Quantitative Parameters Analysis of Lymph Node Status in Rectal Cancer Based on Computed Tomography Imaging

Hongmin CaiChinese 2011

Abstract

Objective To evaluate lymph nodes in rectal cancer patients quantitatively using a computer-aided automated approach,and assess its accuracy in differentiating benign and malignant lymph nodes.Methods Patients(147) with newly diagnosed rectal cancer,underwent enhanced computed tomography(CT) examination.Patients were assigned to the benign node or malignant node group according to histopathological analysis of node samples.Images were reviewed by two radiologists blindly on workstation and making the consensus.All CT-detected lymph nodes were segmented using the Matlab software,and short-axis diameter,nodal density,fractal dimension and heterogeneity quantitative parameters of each node were measured.The relationships between lymph node metastases and CT findings were analyzed statistically using t test and X2 test.Results Benign and malignant lymph nodes were found in each size group(3,5,8 or 10 mm short-axis diameter).The fractal dimension,density and heterogeneity were higher in malignant nodes than in benign nodes,and there was a significant difference in heterogeneity between metastatic and non-metastatic lymph nodes(fractal dimension:metastasis 1.1244,non-metastasis 1.0878;P0.05;density:metastasis 108.4,non-metastasis 98.9;P0.05;heterogeneity:metastasis 0.3035,non-metastasis 0.2225;P0.05).Conclusion The involvement of lymph nodes in rectal carcinoma were found in each size group,but also the nodal density,fractal dimension and heterogeneity quantitative parameters of the lymph nodes were helpful to consider these lymph nodes stasus.Computer-aided quantitative analysis can improve the prediction of node status in rectal cancer.
BibTeX
@inproceedings{cai2011quantitative,
  author = {Hongmin Cai},
  title = {{Quantitative Parameters Analysis of Lymph Node Status in Rectal Cancer Based on Computed Tomography Imaging}},
  booktitle = {Chinese},
  year = {2011},
}

Loading related content...

Figure