Quantification of radiodermatitis through image processin

• Autor(a) principal: Naranjo, Ignacio Agustín Verdugo
• Data de Publicação: 2019
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Biblioteca Digital de Teses e Dissertações da USP
• Texto Completo: https://www.teses.usp.br/teses/disponiveis/59/59135/tde-27092021-162825/
• topic: Image processing; Processamento de imagens; Radiodermatitis; Radiodermite; Radioterapia; Radiotherapy

Radiodermatitis is an ionizing radiation acute reaction of the skin. The appearance of the lesion presents a red visual appearance known as erythema, caused by an increase in subpapillary vascular plexus blood volume. Erythema rating is currently done qualitatively using the \"RTOG / EORTC Late Radiation Morbidity Scoring Schema.\" At present, there is no quantitative method to assess the degree of injury that is affecting the skin throughout Breast Cancer therapy used in the clinical setting. This study proposes a novel method using digital, polarized light images to evaluate erythema. After the approval of the Research and Ethical Committee of the University of São Paulo, 23 breast cancer patients (>18 years old) were randomly chosen with different skin colour, TUMOUR types, surgery history, treatment type, and were followed up throughout their treatments. Circularly polarized light digital images of the patients were taken along the treatment. For each visit day, eight pictures were taken in 2 different positions (frontal and lateral) in different setups. Image registration between images of different days, for the same patients, was done using anatomical regions, skins marks, and tattoos in the border of the treatment field. The ROIs chosen for RGB colour-space analysis were the ones that reached at least grade 1 during treatment. To decrease the effect flash illumination variations from day to day, image intensities were normalized by the average intensity of a white stripe added outside the treatment area near the ROI. Each erythema was independently evaluated by physicians using the RTOG schema so that it could to test and validate the image method under development. For analysis, different groups were analysed: white ,brown skin, and dark skin. All three groups included hypofractionated vs. conventional treatment. All groups included patients with radiodermatitis grade 0, 1 and 2. It was verified that the RGB normalized intensities decrease as the radiodermatitis grade increases and that brown skin presents a more pronounced decrease. The most sensitive channel to radiodermatitis grade was the green one. The most statistically significant sensitivity in the image method was found in the differentiation between radiodermatitis grade 0 and 1 for the white and brown skin patients. The present study demonstrated a novel approach to evaluate radiodermatitis quantitatively. Despite similar past attempts in the literature, they all lack in the number of patients and the diversity of patients. This work presented a simple methodology that has to be further developed as an objective radiodermatitis quantification methodology to help the physician practice.