Application of architecture using AI in the training of a set of pixels of the image at aid decision-making diagnostic cancer

Abstract

Medical Medicine specialists are using images more and more as support in decision making in the identification of pathologies with greater complexity and severity in the specific case. Predictive and diagnostic models in images with neoplasia (collagen V) associated with exposure to asbestos fibers were studied. In this article we intend to initially assess machine learning on the basis containing 100 images classified by specialists with characteristics of normality and abnormality to aid diagnosis. Therefore, the objective is to analyze and use the concepts of learning by the technique of artificial intelligence with neural networks that culminated in significant advances of 81% of correct answers in the image diagnostics and to propose the application of the Paraconsistent Standard Analyser Unit of Artificial Neural Networks in order to categorize the degree of abnormality (normal, almost normal, almost abnormal, abnormal) with the use of the extreme and non-extreme states of Paraconsistent Logic and thus support specialists in decision making in the diagnosis of cancer.