Image Analysis of Benign and Malignant Neck Masses

Abstract

Current techniques for differentiating benign and malignant neck masses in computed tomography scans rely on basic features such as size and image density. Advances in digital imaging have led to development of sophisticated computerized image processing tools. Four different image processing algorithms were used to compare 10 patients with neck masses positive for squamous cell carcinoma and 11 patients with benign neck nodes. Statistically significant differences were seen between the malignant and benign masses for edge analysis and Fourier analysis. Significant differences were also seen in several texture parameters, although these results were considered less reliable. There were no differences between the two groups when compared by histogram analysis. It is concluded that both edge analysis and Fourier analysis can provide additional information to distinguish benign and malignant neck masses.