Diagnosis of abdominal mass in ultrasound images using linear collaborative discriminant regression classification

An abdominal ultrasound image is a practical way of checking internal organs. This paper intends to develop an advanced model for diagnosing abdominal masses using US images. This detection technique is accomplished in two stages including Feature extraction and Classification. During feature extraction, texture feature is extracted from US image by adaptive gradient location and orientation histogram (AGLOH). Later in the classification stage, linear collaborative discriminant regression classification (LCDRC) model is used to classify whether the image is normal or abnormal. The classification error produced by the collaborative demonstration is lesser when evaluated with the error produced by the demonstration of single class. Therefore, an improved diagnosis precision is achieved. The features of the proposed AGLOH method are compared with conventional techniques such as gradient location and orientation histogram (GLOH). Further, the classifier of the proposed LCDRC method is compared with conventional techniques such as SVM and NN and validates the effectiveness of the proposed method.