Thyroid nodules investigation based on an artificial neural network

Abstract

The image processing techniques widely use ultrasound images. Among them, the artificial neural network techniques (ANN) are successfully used in the field of segmentation and recognition of various objects in digital images. This paper proposes a more robust and efficient method for segmenting thyroid nodules in ultrasound images. We proposed and successfully developed a Computer Aided Diagnosis (CAD) software application based on an ANN to automatically identify the presence of the thyroid nodules. The CAD algorithm has the following steps: de-noising, segmentation and ANN classification. Finally, we tested and validated the CAD efficiency.