Automatic investigation of the thyroid nodules from US images
Dorin Bibicu, Luminița Moraru, Simona Moldovanu · AIP conference proceedings · 2012
The aim of the current study is to develop a standalone Computer Aided Diagnosis (CADi) application in order to improve the accuracy of diagnosis of the thyroid nodules in ultrasound images. The automatic investigation process is based on a new algorithm that allows the digital ultrasound images processing by filtering, binarization and segmentation techniques. The presence of thyroid nodules in processed images is automatically highlighted. The software application allows the possibility to perform geometric measurements on the investigated images. The processed images, the automatic diagnostic and the geometric measurements can be saved in a database attached to the application. In our experimental analysis 90 US images were used: 45 images of healthy thyroid and 45 that present thyroid nodules. The CADi application has a good accuracy of 87.77%.