OBJECTIVE: To assess computed tomography (CT) evaluation of mediastinal nodes in non-small cell lung cancer to predict metastatic involvement by measurement of their axis and surface area in the coronal plane, as compared to standard short-axis measures in the axial plane. METHODS: Evaluation of mediastinal nodes was retrospectively performed on CT scans of 100 patients before thoracotomy. In all patients, mediastinal dissection was performed in the appropriate stations (n=264) according to the side (59 right, 41 left) of the tumor. Measurements of short axis and surface area of the largest node in each dissected station were performed on axial and coronal planes. RESULTS: By using the standard threshold of axial short axis ≥10 mm, sensitivity and specificity were 25% and 98%, respectively. Areas under receiver operating characteristic curves were 0.828 and 0.821 for axial short axis and axial surface area data. For comparison, areas under receiver operating characteristic curves were 0.843 and 0.845 for coronal short axis and coronal surface area data, respectively. So, for a specificity of 98%, sensitivity was 29% for coronal short axis ≥11 mm and 33% for coronal surface area ≥123 mm(2). When using axial short axis ≥10 mm or coronal surface area ≥120 mm(2), sensitivity was 45%, whereas specificity remained at 96%. CONCLUSION: Coronal measurements of mediastinal nodes give a slightly albeit non-significant improvement of diagnostic accuracy over axial ones. If both axial short axis and coronal surface area are taken into account, accuracy is improved.
OBJECTIVE: To assess computed tomography (CT) evaluation of mediastinal nodes in non-small cell lung cancer to predict metastatic involvement by measurement of their axis and surface area in the coronal plane, as compared to standard short-axis measures in the axial plane. METHODS: Evaluation of mediastinal nodes was retrospectively performed on CT scans of 100 patients before thoracotomy. In all patients, mediastinal dissection was performed in the appropriate stations (n=264) according to the side (59 right, 41 left) of the tumor. Measurements of short axis and surface area of the largest node in each dissected station were performed on axial and coronal planes. RESULTS: By using the standard threshold of axial short axis ≥10 mm, sensitivity and specificity were 25% and 98%, respectively. Areas under receiver operating characteristic curves were 0.828 and 0.821 for axial short axis and axial surface area data. For comparison, areas under receiver operating characteristic curves were 0.843 and 0.845 for coronal short axis and coronal surface area data, respectively. So, for a specificity of 98%, sensitivity was 29% for coronal short axis ≥11 mm and 33% for coronal surface area ≥123 mm(2). When using axial short axis ≥10 mm or coronal surface area ≥120 mm(2), sensitivity was 45%, whereas specificity remained at 96%. CONCLUSION: Coronal measurements of mediastinal nodes give a slightly albeit non-significant improvement of diagnostic accuracy over axial ones. If both axial short axis and coronal surface area are taken into account, accuracy is improved.