An automatic computer-aided diagnosis system for liver tumours on computed tomography images

Liver cancer, one of the more common cancer diseases that cause a large number of deaths every year, can be reduced by early detection and diagnosis. Computer-Aided Diagnosis (CAD) can play a key role in the early detection and diagnosis of liver cancer. This paper develops a novel computer-aided diagnosis system focussing on the discriminating power of statistical texture descriptors in characterizing hepatocellular (malignant) from hemangioma (benign) liver tumours. The CAD system consists of three stages: (i) automatic tumour segmentation, (ii) texture feature extraction and (iii) tumour characterization using a classifier. Specifically, four features sets, the original gray level; co-occurrence of gray level; wavelet coefficient statistics and contourlet coefficient statistics are extracted from the tumour region of interest. A probabilistic neural network classifier is used to investigate the ability of each feature set in differentiating malignant from benign tissues. The performance of the CAD system evaluated using a database of images indicates that the highest accuracy achieved is 96.7% and the highest sensitivity and specificity are 97.3% and 96%, respectively that had been obtained with the contourlet coefficient co-occurrence features. The experimental results suggest that the developed CAD system has great potential and promise in the automatic diagnosis of both benign and malignant tumours of liver.

Figure optionsDownload as PowerPoint slideHighlights
► A computer-aided diagnosis system for automatic diagnosis of liver tumours from CT images is proposed.
► Tumour region of interest is automatically segmented from CT images.
► Gray level textures, wavelet coefficient textures and contourlet coefficient textures are extracted.
► PNN classifier is employed in classifying tumour as benign or malignant with an accuracy of 96.7%.