Thyroid Nodule Detection using Artificial Neural Network
Xhitij A. Kesarkar, Kshama Vishwanath Kulhalli · 2021
Thyroid gland is one of the largest endocrine gland and is located below the skin and muscles at front of neck. It has an important role in maintaining the metabolism of body. Modalities like Ultrasonography, Computer Tomography (CT),Magnetic Resonance Imaging(MRI) and Computer Aided Diagnosis(CAD) are used for identification and classification of abnormalities in thyroid gland. This paper proposes a computer based system for classifying nodules in thyroid as benign or malignant in ultrasound images depending upon the extracted features. The information related to texture is calculated from both Region of Interest (ROI) blocks. Enhancement is done to improve the image followed by segmentation using Active contour without edge(ACWE).With help of this extracted features, the Multi-layer perceptron(MLP) classifies the nodule as malignant or benign. The classification results obtained the height of accuracy, sensitivity, specificity, predicted positive value and predicted negative value at 93.84%, 97.82%, 84.21%,93.75% and 94.11% respectively. These results suggests that proposed scheme accomplished the classification of ultrasound thyroid nodules as benign or malignant.