Resumen

FERNANDEZ-LOPEZ, Claudia; RUIZ-GONZALEZ, Yusely; MOLLINEDA-DIOGO, Niurka  y  LORENZO-GINORI, Juan V. Evaluation of segmentation accuracy and area measurement in images of skin lesions. EAC [online]. 2022, vol.43, n.2, pp. 76-90.  Epub 17-Oct-2022. ISSN 1815-5928.

The healing of skin lesions is a complex process that can be supervised through periodic inspection of the condition of the lesions. With the estimation of their areas, it is possible to evaluate the response to the applied treatment, predict possible healing results and differentiate a lesion that will heal easily from another that will require prolonged treatment. This information is used in drug candidate studies, therefore high-precision measurements are required. Manual techniques that involve contact with the lesions are usually used for measurements, which is risky, prone to errors and inter- and intra-specialist variations. In this research, the segmentation of photographic images of lesions on the skin of laboratory animals was developed to select the most appropriate alternative among possible algorithms implemented for estimating the area of the lesions. To this end, the quality of several segmentation by thresholding methods to detect lesions was evaluated in a supervised manner, with the help of reference images previously segmented by experts. Different quality indicators were used, such as the Jaccard coefficient, the relative error in the automatic estimation of the area with respect to the reference, and non-parametric statistical tests. The results showed differences between the applied algorithms and coincidences between descriptive statistics and non-parametric statistics and might contribute to select an appropriate alternative among those tested. A preliminary study on the feasibility of the proposed method for the analysis of diabetic feet ulcers was also made. This research has a direct application in the pharmaceutical and health sector.

Palabras clave : lesions area; segmentation; thresholding; segmentation quality.

        · resumen en Español     · texto en Español     · Español ( pdf )