Maho Tsubakimoto1, Tsuneo Yamashiro2, Yukari Tamashiro3, Sadayuki Murayama4. 1. Department of Radiology, Graduate School of Medical Science, University of the Ryukyus, 207 Uehara, Nishihara, Okinawa 903-0215, Japan. Electronic address: tsubaki@med.u-ryukyu.ac.jp. 2. Department of Radiology, Graduate School of Medical Science, University of the Ryukyus, 207 Uehara, Nishihara, Okinawa 903-0215, Japan. Electronic address: clatsune@yahoo.co.jp. 3. Department of Radiology, Graduate School of Medical Science, University of the Ryukyus, 207 Uehara, Nishihara, Okinawa 903-0215, Japan. Electronic address: k178705@eve.u-ryukyu.ac.jp. 4. Department of Radiology, Graduate School of Medical Science, University of the Ryukyus, 207 Uehara, Nishihara, Okinawa 903-0215, Japan. Electronic address: sadayuki@med.u-ryukyu.ac.jp.
Abstract
PURPOSE: To assess the ability of parameters derived from computed tomography (CT) histograms and the maximum standardized uptake value (SUVmax) of 18F-fludeoxyglucose-positron emission tomography/CT (FDG-PET/CT) images to distinguish solid lung adenocarcinomas from squamous cell carcinomas and to determine if these parameters are correlated. METHODS: This study comprised 43 consecutive patients with solid lung cancer (<3 cm in diameter), who underwent both plain chest CT and FDG-PET/CT (adenocarcinoma, n = 25; squamous cell carcinoma, n = 18). Density histograms of targeted lung cancers were created from chest CT images, and kurtosis and skewness were calculated. On FDG-PET/CT, the SUVmax without/with respiratory gating (RG) were calculated for each lesion. The values for the 4 parameters determined for patients in each diagnostic group were compared by the Mann-Whitney test. The diagnostic characteristics of the parameters were assessed by receiver operating characteristic (ROC) curve analysis. Differences between these parameters were assessed by the chi-square test. SUVmax with RG, kurtosis, and skewness were combined for binary logistic regression analysis, and the differences between the combined parameters and SUVmax with RG were also assessed. Spearman rank correlation analysis was used to determine the correlations for kurtosis or for skewness with SUVmax without/with RG. RESULTS: The differences in kurtosis and SUVmax without/with RG between the diagnostic groups were significant (kurtosis, P < 0.004; SUVmax without/with RG both P < 0.0001). ROC curve analysis indicated that each parameter (kurtosis value, skewness value, SUVmax, without/with RG) provided low-high ability to differentiate between 2 groups (area under the curve [AUC]: 0.760, 0.593, 0.900, 0.931, respectively). The ROC of the combined parameters provided the highest ability (AUC: 0.949). Both kurtosis and skewness were significantly correlated with SUVmax without/with RG. Kurtosis and SUVmax with RG were most strongly correlated (ρ = 0.618). CONCLUSION: Quantitative CT histogram values and SUV assessment can differentiate solid lung adenocarcinomas from squamous cell carcinomas. Kurtosis and SUVmax values were strongly correlated. The addition of RG and further combination of the parameters improved the results.
PURPOSE: To assess the ability of parameters derived from computed tomography (CT) histograms and the maximum standardized uptake value (SUVmax) of 18F-fludeoxyglucose-positron emission tomography/CT (FDG-PET/CT) images to distinguish solid lung adenocarcinomas from squamous cell carcinomas and to determine if these parameters are correlated. METHODS: This study comprised 43 consecutive patients with solid lung cancer (<3 cm in diameter), who underwent both plain chest CT and FDG-PET/CT (adenocarcinoma, n = 25; squamous cell carcinoma, n = 18). Density histograms of targeted lung cancers were created from chest CT images, and kurtosis and skewness were calculated. On FDG-PET/CT, the SUVmax without/with respiratory gating (RG) were calculated for each lesion. The values for the 4 parameters determined for patients in each diagnostic group were compared by the Mann-Whitney test. The diagnostic characteristics of the parameters were assessed by receiver operating characteristic (ROC) curve analysis. Differences between these parameters were assessed by the chi-square test. SUVmax with RG, kurtosis, and skewness were combined for binary logistic regression analysis, and the differences between the combined parameters and SUVmax with RG were also assessed. Spearman rank correlation analysis was used to determine the correlations for kurtosis or for skewness with SUVmax without/with RG. RESULTS: The differences in kurtosis and SUVmax without/with RG between the diagnostic groups were significant (kurtosis, P < 0.004; SUVmax without/with RG both P < 0.0001). ROC curve analysis indicated that each parameter (kurtosis value, skewness value, SUVmax, without/with RG) provided low-high ability to differentiate between 2 groups (area under the curve [AUC]: 0.760, 0.593, 0.900, 0.931, respectively). The ROC of the combined parameters provided the highest ability (AUC: 0.949). Both kurtosis and skewness were significantly correlated with SUVmax without/with RG. Kurtosis and SUVmax with RG were most strongly correlated (ρ = 0.618). CONCLUSION: Quantitative CT histogram values and SUV assessment can differentiate solid lung adenocarcinomas from squamous cell carcinomas. Kurtosis and SUVmax values were strongly correlated. The addition of RG and further combination of the parameters improved the results.