Preliminary study on objective evaluation algorithm of human infrared thermogram seriality and its clinical application in population with metabolic syndrome

Discussion

The clinical research of IRT has long relied on the textual description of thermograms, which is highly subjective and obviously impedes the clinical application. A major current research focus in the field of IRT is how to use quantitative methods to objectively assess the subtle features of thermograms, thereby achieving greater clinical significance.

Based on previous studies on infrared thermograms of healthy participants and patients with MS, the team found that with the increase of metabolic component abnormalities, the temperature order of each region of thermogram was disrupted, and the regional temperature distribution was inverted.27 In this study, a recursive stepwise screening method was used to obtain a stable thermal sequence structure of ROI in healthy participants, that is, T _palmlower leglower abdomen. Based on the three significant ROIs, the thermogram sequence of patients with MS was judged, and the thermal sequence values were subsequently calculated. Through internal verification, it was proved that the infrared thermogram seriality had good auxiliary diagnostic value, high sensitivity and certain specificity for MS and could be applied to the screening of MS diseases. Moreover, because the thermogram seriality only considers the relative temperature of different ROIs, it can avoid the changes of absolute temperature of thermogram caused by factors such as production equipment, calibration method and environmental temperature, and the results are relatively stable and reliable.

In addition, the results show that human surface temperature changes with age. The difference between the thermal sequence values in patients with MS and those of healthy individuals gradually decreases with age, especially in women. It can be seen from this that thermogram seriality was affected by sex and age. This thermal sequence is more stable in male and young and middle-aged patients. Because the screening of MS using IRT is mainly aimed at young groups, this algorithm has great clinical application value. Furthermore, this study revealed significant positive correlations between BMI and thermal sequences in both healthy populations and MS, suggesting that obesity exerts a more pronounced impact on thermal sequence patterns. Obesity induces increased metabolic heat production,28 while also causing systemic fat deposition with varying degrees across the body, thus diminishing the conduction of core abdominal heat. Consequently, obese individuals typically exhibit a distinct ‘abdominal cooling-palmar warming’ thermal sequence.29 The findings were observed by Chudecka et al,30 who revealed significant positive correlations between BMI and hand temperatures, contrasted by negative correlations with abdominal thermal patterns. Collectively, thermal sequence could sensitively reflect the metabolic state of the body, detect pathological changes and assist in determining disease progression.

Regarding the algorithm, the following shortcomings exist: (1) Its diagnostic value for diseases other than MS has not yet been sufficiently validated. (2) The algorithm may lack sufficient sensitivity, and thus may not be applicable, in situations where multiple ROIs simultaneously show elevated temperatures leading to no obvious change in the overall thermal sequence, or where temperature changes within ROIs are too small to cause a noticeable alteration in the thermal sequence. In addition, the study has some limitations: (1) The manual segmentation of ROIs in the thermograms may be influenced by operator subjectivity, potentially introducing bias into the analysis. (2) Randomised resampling was used for female patients with MS, which increased the risk of model overfitting, reduced sample diversity and may have led to deficiencies in capturing differences in thermal sequence patterns between healthy female participants and female patients with MS. (3) Since thermogram seriality was to rank the relative temperature of each ROI, the absolute temperature of each ROI was converted to rank order in the algorithm, which may result in the loss of some valid information. (4) Thermogram seriality is associated with sex and age. When differentiating MS cases, the AUC, sensitivity and specificity of men and young people are significantly higher than those of women and older people, which causes a certain degree of interference in the diagnosis of MS.

To enhance the model’s performance, stability and clinical applicability, future studies should include larger, more diverse and well-balanced prospective cohorts encompassing multiple regions and ethnicities, not only in patients with MS but also in those with other diseases, to further validate the diagnostic value of thermogram seriality. Regarding the issue of manual ROI segmentation, future work may benefit from the incorporation of more automated and standardised image segmentation techniques to improve accuracy and consistency in ROI segmentation. Furthermore, during thermogram analysis, absolute temperature should be cross-referenced with factors such as age, sex and BMI to avoid the loss of important information. Particular emphasis should be placed on the correct selection of ROIs, as thermograms can sensitively reflect disease-affected body areas, and the selection of ROI directly determines the diagnostic value of seriality.

Back to top