Evaluation of mean skin temperature formulas by infrared thermography

Abstract

 To study the reliabiliity of formulas for calculating mean skin temperature (T _sk), values were computed by 18 different techniques and were compared with the mean of 10,841 skin temperatures measured by infrared thermography. One hundred whole-body infrared thermograms were scanned in ten resting males while changing the air temperature from 40° C to 4° C. Local, regional average and mean skin temperatures were obtained using an image processing system. The agreement frequency, defined as the percentage of the calculated T _sk values which agreed with the corresponding infrared thermographic T _sk within ±0.2° C, ranged for with the various formulas from 7% to 80%. In many sites, the local skin temperature did not coincide with the regional average skin temperature. When the local skin temperatures which showed the highest percentage similarity to the regional average skin temperature within ±0.4° C were applied to the formula, the agreement frequency was markedly improved for all formulas. However, the agreement frequency was not affected by changing the weighting factors from specific constants to individually measured values of regional surface area. By applying the physiologically reliable accuracy range of ±0.2° C in the moderate and ±0.4° C in the cool condition, agreement frequencies of at least 95% were observed in formulas involving seven or more skin temperature measurement sites, including the hand and foot. We conclude that calculation of a reliable mean skin temperature must involve more than seven skin temperature measurement sites regardless of ambient temperature. Optimal sites for skin temperature measurement are proposed for various formulas.