Validação do Índice Carneiro de Contraste (ICC) como métrica de avaliação de imagens mamográficas

Abstract

Breast cancer is a worrying and impactful disease, affecting millions of women around the world. It is the most common cancer among women, being the second disease that most affects women in Brazil, a statistic that is surpassed by non-melanoma skin cancer. Early detection is essential for effective treatment, and mammography is a valuable tool in this process. Breast density is a factor that can make it difficult to detect breast cancer on mammograms, since dense breasts have more fibroglandular tissue, which can hide cancerous lesions. Image processing techniques are used to improve the contrast in mammograms, making lesions more visible. Quantitative metrics, such as the Carneiro Contrast Index (CCI), are used to objectively measure contrast in images and evaluate the effectiveness of techniques used on images as part of digital processing. The aim of this work is to validate the CCI quantitative contrast analysis technique on phantom images, comparing it with other metrics and analysis techniques. During the development of the work, phantom images from different databases were analyzed, collecting data on CCI, EME, standard deviation, tube current, peak kilovoltage and radiation doses. Analysis of the results revealed that CCI and EME vary across images, with CCI showing greater overall contrast sensitivity. Acquisition parameters such as the tube current and peak kilovoltage play a key role in image quality and contrast. Increasing peak kilovoltage can result in images with greater gray scale and lower contrast, while the tube current affects the amount of radiation and signal strength in the image. The results suggest that CCI is a sensitive metric for evaluating variations in the contrast of mammography images, and the appropriate choice of acquisition parameters is crucial to optimize image quality.