Comparison of the diagnostic accuracy of three infrared imaging methods in evaluating patients with presumptive complex regional pain syndrome, Type 1

Objective: To calculate the sensitivity, specificity, positive predictive value, negative predictive value, and receiver operating characteristics (ROC) curves analysis for three separate and distinct infrared (IR) imaging methodologies to evaluate patients with presumptive Complex Regional Pain Syndrome (CRPS), Type 1. The three IR imaging methodologies evaluated in this study were: (1) functional infrared imaging (fIR), which obtains and evaluates three distinct IR signature indices; (2) quantitative IR imaging, which obtains side-to-side computer-generated quantitative data of the region of interest (ROI); and (3) qualitative IR analysis of the ROI. 

Description: CRPS, formally known as Reflex Sympathetic Dystrophy, is a potentially disabling condition characterized by regional pain. This poorly understood and difficult to diagnose disease is the result of a multifactorial interplay between altered somatosensory, motor, autonomic, and inflammatory systems. Peripheral and central sensitization is a common feature in CRPS. 

Setting: All studies were performed at the Colorado Infrared Imaging Center, a freestanding diagnostic infrared imaging center. 

Methods: A retrospective chart review of 299 consecutive patients referred to the infrared diagnostic imaging center to rule out CRPS were evaluated to compare the physical examination findings with three separate and distinct IR imaging methodologies. Their primary care physicians referred the patients with a presumptive diagnosis of CRPS. Prior to IR imaging, all patients underwent a focused clinical examination utilizing the modified (Budapest) International Association for the Study of Pain criteria for diagnosing CRPS as the gold standard. The three separate and distinct IR imaging methodologies were compared with the clinical examination findings. Test validity measures (sensitivity, specificity, positive predictive value, negative predictive value, p-value-difference between the specified test and chance) were calculated. Receiver operating characteristics (ROC) were analysed to determine the area under the curve (AUC) for each of the three individual IR diagnostic testing methodologies. 

Results: 1. The fIR method, utilizing three distinct IR signature indices, demonstrated a high sensitivity of 83.9% (95% confidence interval (CI): 72.3 and 92.0), specificity of 99.2% (95% CI: 97.0% and 99.9%), positive predictive value of 96.3% (95% CI: 87.3 and 99.6), negative predictive value of 95.9% (95% CI: 92.6 and 98.0) and had a significant association with the gold standard method (p < .0001). The ROC curve analysis (95% CI) revealed a 91.5% (86.9%, 96.2%) area under the curve (AUC). 

2. The quantitative IR imaging method, which obtains a side-to-side computer generated quantitative data of the ROI, demonstrated a sensitivity of 38.7% (95% CI: 26.6% and 51.9%), specificity of 84.8% (95% CI: 79.6 and 89.1), positive predictive value of 40.0% (95% CI: 27.6 and 53.5), negative predictive value of 84.1% (95% CI: 78.8% and 88.5) and a significant inverse association with the gold standard method (p<.0004) when incorporating an equal or greater to 1°C delta T (AT). The ROC curve analysis (95% CI) revealed a low 61.8% (55.2%, 68.3%) AUC when utilizing this diagnostic IR imaging methodology for evaluating patients with presumptive CRPS. 

3. The qualitative IR imaging method, which compares an equal or greater to 1°C AT utilizing a 10°C colour palette comparing a side-to-side ROI, demonstrated a sensitivity of 91.9% (95% CI: 82.2 and 97.3), specificity of 14.4% (95% CI: 10.1 and 19.5), positive predictive value of 21.9% (95% CI: 17.1 and 27.5), negative predictive value of 87.2% (95% CI: 72.6 and 95.7 and was not significantly associated with the gold standard method (p>0.05). The ROC curve analysis (95% CI) revealed a 53.1% (95% CI: 49.1% to 57.2%) area under the curve (AUC) when utilizing this diagnostic IR imaging methodology for evaluating patients with presumptive CRPS. 

Conclusions: fIR is a highly sensitive and specific IR imaging diagnostic methodology that formulates an accurate interpretive impression (assessment) in patients with presumptive CRPS. Quantitative and qualitative IR imaging methodologies performed poorly on test validation measures; these IR imaging methodologies are not recommended for the evaluation of patients with presumptive CRPS due to lack of validity, thereby making these methods unable to differentiate between patients with chronic pain syndromes that mimic CRPS.