Digital planimetry with modern electronic devices requires verification for the use in studies and in real clinical practice.
OBJECTIVE
To assess the accuracy and reproducibility of measuring the surface area of phantom skin lesions of irregular shape on a curved surface using the LesionMeter mobile application.
MATERIAL AND METHODS
We conducted an experimental comparative study of measuring the surface area of irregular figures on a curved surface using the application and the reference method (template). A correlation analysis was carried out, bias and relative bias of measurements were determined with an additional estimation of allocation, statistical significance, quality control of linear models and consistency of measurement results. The reproducibility of the method was also evaluated.
RESULTS
Pearson’s correlation coefficient was 0.990 with 95% CI 0.985; 0.993 (p<2.2e-16 (excellent coherence)). Blend-Altman plot: mean absolute error (bias) = –0.095 (95% CI –0.16; –0.027) cm2; upper limit of bias = +0.62 (95% CI 0.51; 0.74) cm2; lower limit of bias = –0.82 (95% CI –0.93; –0.70) cm2. The maximum bias (underestimation of phantom wound area by the application) was obtained for objects with surface area 2-7.5 cm2. The relative bias measurement for figures with surface area ≈0.5 cm2 was 15% (95% CI 12; 17%). Figure area enlargement by 1 cm2 was followed by relative bias decrease by 1% (95% CI –1.6; –0.4%).
CONCLUSION
The accuracy and reliability of measuring the phantom wounds on a curved surface with the LesionMeter application correspond to accuracy of the reference method (template) and literature data on similar planimetric methods.