Takahiro Nakajima1, Masato Shingyoji2, Takashi Anayama3, Hideki Kimura4, Kazuhiro Yasufuku5, Ichiro Yoshino6. 1. Department of General Thoracic Surgery, Chiba University Graduate School of Medicine, Chiba, Japan; Division of Thoracic Diseases, Chiba Cancer Center, Chiba, Japan; Division of Thoracic Surgery, Toronto General Hospital, University Health Network. Electronic address: takahiro_nakajima@med.miyazaki-u.ac.jp. 2. Division of Thoracic Diseases, Chiba Cancer Center, Chiba, Japan. 3. Department of Surgery II, Kochi Medical School, Kochi University, Kochi, Japan; Division of Thoracic Surgery, Toronto General Hospital, University Health Network. 4. Division of Thoracic Diseases, Chiba Cancer Center, Chiba, Japan; Department of Thoracic Surgery, Saiseikai Narashino Hospital, Chiba, Japan. 5. Division of Thoracic Surgery, Toronto General Hospital, University Health Network. 6. Department of General Thoracic Surgery, Chiba University Graduate School of Medicine, Chiba, Japan.
Abstract
OBJECTIVE: The aim of this study was to analyze the spectral features of the radiofrequency of lymph nodes during endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) and to determine its diagnostic value for detecting metastatic nodes in patients with lung cancer. METHODS: Ultrasound spectrums of lymph nodes during EBUS-TBNA were retrospectively analyzed. A linear regression of frequency spectrum and the ultrasonic spectral parameters midband-fit, slope, and intercept were calculated. Mean values for these parameters within lymph nodes were computed. The cutoff values for each parameter for distinguishing metastatic vs benign lymph nodes were first determined within the training set; these cutoff values were then applied to the testing set for validation. RESULTS: Overall, 362 lymph nodes (112 metastatic, 250 benign) were analyzed as the training set, and 284 lymph nodes (74 metastatic, 210 benign) were evaluated as the testing set. In the training set, all of the parameters showed a significant difference between metastatic and benign lymph nodes (P < .001). The metastatic nodes tended to show low midband-fit, high slope, and low intercept. When midband-fit and intercept were combined, the diagnostic accuracy was maximized in the training set. In the testing set, the combination of intercept and slope produced the highest diagnostic accuracy, with the following outcomes: sensitivity, 79.7%; specificity, 84.3%; diagnostic accuracy, 83.1%; positive predictive value, 64.1%; and negative predictive value, 92.2%. CONCLUSIONS: Metastatic lymph nodes possess unique ultrasonic spectrum features, and spectrum analysis can be used as a novel diagnostic tool for differentiating between benign and malignant nodes in patients with lung cancer.
OBJECTIVE: The aim of this study was to analyze the spectral features of the radiofrequency of lymph nodes during endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) and to determine its diagnostic value for detecting metastatic nodes in patients with lung cancer. METHODS: Ultrasound spectrums of lymph nodes during EBUS-TBNA were retrospectively analyzed. A linear regression of frequency spectrum and the ultrasonic spectral parameters midband-fit, slope, and intercept were calculated. Mean values for these parameters within lymph nodes were computed. The cutoff values for each parameter for distinguishing metastatic vs benign lymph nodes were first determined within the training set; these cutoff values were then applied to the testing set for validation. RESULTS: Overall, 362 lymph nodes (112 metastatic, 250 benign) were analyzed as the training set, and 284 lymph nodes (74 metastatic, 210 benign) were evaluated as the testing set. In the training set, all of the parameters showed a significant difference between metastatic and benign lymph nodes (P < .001). The metastatic nodes tended to show low midband-fit, high slope, and low intercept. When midband-fit and intercept were combined, the diagnostic accuracy was maximized in the training set. In the testing set, the combination of intercept and slope produced the highest diagnostic accuracy, with the following outcomes: sensitivity, 79.7%; specificity, 84.3%; diagnostic accuracy, 83.1%; positive predictive value, 64.1%; and negative predictive value, 92.2%. CONCLUSIONS: Metastatic lymph nodes possess unique ultrasonic spectrum features, and spectrum analysis can be used as a novel diagnostic tool for differentiating between benign and malignant nodes in patients with lung cancer.