Advance Thermography diagnostics: automatic algorithm to find out acute sinusitis

Abstract

Development of medical imaging techniques (including medical thermography) provided an opportunity to mine data that can be explored for the development of clinical decision support systems. Aim. Find out quantitative thermography criteria for acute rhinosinusitis and implement these criteria into automated analysis protocols to create a diagnostic complex suitable for use by professionals of any medical speciality. Material, methods. After necessary adaptation, facial thermography by mean of thermography camera ТВС300-меd LLC “СТК СИЛАР” Russia, (384x288 pixel, 30 mK) was performed in 100 healthy volunteers and 305 patients with suspected acute rhinosinusitis (in 173 of them diagnosis was supported). Resulted thermograms were processed in “cloud service” TVision (LLC “Dignosys”, Russia). Point and regions of interest (ROI), namely, point in the middle of the line, connected inner edge of eyebrows (Тср) and ROI, covered the projections of maxillary sinuses were marked automatically. Difference between temperature in Тср point and mean temperature in each ROI (ΔТr/ΔТl), as well as mean temperature between ROI were calculated (ΔTпаз). Result. The best results ( sensitivity 82%, specificity 74%, accuracy 78%) were received when complex of ΔТr/ΔТl ≤0 and module |ΔTпаз| ≥0,5°C (both or at least one criterion) was applied. Based on these data, automatic algorithm of thermography acute rhinosinusitis diagnostics was created to use in primary medical care routine as clinical decision support systems.