To denoise the diffusion weighted images （DWls） featured as multi-boundary, which was very important for the calculation of accurate DTIs （diffusion tensor magnetic resonance imaging）, a modified Wiener filter was...To denoise the diffusion weighted images （DWls） featured as multi-boundary, which was very important for the calculation of accurate DTIs （diffusion tensor magnetic resonance imaging）, a modified Wiener filter was proposed. Through analyzing the widely accepted adaptive Wiener filter in image denoising fields, which suffered from annoying noise around the edges of DWIs and in turn greatly affected the denoising effect of DWIs, a local-shift method capable of overcoming the defect of the adaptive Wiener filter was proposed to help better denoising DWIs and the modified Wiener filter was constructed accordingly. To verify the denoising effect of the proposed method, the modified Wiener filter and adaptive Wiener filter were performed on the noisy DWI data, respectively, and the results of different methods were analyzed in detail and put into comparison. The experimental data show that, with the modified Wiener method, more satisfactory results such as lower non-positive tensor percentage and lower mean square errors of the fractional anisotropy map and trace map are obtained than those with the adaptive Wiener method, which in turn helps to produce more accurate DTIs.展开更多
目的构建基于MRI扩散加权成像(diffusion-weighted imaging,DWI)的深度学习模型,讨论其对急性缺血性卒中静脉溶栓治疗患者90天预后的预测价值。材料与方法回顾性分析了2家医院进行静脉溶栓治疗的677名急性缺血性卒中(acute ischemic str...目的构建基于MRI扩散加权成像(diffusion-weighted imaging,DWI)的深度学习模型,讨论其对急性缺血性卒中静脉溶栓治疗患者90天预后的预测价值。材料与方法回顾性分析了2家医院进行静脉溶栓治疗的677名急性缺血性卒中(acute ischemic stroke,AIS)患者的临床及影像学资料,通过影像储存和传输系统(picture archiving and communication systems,PACS)收集患者MRI-DWI图像,使用深度神经网络提取患者图像特征。我们将数据集1(医院1)随机分为训练集(70%)和测试集(30%),建立基于临床特征(模型A)和MRI-DWI影像组学特征(模型B)的传统机器学习模型,基于MRI-DWI深度学习特征的深度学习模型(模型C)以及结合临床特征和深度学习特征的组合模型(模型D),预测AIS患者接受静脉溶栓治疗后90天预后[通过评估改良Rankin评分(modified Rankin Scale,mRS),评分<2分表示预后良好]。数据集2(医院2)用于外部验证。通过受试者工作特征(receiver operating characteristic,ROC)曲线及其曲线下面积(area under the curve,AUC)评估模型的预测性能。为了比较不同模型的AUC值差异是否有统计学意义,进一步采用DeLong检验进行统计分析,评估各模型之间AUC差异的显著性。结果基于临床特征和DWI-MRI影像组学特征的机器学习模型A和模型B以及深度学习模型C的AUC分别为0.705[95%置信区间(confidence interval,CI):0.613~0.792]、0.846(95%CI:0.777~0.906)和0.877(95%CI:0.811~0.934)。结合临床和深度学习特征的组合模型D在预测AIS患者静脉溶栓后90天预后方面表现出显著优势,其AUC值为0.930(95%CI:0.890~0.963)。此外,深度学习模型在外部验证数据集中同样显示出良好的性能,模型C和模型D的AUC分别为0.887(95%CI:0.798~0.960)和0.947(95%CI:0.891~0.984)。结论基于MRI-DWI的影像组学特征在预测接受静脉溶栓治疗的AIS患者的90天预后中发挥重要作用。深度学习方法在AIS溶栓治疗预后的预测模型中优于传统机器学习方法。结合临床特征和MRI-DWI特征的深度学习模型可为临床个性化评估AIS患者预后及制订治疗方案提供有力工具。展开更多
基金Project(2009AA04Z214) supported by the National High Technology Research and Development Program of ChinaProject(07JJ6133) supported by the Natural Science Foundation of Hunan Province, China
文摘To denoise the diffusion weighted images （DWls） featured as multi-boundary, which was very important for the calculation of accurate DTIs （diffusion tensor magnetic resonance imaging）, a modified Wiener filter was proposed. Through analyzing the widely accepted adaptive Wiener filter in image denoising fields, which suffered from annoying noise around the edges of DWIs and in turn greatly affected the denoising effect of DWIs, a local-shift method capable of overcoming the defect of the adaptive Wiener filter was proposed to help better denoising DWIs and the modified Wiener filter was constructed accordingly. To verify the denoising effect of the proposed method, the modified Wiener filter and adaptive Wiener filter were performed on the noisy DWI data, respectively, and the results of different methods were analyzed in detail and put into comparison. The experimental data show that, with the modified Wiener method, more satisfactory results such as lower non-positive tensor percentage and lower mean square errors of the fractional anisotropy map and trace map are obtained than those with the adaptive Wiener method, which in turn helps to produce more accurate DTIs.
文摘目的构建基于MRI扩散加权成像(diffusion-weighted imaging,DWI)的深度学习模型,讨论其对急性缺血性卒中静脉溶栓治疗患者90天预后的预测价值。材料与方法回顾性分析了2家医院进行静脉溶栓治疗的677名急性缺血性卒中(acute ischemic stroke,AIS)患者的临床及影像学资料,通过影像储存和传输系统(picture archiving and communication systems,PACS)收集患者MRI-DWI图像,使用深度神经网络提取患者图像特征。我们将数据集1(医院1)随机分为训练集(70%)和测试集(30%),建立基于临床特征(模型A)和MRI-DWI影像组学特征(模型B)的传统机器学习模型,基于MRI-DWI深度学习特征的深度学习模型(模型C)以及结合临床特征和深度学习特征的组合模型(模型D),预测AIS患者接受静脉溶栓治疗后90天预后[通过评估改良Rankin评分(modified Rankin Scale,mRS),评分<2分表示预后良好]。数据集2(医院2)用于外部验证。通过受试者工作特征(receiver operating characteristic,ROC)曲线及其曲线下面积(area under the curve,AUC)评估模型的预测性能。为了比较不同模型的AUC值差异是否有统计学意义,进一步采用DeLong检验进行统计分析,评估各模型之间AUC差异的显著性。结果基于临床特征和DWI-MRI影像组学特征的机器学习模型A和模型B以及深度学习模型C的AUC分别为0.705[95%置信区间(confidence interval,CI):0.613~0.792]、0.846(95%CI:0.777~0.906)和0.877(95%CI:0.811~0.934)。结合临床和深度学习特征的组合模型D在预测AIS患者静脉溶栓后90天预后方面表现出显著优势,其AUC值为0.930(95%CI:0.890~0.963)。此外,深度学习模型在外部验证数据集中同样显示出良好的性能,模型C和模型D的AUC分别为0.887(95%CI:0.798~0.960)和0.947(95%CI:0.891~0.984)。结论基于MRI-DWI的影像组学特征在预测接受静脉溶栓治疗的AIS患者的90天预后中发挥重要作用。深度学习方法在AIS溶栓治疗预后的预测模型中优于传统机器学习方法。结合临床特征和MRI-DWI特征的深度学习模型可为临床个性化评估AIS患者预后及制订治疗方案提供有力工具。
文摘目的旨在评估动态对比增强磁共振成像(dynamic contrast-enhancement magnetic resonance imaging,DCE-MRI)结合扩散加权成像(diffusion weighted imaging,DWI)在预测前列腺癌(prostate cancer,PCa)Ki-67表达和Gleason评分中的诊断效能。材料与方法回顾性分析了2019年1月至2023年10月自贡市第四人民医院收治的66例PCa患者的临床及影像资料。结合T2WI、DWI序列和由DWI自动计算出的表观扩散系数(apparent diffusion coeffieient,ADC),在DCE-MRI图像上手动勾画肿瘤感兴趣区(region of interest,ROI),计算ROI药代动力学参数,包括容积转运常数(volume transfer contrast,K^(trans))、速率常数(rate contrast,K_(ep))、血管外细胞外容积分数(extravascular extracellular volume fraction,Ve),并测量ADC值。根据靶向穿刺病理诊断Gleason评分和Ki-67表达水平,分为Ki-67高表达组(Ki-67>10%)和低表达组(Ki-67≤10%),Gleason评分低级别(GG 1~2)和高级别(GG 3~5)组。组间差异比较使用两独立样本t检验或非参数检验,采用Spearman相关分析评价DCE-MRI参数和ADC值与Ki-67、Gleason评分的相关性,并建立logistic回归模型,通过受试者工作特征(receiver operating characteristic,ROC)曲线评估诊断效能。结果ADC值与Ki-67表达、Gleason评分均呈负相关(P<0.001),K^(trans)、K_(ep)、Ve与Ki-67表达均呈正相关(P<0.001),K^(trans)、K_(ep)与Gleason评分均呈正相关(P<0.001)。Ki-67高、低表达组K^(trans)、K_(ep)、Ve、ADC值比较差异均具有统计学意义(P<0.01),Gleason评分高、低级别组K^(trans)、K_(ep)、ADC值比较差异均具有统计学意义(P<0.01);Ki-67表达的ROC曲线分析显示,联合模型K^(trans)+K_(ep)+Ve+ADC诊断效能最好,曲线下面积(area under the curve,AUC)为0.940;Gleason评分分级的ROC曲线分析显示,联合模型K^(trans)+K_(ep)+ADC诊断效能最好,AUC为0.861。结论DCE-MRI的药代动力学参数和ADC值相结合,在预测PCa的Ki-67表达和Gleason评分中显示出高诊断效能。联合使用DCE-MRI定量参数与ADC值可提高PCa病理分级和生物侵袭性的预测准确性。
文摘目的探讨基于动态对比增强磁共振成像(dynamic contrast-enhanced magnetic resonance imaging,DCE-MRI)和扩散加权成像(diffusion-weighted imaging,DWI)的瘤内及瘤周影像组学预测乳腺癌人表皮生长因子受体2(human epidermal growth factor receptor-2,HER-2)状态的价值。材料与方法回顾性分析246例经术后病理证实的乳腺癌患者的临床及影像学资料,按7∶3比例随机分为训练组和验证组。采用ITK-SNAP软件手动勾画病灶瘤内感兴趣区,使用PHIgo-AK软件进行瘤周的扩展并提取瘤内及瘤周的影像组学特征。采用最小冗余最大相关(max-relevance and min-redundancy,mRMR)算法等选择DCE-MRI、DWI瘤内及瘤周的最优特征数。分别建立单序列及联合序列的影像组学模型,采用受试者工作特征(receiveroperating characteristic,ROC)曲线对各模型的预测效能进行分析,并计算曲线下面积(area under the curve,AUC),选出预测效能最高的模型,在训练组中从临床及常规影像学特征中通过单因素logistic回归筛选出预测HER-2状态的独立危险因素,结合预测效能最高模型的影像组学标签评分(radiomic score,rad-score)建立融合模型,并以诺模图(nomogram)展示,采用AUC值,决策曲线分析(decision curve analysis,DCA)评估模型的效能及临床价值。结果基于DCE-MRI和DWI瘤内及瘤周的影像组学联合模型预测HER-2状态的AUC值在训练组和验证组分别为0.953和0.948,效能最高。肿瘤最大径是区分乳腺癌HER-2状态的独立危险因素,最终结合rad-score和肿瘤最大径建立的融合模型对乳腺癌HER-2状态有良好的预测效能,在训练组的AUC值为0.961,验证组为0.958。结论基于DCE-MRI和DWI瘤内及瘤周的影像组学方法对乳腺癌HER-2状态的预测具有良好的价值。
