目的基于1.5 T MRI,比较常规T2液体衰减反转恢复(T2-fluid attenuated inversion recovery,T2-FLAIR)序列传统重建和快速T2-FLAIR深度学习(deep learning,DL)重建在颅内占位性病变患者的图像质量与临床价值,并探究最佳DL重建参数。材料...目的基于1.5 T MRI,比较常规T2液体衰减反转恢复(T2-fluid attenuated inversion recovery,T2-FLAIR)序列传统重建和快速T2-FLAIR深度学习(deep learning,DL)重建在颅内占位性病变患者的图像质量与临床价值,并探究最佳DL重建参数。材料与方法前瞻性纳入颅内占位性病变患者104例,分别采集常规T2-FLAIR和快速T2-FLAIR[并行采集(parallel imaging,PI)加速因子2]。常规T2-FLAIR采用传统重建,记为NDL组;快速T2-FLAIR选择DL重建等级2、3和4,记为PI-DL2、PI-DL3和PI-DL4组。由两位医师采用盲法对四组图像进行定量评价和定性评价,并记录病变大小和数量。定量评价包括信噪比(signal-to-noise ratio,SNR)和对比噪声比(contrast-to-noise ratio,CNR);定性评价包括图像锐利度、噪声、灰白质对比度、伪影、病变显示、诊断信心和整体图像质量。结果常规T2-FLAIR扫描时间为2 min 8 s,快速T2-FLAIR扫描时间为1 min 20 s,时间缩短约37.5%。定量分析显示,与NDL组相比,各DL重建组(等级2、3、4)的SNR均有提高,且随DL等级增加而提高(P<0.05);PI-DL4组的CNR高于其他三组(P<0.05),而PI-DL2组在胼胝体压部、脑桥和小脑区域的CNR与NDL组差异无统计学意义(P>0.05)。定性评价方面,两位诊断医师评价一致性良好;PI-DL4组在图像锐利度、噪声控制和整体图像质量方面表现最佳(P<0.05);PI-DL4与PI-DL3组灰白质对比度、病变显示和诊断信心差异无统计学意义(P>0.05);PI-DL2组与NDL组在各项定性评价指标上差异均无统计学意义(P>0.05)。在病变检出方面,DL组检出率高于NDL组,病变大小测量差异无统计学意义(P>0.05)。结论在1.5 T MRI中,将DL重建算法与PI加速技术联合使用,可显著提高T2-FLAIR序列的图像质量和病变显示能力,并有效缩短扫描时间。由于DL等级4可能会导致部分病变边缘对比度降低,因此颅内占位性病变推荐使用DL等级3作为T2-FLAIR序列的最佳重建参数。展开更多
Objective:To retrospectively evaluate the diagnostic efficacy of traditional MRI and T2 Mapping quantitative imaging technology for knee joint cartilage injury,clarify the differences in diagnostic value of the two im...Objective:To retrospectively evaluate the diagnostic efficacy of traditional MRI and T2 Mapping quantitative imaging technology for knee joint cartilage injury,clarify the differences in diagnostic value of the two imaging methods in different injury grades and different cartilage subregions,and provide evidence-based basis for the accurate diagnosis of clinical cartilage injury.Methods:Clinical and imaging data of 286 patients with knee joint lesions admitted to the Affiliated Hospital of Xiangtan Medicine and Health Vocational College from January 2020 to June 2023 were collected retrospectively.All patients underwent both traditional MRI sequences and T2 Mapping sequences.The knee joint cartilage was divided into 14 subregions.Two senior radiologists independently diagnosed the images of the two imaging technologies using a blind method and recorded the cartilage injury grades.The sensitivity,specificity,accuracy,positive predictive value,negative predictive value,and area under the receiver operating characteristic curve(AUC)of the two technologies for diagnosing cartilage injury were calculated and compared,and the differences in their diagnostic efficacy in different injury grades and different subregions were analyzed.Results:A total of 4004 cartilage subregions from 286 patients were included in the analysis,including 1836 injured subregions and 2168 normal subregions.The overall sensitivity(89.7%),accuracy(91.2%),and AUC(0.946)of T2 Mapping quantitative imaging for diagnosing cartilage injury were significantly higher than those of traditional MRI(76.3%,82.5%,and 0.852 respectively),with statistically significant differences(p<0.001);there was no significant difference in specificity between the two(93.5%vs 90.8%,p=0.062).Subgroup analysis showed that T2 Mapping had the most significant diagnostic advantage in early cartilage injury(Grade 1),with sensitivity(78.5%)33.2%higher than that of traditional MRI(45.3%)(p<0.001).Conclusion:The diagnostic efficacy of T2 Mapping quantitative imaging for knee joint cartilage injury is significantly superior to that of traditional MRI,especially in the detection of early cartilage injury and accurate evaluation of weight-bearing area injury.Data verify its clinical applicability and reliability.It can be used as an important supplementary method to traditional MRI,and is recommended for the early diagnosis,grading evaluation,and clinical follow-up of cartilage injury.展开更多
文摘目的基于1.5 T MRI,比较常规T2液体衰减反转恢复(T2-fluid attenuated inversion recovery,T2-FLAIR)序列传统重建和快速T2-FLAIR深度学习(deep learning,DL)重建在颅内占位性病变患者的图像质量与临床价值,并探究最佳DL重建参数。材料与方法前瞻性纳入颅内占位性病变患者104例,分别采集常规T2-FLAIR和快速T2-FLAIR[并行采集(parallel imaging,PI)加速因子2]。常规T2-FLAIR采用传统重建,记为NDL组;快速T2-FLAIR选择DL重建等级2、3和4,记为PI-DL2、PI-DL3和PI-DL4组。由两位医师采用盲法对四组图像进行定量评价和定性评价,并记录病变大小和数量。定量评价包括信噪比(signal-to-noise ratio,SNR)和对比噪声比(contrast-to-noise ratio,CNR);定性评价包括图像锐利度、噪声、灰白质对比度、伪影、病变显示、诊断信心和整体图像质量。结果常规T2-FLAIR扫描时间为2 min 8 s,快速T2-FLAIR扫描时间为1 min 20 s,时间缩短约37.5%。定量分析显示,与NDL组相比,各DL重建组(等级2、3、4)的SNR均有提高,且随DL等级增加而提高(P<0.05);PI-DL4组的CNR高于其他三组(P<0.05),而PI-DL2组在胼胝体压部、脑桥和小脑区域的CNR与NDL组差异无统计学意义(P>0.05)。定性评价方面,两位诊断医师评价一致性良好;PI-DL4组在图像锐利度、噪声控制和整体图像质量方面表现最佳(P<0.05);PI-DL4与PI-DL3组灰白质对比度、病变显示和诊断信心差异无统计学意义(P>0.05);PI-DL2组与NDL组在各项定性评价指标上差异均无统计学意义(P>0.05)。在病变检出方面,DL组检出率高于NDL组,病变大小测量差异无统计学意义(P>0.05)。结论在1.5 T MRI中,将DL重建算法与PI加速技术联合使用,可显著提高T2-FLAIR序列的图像质量和病变显示能力,并有效缩短扫描时间。由于DL等级4可能会导致部分病变边缘对比度降低,因此颅内占位性病变推荐使用DL等级3作为T2-FLAIR序列的最佳重建参数。
基金Application Research of MRI Physiological Quantitative Imaging Technology in the Diagnosis of Cartilage Injury(Project No.:RCYJ2021-04)。
文摘Objective:To retrospectively evaluate the diagnostic efficacy of traditional MRI and T2 Mapping quantitative imaging technology for knee joint cartilage injury,clarify the differences in diagnostic value of the two imaging methods in different injury grades and different cartilage subregions,and provide evidence-based basis for the accurate diagnosis of clinical cartilage injury.Methods:Clinical and imaging data of 286 patients with knee joint lesions admitted to the Affiliated Hospital of Xiangtan Medicine and Health Vocational College from January 2020 to June 2023 were collected retrospectively.All patients underwent both traditional MRI sequences and T2 Mapping sequences.The knee joint cartilage was divided into 14 subregions.Two senior radiologists independently diagnosed the images of the two imaging technologies using a blind method and recorded the cartilage injury grades.The sensitivity,specificity,accuracy,positive predictive value,negative predictive value,and area under the receiver operating characteristic curve(AUC)of the two technologies for diagnosing cartilage injury were calculated and compared,and the differences in their diagnostic efficacy in different injury grades and different subregions were analyzed.Results:A total of 4004 cartilage subregions from 286 patients were included in the analysis,including 1836 injured subregions and 2168 normal subregions.The overall sensitivity(89.7%),accuracy(91.2%),and AUC(0.946)of T2 Mapping quantitative imaging for diagnosing cartilage injury were significantly higher than those of traditional MRI(76.3%,82.5%,and 0.852 respectively),with statistically significant differences(p<0.001);there was no significant difference in specificity between the two(93.5%vs 90.8%,p=0.062).Subgroup analysis showed that T2 Mapping had the most significant diagnostic advantage in early cartilage injury(Grade 1),with sensitivity(78.5%)33.2%higher than that of traditional MRI(45.3%)(p<0.001).Conclusion:The diagnostic efficacy of T2 Mapping quantitative imaging for knee joint cartilage injury is significantly superior to that of traditional MRI,especially in the detection of early cartilage injury and accurate evaluation of weight-bearing area injury.Data verify its clinical applicability and reliability.It can be used as an important supplementary method to traditional MRI,and is recommended for the early diagnosis,grading evaluation,and clinical follow-up of cartilage injury.
文摘目的探讨增强T2^(*)加权血管成像(enhanced T2 star-weighted angiography,ESWAN)序列中R2^(*)值、相位值、幅度值在T2WI低信号肾脏病变良恶性鉴别诊断中的可行性。材料与方法回顾性收集行ESWAN检查、经病理组织学证实的145例T2WI低信号肾脏病变患者(共145个病灶,恶性病变112个,良性病变33个)的术前MRI图像。在肿瘤最大面积的层面上绘制肿瘤T2WI低信号的感兴趣区。通过Kruskal-Wallis检验、卡方检验对参数进行比较,将有统计学意义的参数进行联合,通过多变量logistic回归建立模型,分析差异有统计学意义的参数,并且绘制其鉴别T2WI低信号肾脏病变良恶性的受试者工作特征(receiver operating characteristic,ROC)曲线,采用DeLong检验评价其诊断效能。结果R2^(*)值和幅度值鉴别T2WI低信号肾脏病变良恶性差异具有统计学意义(P=0.001)。R2^(*)值的ROC曲线下面积(area under the curve,AUC)为0.891[95%置信区间(confidence interval,CI):0.829~0.937,P<0.001],敏感度、特异度分别为97.3%、72.7%;幅度值的AUC为0.869(95%CI:0.803~0.920,P<0.001),敏感度、特异度分别为86.6%、81.8%;相位值的AUC为0.563(95%CI:0.478~0.645,P=0.249),敏感度、特异度分别为67.9%、54.6%;R2^(*)值联合幅度值的AUC为0.886(95%CI:0.823~0.933,P<0.001),敏感度、特异度分别为97.3%、72.7%;R2^(*)值联合病变长径的AUC为0.894(95%CI:0.832~0.939,P<0.001),敏感度、特异度分别为92.0%、81.8%;幅度值联合病变长径的AUC为0.858(95%CI:0.790~0.910,P<0.001),敏感度、特异度分别为75.9%、90.9%。结论R2^(*)值、R2^(*)值联合病变长径、R2^(*)值联合幅度值是鉴别T2WI低信号肾脏病变良恶性的有效方法,R2^(*)值联合病变长径具有更好的诊断性能。
