Gastric Lymph nodes detection based on visual saliency and dictionary learning

Abstract

Lymph nodes are key signs of cancer diagnosis for abdominal CT scans assessment. A lymph node detection scheme based on CT sequences images which only contain stomach and its surrounding tissue is proposed in this paper. However, lymph nodes are easily confused with its surrounding fat and vessels, and we firstly uses an improved visual attention algorithm to extract all suspicious targets. Then feature extraction and dictionary learning are utilized for classification and recognition to real lymph nodes. Experimental results show that the proposed detection scheme can detect effectively and accurately Lymph nodes.