and Technology  Journal

keywords

• Acute abdomen; Appendicitis; Infrared imaging; Machine learning Clinical research; Computerized tomography; Diagnosis; Infrared imaging; Learning algorithms; Learning systems; Machine learning; Acute abdomen; Acute appendicitis; Appendicitis; Clinical practices; Diagnostic accuracy; Healthy controls; Observational study; Proof of concept; Thermography (imaging)

Digital Object Identifier (DOI)

• 10.1016/j.infrared.2019.103165

start page

end page

volume

• 105