乳腺癌灶的精确分割是乳腺癌计算机辅助诊断的重要前提.在动态对比增强核磁共振成像(Dynamic contrastenhanced magnetic resonance imaging,DCE-MRI)的图像中,乳腺癌灶具有对比度低、边界模糊及亮度不均匀等特点,传统的活动轮廓模型方...乳腺癌灶的精确分割是乳腺癌计算机辅助诊断的重要前提.在动态对比增强核磁共振成像(Dynamic contrastenhanced magnetic resonance imaging,DCE-MRI)的图像中,乳腺癌灶具有对比度低、边界模糊及亮度不均匀等特点,传统的活动轮廓模型方法很难取得准确的分割结果.本文提出一种结合马尔科夫随机场(Markov random field,MRF)能量和模糊速度函数的活动轮廓模型的半自动分割方法来完成乳腺癌灶的分割,相对于专业医生的手动分割,本文方法具有速度快、可重复性高和分割结果相对客观等优点.首先,计算乳腺DCE-MRI图像的MRF能量,以增强目标区域与周围背景的差异.其次,在能量图中计算每个像素点的后验概率,建立基于后验概率驱动的活动轮廓模型区域项.最后,结合Gabor纹理特征、DCE-MRI时域特征和灰度特征构建模糊速度函数,将其引入到活动轮廓模型中作为边缘检测项.在乳腺癌灶边界处,该速度函数趋向于零,活动轮廓曲线停止演变,完成对乳腺癌灶的分割.实验结果表明,所提出的方法有助于乳腺癌灶在DCE-MRI图像中的准确分割.展开更多
Objective： To prospectively compare the discriminative capacity of dynamic contrast enhanced-magnetic resonance imaging（DCE-MRI） with that of^18F-fluorodeoxyglucose（^18F-FDG） positron emission tomography/computed...Objective： To prospectively compare the discriminative capacity of dynamic contrast enhanced-magnetic resonance imaging（DCE-MRI） with that of^18F-fluorodeoxyglucose（^18F-FDG） positron emission tomography/computed tomography（PET/CT） in the differentiation of malignant and benign solitary pulmonary nodules（SPNs）.Methods： Forty-nine patients with SPNs were included in this prospective study. Thirty-two of the patients had malignant SPNs, while the other 17 had benign SPNs. All these patients underwent DCE-MRI and ^18F-FDG PET/CT examinations. The quantitative MRI pharmacokinetic parameters, including the trans-endothelial transfer constant（K^trans）, redistribution rate constant（Kep）, and fractional volume（Ve）, were calculated using the Extended-Tofts Linear two-compartment model. The ^18F-FDG PET/CT parameter, maximum standardized uptake value（SUV（max））, was also measured. Spearman＇s correlations were calculated between the MRI pharmacokinetic parameters and the SUV（max） of each SPN. These parameters were statistically compared between the malignant and benign nodules. Receiver operating characteristic（ROC） analyses were used to compare the diagnostic capability between the DCE-MRI and ^18F-FDG PET/CT indexes.Results： Positive correlations were found between K^trans and SUV（max）, and between K（ep） and SUV（max）（P〈0.05）.There were significant differences between the malignant and benign nodules in terms of the K^trans, K（ep） and SUV（max） values（P〈0.05）. The areas under the ROC curve（AUC） of K^trans） K（ep） and SUV（max） between the malignant and benign nodules were 0.909, 0.838 and 0.759, respectively. The sensitivity and specificity in differentiating malignant from benign SPNs were 90.6% and 82.4% for K^trans; 87.5% and 76.5% for K（ep）; and 75.0% and 70.6%for SUV（max）, respectively. The sensitivity and specificity of K^trans and K（ep） were higher than those of SUV（max）, but there was no significant differe展开更多
文摘乳腺癌灶的精确分割是乳腺癌计算机辅助诊断的重要前提.在动态对比增强核磁共振成像(Dynamic contrastenhanced magnetic resonance imaging,DCE-MRI)的图像中,乳腺癌灶具有对比度低、边界模糊及亮度不均匀等特点,传统的活动轮廓模型方法很难取得准确的分割结果.本文提出一种结合马尔科夫随机场(Markov random field,MRF)能量和模糊速度函数的活动轮廓模型的半自动分割方法来完成乳腺癌灶的分割,相对于专业医生的手动分割,本文方法具有速度快、可重复性高和分割结果相对客观等优点.首先,计算乳腺DCE-MRI图像的MRF能量,以增强目标区域与周围背景的差异.其次,在能量图中计算每个像素点的后验概率,建立基于后验概率驱动的活动轮廓模型区域项.最后,结合Gabor纹理特征、DCE-MRI时域特征和灰度特征构建模糊速度函数,将其引入到活动轮廓模型中作为边缘检测项.在乳腺癌灶边界处,该速度函数趋向于零,活动轮廓曲线停止演变,完成对乳腺癌灶的分割.实验结果表明,所提出的方法有助于乳腺癌灶在DCE-MRI图像中的准确分割.
基金supported by the Jiangsu Province Natural Science Foundation (No. BK20161291)the Nantong Science Foundation of China (No. MS2201507)the Nantong Municipal Commission of Health and Family Planning Young Fund (No. WQ2014047)
文摘Objective： To prospectively compare the discriminative capacity of dynamic contrast enhanced-magnetic resonance imaging（DCE-MRI） with that of^18F-fluorodeoxyglucose（^18F-FDG） positron emission tomography/computed tomography（PET/CT） in the differentiation of malignant and benign solitary pulmonary nodules（SPNs）.Methods： Forty-nine patients with SPNs were included in this prospective study. Thirty-two of the patients had malignant SPNs, while the other 17 had benign SPNs. All these patients underwent DCE-MRI and ^18F-FDG PET/CT examinations. The quantitative MRI pharmacokinetic parameters, including the trans-endothelial transfer constant（K^trans）, redistribution rate constant（Kep）, and fractional volume（Ve）, were calculated using the Extended-Tofts Linear two-compartment model. The ^18F-FDG PET/CT parameter, maximum standardized uptake value（SUV（max））, was also measured. Spearman＇s correlations were calculated between the MRI pharmacokinetic parameters and the SUV（max） of each SPN. These parameters were statistically compared between the malignant and benign nodules. Receiver operating characteristic（ROC） analyses were used to compare the diagnostic capability between the DCE-MRI and ^18F-FDG PET/CT indexes.Results： Positive correlations were found between K^trans and SUV（max）, and between K（ep） and SUV（max）（P〈0.05）.There were significant differences between the malignant and benign nodules in terms of the K^trans, K（ep） and SUV（max） values（P〈0.05）. The areas under the ROC curve（AUC） of K^trans） K（ep） and SUV（max） between the malignant and benign nodules were 0.909, 0.838 and 0.759, respectively. The sensitivity and specificity in differentiating malignant from benign SPNs were 90.6% and 82.4% for K^trans; 87.5% and 76.5% for K（ep）; and 75.0% and 70.6%for SUV（max）, respectively. The sensitivity and specificity of K^trans and K（ep） were higher than those of SUV（max）, but there was no significant differe
