BACKGROUND: Endoscopic ultrasonography (EUS) has been shown to be useful for detecting lymph node metastasis in esophageal cancer. The evaluation of nodal metastasis requires both objective and subjective analyses. In the present study, mediastinal lymph nodes in superficial esophageal carcinoma (SEC) were examined by both EUS appearance and histography, using NIH image software. METHODS: One hundred and seventy-one lymph nodes of 56 patients with SEC were detected by EUS. These lymph nodes were diagnosed by type classification, based on boundary and internal echo, and by the construction of internal echo histograms using NIH image software. The results were compared with the histological findings. RESULTS: . The sensitivity, specificity, and accuracy in assessing mediastinal lymph node metastasis by type classification were 83.3%, 88.2%, and 87.7%, respectively. The mean and SD of the histogram correlated well with histological findings and type classification ( P < 0.0001). All lymph nodes with a mean value of less than 185 of the histogram were negative nodes. When positive nodes by type classification were reevaluated according to the threshold value of 185 using the histogram, the sensitivity, specificity, and accuracy improved to 83.3%, 100%, and 98.2%, respectively. CONCLUSIONS: Type classification assisted by histography improved the diagnostic accuracy of mediastinal lymph node metastasis in SEC.
BACKGROUND: Endoscopic ultrasonography (EUS) has been shown to be useful for detecting lymph node metastasis in esophageal cancer. The evaluation of nodal metastasis requires both objective and subjective analyses. In the present study, mediastinal lymph nodes in superficial esophageal carcinoma (SEC) were examined by both EUS appearance and histography, using NIH image software. METHODS: One hundred and seventy-one lymph nodes of 56 patients with SEC were detected by EUS. These lymph nodes were diagnosed by type classification, based on boundary and internal echo, and by the construction of internal echo histograms using NIH image software. The results were compared with the histological findings. RESULTS: . The sensitivity, specificity, and accuracy in assessing mediastinal lymph node metastasis by type classification were 83.3%, 88.2%, and 87.7%, respectively. The mean and SD of the histogram correlated well with histological findings and type classification ( P < 0.0001). All lymph nodes with a mean value of less than 185 of the histogram were negative nodes. When positive nodes by type classification were reevaluated according to the threshold value of 185 using the histogram, the sensitivity, specificity, and accuracy improved to 83.3%, 100%, and 98.2%, respectively. CONCLUSIONS: Type classification assisted by histography improved the diagnostic accuracy of mediastinal lymph node metastasis in SEC.