分享:
分享到微信朋友圈
X
临床研究
弥散峰度成像联合常规MRI在甲状腺良恶性结节鉴别诊断中的价值
唐启瑛 刘信攸 姜秋利 朱柳红 周建军

Cite this article as: TANG Q Y, LIU X Y, JIANG Q L, et al. Value of diffusion kurtosis imaging combined with conventional MRI in the differential diagnosis of benign and malignant thyroid nodules[J]. Chin J Magn Reson Imaging, 2025, 16(5): 136-142.本文引用格式:唐启瑛, 刘信攸, 姜秋利, 等. 弥散峰度成像联合常规MRI在甲状腺良恶性结节鉴别诊断中的价值[J]. 磁共振成像, 2025, 16(5): 136-142. DOI:10.12015/issn.1674-8034.2025.05.021.


[摘要] 目的 探讨弥散峰度成像(diffusion kurtosis imaging, DKI)联合常规MRI在甲状腺良恶性结节鉴别诊断中的应用价值。材料与方法 选取经手术病理证实的甲状腺结节患者96例,其中39例为良性结节,57例为恶性结节。术前均行常规MRI序列和DKI检查,获得T1WI和T2WI的信号强度比值(signal intensity ratio, SIR)、表观弥散系数(apparent diffusion coefficient, ADC)、平均弥散系数(mean diffusivity, MD)和平均峰度系数(mean kurtosis, MK)及形态学特征。采用单因素和多因素logistic回归分析评估这些参数和特征作为甲状腺恶性结节潜在预测因子的价值,为验证logistic回归模型的稳健性,采用5折交叉验证方法,并绘制受试者工作特征(receiver operating characteristic, ROC)曲线来评价其鉴别良恶性结节的诊断效能。结果 单因素logistic分析提示恶性结节的特征包括ADC值减低(P<0.001)、MD值减低(P<0.001)、MK值增高(P<0.001)、年龄较小(P<0.001)、肿瘤直径较小(P<0.001)、实性成分更多见(P<0.001)和边缘不规则更多见(P<0.001)。在多因素分析中,MD值减低[比值比(odds ratio, OR)=5.046;P=0.001],直径较小(OR=3.817;P=0.001)和边缘不规则(OR=84.876;P<0.001)是甲状腺恶性结节的独立危险因素,5折交叉验证中,三者联合诊断恶性结节的平均ROC曲线下面积为0.968,最佳临界值为0.42,敏感度为91.4%,特异度为84.6%,准确率为88.7%,F1值为0.904。结论 DKI参数MD值联合常规MRI形态学特征可为术前鉴别诊断甲状腺良恶性结节提供影像学诊断依据。
[Abstract] Objective To assess the diagnostic value of unenhanced MRI with diffusion kurtosis imaging (DKI) in differential diagnosis between thyroid benign and malignant nodules.Materials and Methods A total of 96 consecutive patients, each with a single thyroid nodule, were included in this study. Among these, 39 nodules were histopathologically confirmed as benign, while 57 were malignant. All patients underwent thyroid MRI, which included T1-weighted imaging (T1WI), T2-weighted imaging (T2WI), diffusion-weighted imaging (DWI), and DKI. Two radiologists independently evaluated the images, measuring the signal intensity ratio (SIR) on T1WI and T2WI, the apparent diffusion coefficient (ADC) from DWI, as well as mean diffusivity (MD) and mean kurtosis (MK) from DKI. Additionally, morphological features of the nodules were assessed. Univariate and multivariate logistic regression analyses were performed to determine the predictive value of these imaging parameters and morphological features for malignancy. To assess the robustness of the logistic regression model, a 5-fold cross-validation approach was applied. Receiver operating characteristic (ROC) curve analysis was performed to evaluate the diagnostic effectiveness of the continuous variables that were statistically significant in the multivariate analysis.Results The characteristics of thyroid malignant nodules including lower ADC values (P < 0.001), lower MD values (P < 0.001), higher MK values (P < 0.001), younger age (P < 0.001), smaller tumor size (P < 0.001), solid component (P < 0.001), and irregular margins (P < 0.001). Multivariate analysis further revealed that lower MD values (odds ratio = 5.046; P = 0.001), smaller tumor size (odds ratio = 3.817; P = 0.001), and irregular margins (odds ratio = 84.876; P < 0.001) were independent risk factors for thyroid malignant nodules. The combined model yielded an average area under the ROC curve of 0.968 in 5-fold cross-validation, with a sensitivity of 91.4%, specificity of 84.6%, accuracy of 88.7%, and an F1 score of 0.904 at the optimal cutoff value of 0.42.Conclusions MD values derived from DKI, combined with morphological features can provide imaging diagnostic basis for the preoperative differential diagnosis between thyroid benign and malignant nodules.
[关键词] 甲状腺结节;甲状腺癌;磁共振成像;弥散峰度成像;鉴别诊断
[Keywords] thyroid neoplasms;thyroid carcinoma;magnetic resonance imaging;diffusion kurtosis imaging;differential diagnosis

唐启瑛 1, 2   刘信攸 3   姜秋利 4   朱柳红 1, 2   周建军 1, 2*  

1 复旦大学附属中山医院厦门医院放射诊断科,厦门 361015

2 厦门市影像医学临床医学研究中心,厦门 361015

3 复旦大学附属中山医院厦门医院普外科,厦门 361015

4 复旦大学附属中山医院厦门医院病理科,厦门 361015

通信作者:周建军,E-mail: zhoujianjunzs@126.com

作者贡献声明:周建军设计本研究的方案,对稿件重要内容进行了修改;唐启瑛起草和撰写稿件,实施研究,采集、分析和解释本研究的数据;刘信攸、姜秋利、朱柳红参与实施研究,分析和解释本研究的数据,并对稿件重要内容进行了修改;朱柳红获得了福建省自然科学基金项目资助,唐启瑛获得了福建省卫生健康科技计划项目资助;全体作者都同意发表最后的修改稿,同意对本研究的所有方面负责,确保本研究的准确性和诚信。


基金项目: 福建省自然科学基金项目 2022J011425 福建省卫生健康科技计划项目 2022QNB020
收稿日期:2025-02-23
接受日期:2025-05-09
中图分类号:R445.2  R581.1 
文献标识码:A
DOI: 10.12015/issn.1674-8034.2025.05.021
本文引用格式:唐启瑛, 刘信攸, 姜秋利, 等. 弥散峰度成像联合常规MRI在甲状腺良恶性结节鉴别诊断中的价值[J]. 磁共振成像, 2025, 16(5): 136-142. DOI:10.12015/issn.1674-8034.2025.05.021.

0 引言

       甲状腺结节是常见的内分泌系统疾病,最新数据显示,在国内外人群中的发病率达19%~68%[1, 2]。尽管国际各大甲状腺学会已经开发了基于超声的风险分层系统,以最大限度地提高甲状腺超声的诊断性能,但仍存在一些局限性,如诊断甲状腺癌的敏感度高,但特异度较低[3, 4, 5]。此外,一项研究表明,在美国甲状腺学会(American Thyroid Association, ATA)、美国放射学会(American College of Radiology, ACR)、欧洲甲状腺学会(European Thyroid Association, ETA)以及韩国甲状腺学会(Korean Thyroid Association, KTA)四种具有代表性的超声风险分层系统指导下,仍有25%~55%的患者因超声诊断为甲状腺癌而接受了穿刺活检,病理证实为良性结节,从而造成了大量不必要的穿刺[6]。因此,如何通过无创的影像学检查进一步提高甲状腺癌诊断的准确性,减少不必要的穿刺活检,成为临床亟待解决的关键问题。

       MRI拥有良好的软组织分辨力,不仅可以客观地显示病灶的形态、信号特点,还能提供大量微观的功能信息,有望提高甲状腺结节诊断的准确性[7]。根据最新ACR指南,超声甲状腺影像报告与数据系统分类(Thyroid Imaging Reporting and Data System category, TIRADS)4类的甲状腺结节恶性风险约为5%~20%,临床处理需综合考虑结节的大小、位置、患者的意愿和焦虑程度等决定是否穿刺[8],而SONG等[9]的研究显示,基于弥散受限和延迟期反晕征等MRI特征的诺模图模型可显著提高ACR-TIRADS 4类结节的诊断准确性,从而减少不必要的穿刺。弥散峰度成像(diffusion kurtosis imaging, DKI)以非高斯分布水分子弥散模型为基础,可以提供比弥散加权成像(diffusion weighted imaging, DWI)更多的组织结构信息,如组织微结构的复杂性等[10, 11, 12],在全身各器官肿瘤的良恶性鉴别中表现出较大的优势[13, 14, 15]。目前DKI在甲状腺结节中的研究较少,SHI等[16]首次提出DKI与传统DWI相比,在甲状腺病变的诊断方面更具优势,并进行了影像参数与组织病理学特点的对比,认为基于DKI的D值与甲状腺良恶性结节的细胞外结构差异相关性较好,而K值更多地反映细胞内结构的改变,但其样本量相对较小。JIANG等[17]的研究也比较了DKI与体素内不相干运动成像(intravoxel incoherent motion imaging, IVIM)对甲状腺良恶性结节鉴别诊断价值,认为两者的诊断效能相当,然而既往研究尚未探讨过DKI与常规MRI联合应用于甲状腺结节诊断的意义。因此本文旨在通过DKI与常规MRI建立联合诊断模型,探究其在甲状腺良恶性结节鉴别诊断中的应用价值,提高甲状腺癌的无创性诊断准确性,为甲状腺结节的临床诊疗决策提供更多依据。

1 材料与方法

1.1 研究对象

       本研究遵守《赫尔辛基宣言》,经复旦大学附属中山医院厦门医院伦理委员会批准,全体受试者均签署了知情同意书,批准文号:B-2020-002。

       收集自2020年6月至2024年9月复旦大学附属中山医院厦门医院共117名超声诊断为甲状腺结节的患者。纳入标准:(1)超声诊断甲状腺结节,且未接受任何治疗;(2)既往无恶性肿瘤病史;(3)结节直径≥10 mm。排除标准:(1)存在MRI检查禁忌证,如患有严重幽闭恐惧症、体内存在非MRI兼容的金属植入物等;(2)未在本院进行手术治疗;(3)MRI检查与手术间隔时间≥2周;(4)由于严重的运动伪影或磁敏感伪影导致DKI图像无法评估;(5)结节几乎为纯囊性,即T1WI及T2WI呈均匀水样信号,未见明显软组织信号。

1.2 MRI检查方案

       所有受试者MRI数据的采集采用3.0 T MR扫描仪(美国GE,Discovery MR 750w),8通道头颈联合线圈(美国GE,8CH Head/Neck Coil)或32通道甲状腺专用表面线圈(苏州众志,TL320)。患者取仰卧位,双肩尽量下垂,嘱患者平静呼吸,避免吞咽及咳嗽动作。患者均行常规平扫、DWI及DKI扫描。具体参数如下:

       (1)横断位T1WI IDEAL快速自旋回波序列:TR 500 ms,TE 9.6 ms,层厚4 mm,层间距1 mm,FOV 22 cm×22 cm,矩阵256×224,NEX 2;(2)横断位T2WI IDEAL快速自旋回波序列:TR 2760 ms,TE 85 ms,层厚4 mm,层间距1 mm,FOV 22 cm×22 cm,矩阵288×192,NEX 2;(3)DWI序列:TR 4000 ms,TE 67.4 ms,层厚4 mm,层间距1 mm,FOV 22 cm×22 cm,矩阵128×128,NEX 4,b=0、500 s/mm2;(4)DKI序列:TR 3100 ms,TE 115.9 ms,层厚4 mm,层间距1 mm,FOV 22 cm×13 cm,矩阵140×84,NEX 4,b=0、1000、2000 s/mm2,采用张量法,每个b值均施加15个方向的弥散敏感梯度场。

1.3 图像分析

       将扫描所得的DWI及DKI序列图像导入美国GE Healthcare AW4.6工作站,采用工作站自带的Functool软件后处理生成对应的表观弥散系数(apparent diffusion coefficient, ADC)、平均弥散系数(mean diffusivity, MD)、平均峰度系数(mean kurtosis, MK)参数图像。在不知晓病理结果的情况下,由1名具有25年影像诊断经验的放射科主任医师A和1名具有10年影像诊断经验的放射科主治医师B采用双盲法阅片,分别测量病变的ADC、MD及MK值,测量各参数值时,首先结合T1WI、T2WI图像,选取结节最大层面,避开内部出血、坏死、囊变区等,对整个结节实性区域勾画感兴趣区(region of interest, ROI),每个结节测量3次并取其平均值。具体勾画过程示意图见图1

       采用同样的方法测量病灶的T1WI及T2WI信号强度比值(signal intensity ratio, SIR),SIR定义为病灶信号强度与脊柱旁肌肉信号强度的比值。形态学特点包括结节的大小、位置、成分、边缘是否规则及纵横比亦被评估。其中成分包括实性和囊实性,纯囊性结节因无法进行测量,所以不在本研究范围内。纵横比定义为在结节最大的横断面图像上,结节的前后径与左右径的比值。本研究对两位医师的测量结果进行一致性检验,若一致性好,则最终结果部分采用主任医师A的测量数据。

图1  感兴趣区(ROI)勾画示意图。1A~1B:结节最大层面T2加权成像及T1加权成像测量时ROI勾画示意图,避开内部出血、坏死、囊变区等,对整个结节实性区域勾画ROI,紫色线代表ROI,每个结节测量3次并取其平均值,同时测量脊柱旁肌肉信号强度,信号强度比值(SIR)定义为病灶的信号强度与脊柱旁肌肉的信号强度比值;1C~1F:同一层面结节选定区域的弥散加权成像(DWI)、表观弥散系数(ADC)图、平均弥散系数(MD)图、平均峰度系数(MK)图。1E~1F左侧色条从蓝色到红色,对应MD或MK值的依次升高。
Fig. 1  Schematic diagram of region of interest (ROI) delineation. 1A-1B: Schematic diagrams of ROI delineation on T2WI and T1WI at the maximum section of the nodule. ROIs are manually drawn over the solid component of the entire nodule while carefully excluding regions with intralesional hemorrhage, necrosis, or cystic change. The purple line represents ROI. Each nodule is measured three times, with the average value taken. The signal intensity of the paraspinal muscles is also measured. The signal intensity ratio (SIR) is defined as the ratio of the signal intensity of the lesion to that of the paraspinal muscles. 1C-1F: Diffusion weighted imaging (DWI), apparent diffusion coefficient (ADC) map, mean diffusivity (MD) map and mean kurtosis (MK) map of the selected nodule regions at the same level. In the left legend of figures 1E-1F, the color scale from blue to red corresponds to an increase in MD or MK values.

1.4 统计学分析

       采用SPSS(美国IBM,v22.0)及R语言(奥地利,R Foundation,v3.6.0)软件进行统计分析。对连续变量采用t检验,对分类变量采用卡方检验或Fisher精确检验。筛选出P<0.05的因素进行多因素logistic回归分析,得出诊断甲状腺恶性结节的独立危险因素,计算比值比(odds ratio, OR)、95%置信区间(confidence interval, CI)和预测概率值。为验证logistic回归模型的稳健性,采用5折交叉验证方法,将所有样本随机分为5个相等的子集,每次以其中1个子集为测试集,其余为训练集,重复5次并计算平均预测性能指标,包括受试者工作特征(receiver operating characteristic, ROC)曲线的曲线下面积(area under the curve, AUC)、敏感度、特异度、准确率和F1值。采用组内相关系数(intra-class correlation coefficient, ICC)评价2名阅片者测量参数的组间一致性及每位阅片者3次测量参数的组内一致性,评价标准如下:<0.40,较差;0.40~<0.60,一般;0.60~<0.75,好;0.75~1.00,很好。P<0.05为差异有统计学意义。

2 结果

2.1 一般资料

       本研究因患者未在本院进行手术治疗排除6例,因MRI检查与手术间隔时间≥2周排除2例,因严重的运动伪影或磁敏感伪影排除8例,因结节几乎为纯囊性排除5例,最终共纳入96例患者,其中男25例,女71例,年龄19~76(45±12)岁;共收集甲状腺结节96例,经手术病理证实良性结节39例,其中滤泡性腺瘤28例,滤泡增生结节7例,结节性甲状腺肿4例,恶性结节57例,均为乳头状癌。临床和MRI特征见表1

表1  甲状腺良恶性结节的临床及MRI特点
Tab. 1  Clinical and MRI characteristics of benign and malignant thyroid nodules

2.2 形态学结果

       与良性结节组相比,恶性结节组患者年龄更小(P<0.001),肿瘤直径更小(P<0.001),实性成分更多见(P<0.001),边缘不规则更多见(P<0.001)。良性结节组和恶性结节组在性别(P=0.069)、位置(P=0.734)和纵横比(P=0.885)上差异无统计学意义。

2.3 定量测量结果

       所有结节的T1WI SIR、T2WI SIR、ADC、MD、MK值见表2。恶性结节组的ADC、MD值低于良性结节组,差异具有统计学意义(P<0.001),MK值高于良性结节组,差异具有统计学意义(P<0.001)(图2图3)。ADC、MD和MK诊断恶性结节的AUC分别为0.848(95% CI:0.760~0.913)、0.886(95% CI:0.805~0.942)和0.799(95% CI:0.705~0.874)。

       ADC、MD和MK鉴别良恶性结节的最佳临界值分别为1.60×10-3 mm2/s(敏感度77.2%,特异度79.5%)、1.99×10-3 mm2/s(敏感度79.0%,特异度89.7%)和0.76(敏感度68.4%,特异度90.0%)。良性结节组和恶性结节组的T1WI SIR(P=0.422)、T2WI SIR(P=0.599)差异无统计学意义。2名医师测量良、恶性结节组T1WI、T2WI的SIR值、ADC、MD及MK值的组间ICC值分别为0.926和0.936、0.974和0.971、0.910和0.929、0.914和0.927、0.912和0.926,均具有很好的一致性,2名医师3次测量值的组内ICC值均>0.75(表3)。

图2  73岁男性患者,甲状腺乳头状癌。2A:横断位T2WI显示直径17 mm的实性结节,局部边缘不规则,纵横比>1,信号强度比(SIR)为3.18;2B:横断位T1WI,SIR为1.54;2C:表观弥散系数(ADC)图,ADC值=2.18×10-3 mm2/s;2D:平均弥散系数(MD)图,MD值=1.90×10-3 mm2/s;2E:平均峰度系数(MK)图,MK值=0.81;2F:组织病理学苏木精-伊红(H&E)染色(×40)。
Fig. 2  A 73-year-old male with papillary thyroid carcinoma (arrow) in the right thyroid lobe. 2A: Axial T2-weighted imaging shows a 17 mm solid nodule with locally irregular margin and a positive taller-than-wide sign: the signal intensity ratio (SIR) is 3.18. 2B: Axial T1-weighted imaging: the SIR is 1.54. 2C: Apparent diffusion coefficient (ADC) map; ADC value = 2.18×10−3 mm2/s. 2D: Mean diffusivity (MD) map; MD value = 1.90×10-3 mm2/s. 2E: Mean kurtosis (MK) map; MK value=0.81. 2F: Histopathological hematoxylin and eosin staining (× 40).
图3  44岁女性患者,甲状腺腺瘤。3A:横断位T2加权成像显示直径28 mm的实性结节,边缘规则,纵横比<1,信号强度比(SIR)为4.58;3B:横断位T1加权成像,SIR为1.33;3C:表观弥散系数(ADC)图,ADC值=2.78×10-3 mm2/s;3D:平均弥散系数(MD)图,MD值=3.05×10-3 mm2/s;3E:平均峰度系数(MK)图,MK值=0.39;3F:组织病理学苏木精-伊红(H&E)染色(×40)。
Fig. 3  A 44-year-old female with thyroid adenoma (arrow) in the right thyroid lobe. 3A: Axial T2-weighted imaging shows a 28 mm solid nodule with a regular margin and a negative taller-than-wide sign: the signal intensity ratio (SIR) is 4.58. 3B: Axial T1-weighted imaging: the SIR is 1.33. 3C: Apparent diffusion coefficient (ADC) map; ADC value = 2.78×10−3 mm2/s. 3D: Mean diffusivity (MD) map; MD value = 3.05×10-3 mm2/s. 3E: Mean kurtosis (MK) map; MK value = 0.39. 3F: Histopathological hematoxylin and eosin staining (× 40).
表2  甲状腺良恶性结节的常规MRI及DKI定量参数
Tab. 2  Conventional MRI and DKI quantitative measurements of benign and malignant thyroid nodules
表3  两位医师测量数据的组内及组间一致性检验结果
Tab. 3  Intra- and inter- group consistency for the measurements of the two radiologists

2.4 多因素logistic分析及诊断效能

       在上述分析结果中筛选出P<0.05的因素,包括年龄、肿瘤最大径、成分、边缘、ADC值、MD值、MK值,纳入多因素logistic回归发现,MD值减低(OR=5.046;95% CI:2.008~25.261;P=0.001),直径较小(OR=3.817;95% CI:1.728~19.916;P=0.001)和边缘不规则(OR=84.876;95% CI:7.527~157.045;P<0.001)是甲状腺恶性结节的独立危险因素(表4)。由logistic回归分析生成MD值低、直径小和边缘不规则的联合预测因子。为验证模型的稳健性,采用5折交叉验证对logistic回归模型进行内部验证,交叉验证的平均F1值为0.904,证明该模型在阳性病例的识别中兼具较高的准确性和召回能力,具备较好的临床实用性(表5)。交叉验证的平均AUC为0.968(95% CI:0.947~0.987),最佳临界值为0.42,敏感度为91.4%,特异度为84.6%,准确率为88.7%(图4)。

图4  平均弥散系数(MD)值减低、直径较小和边缘不规则三者联合诊断甲状腺恶性结节的受试者工作特征曲线,5折交叉验证的平均曲线下面积为0.968,最佳临界值为0.42,敏感度为91.4%,特异度为84.6%,准确率为88.7%,F1值为0.904。
Fig. 4  Receiver operating characteristic (ROC) curve of the diagnostic performance of the combined predictor of mean diffusivity (MD), diameter, and margin in identifying malignant thyroid nodules. The combined model yielded an average area under the ROC curve of 0.968 in 5-fold cross-validation, with a sensitivity of 91.4%, specificity of 84.6%, accuracy of 88.7%, and an F1 score of 0.904 at the optimal cutoff value of 0.42.
表4  多因素logistic回归分析结果
Tab. 4  Multivariate logistic regression analysis of variables
表5  logistic回归模型5折交叉验证的性能指标
Tab. 5  Performance metrics of the logistic regression model in 5-fold cross-validation

3 讨论

       本研究采用DKI参数联合常规MRI形态学特征对甲状腺结节进行术前评估,发现基于MD值减低、直径较小和边缘不规则的多因素模型可以很好地预测甲状腺结节恶性风险,5折交叉验证中,三者联合诊断恶性结节的平均ROC曲线下面积为0.968,最佳临界值为0.42,敏感度为91.4%,特异度为84.6%,准确率为88.7%,F1值为0.904。

3.1 DKI定量参数对鉴别甲状腺良恶性结节的价值

       传统DWI假设水分子在随机运动情况下满足高斯分布,然而在活体内水分子的弥散受到局部组织结构和细胞形态的影响,其运动必然是非高斯分布的。DKI则是基于体内水分子运动符合非高斯分布的模型,通过拟合计算出的MD和MK两个主要参数,MD值能更真实地反映水分子的弥散水平,MK值则反映了水分子弥散偏离高斯分布的程度,从而间接反映组织结构的复杂性[11, 18, 19]。本研究结果显示,甲状腺恶性结节的MD值低于良性结节,与以往研究结果一致[16]。主要原因是甲状腺癌的细胞密度比良性结节高,尤其是乳头状癌含有纤维、钙化的砂粒体和不规则的乳头状结构,均导致水分子的弥散受限,即MD值的减低[20]。SHI等[16]的研究中甲状腺恶性结节具有较高的MK值,反映了较高程度的细胞微观结构异质性。然而,本研究多因素logistic回归分析结果并未表明MK是恶性结节的独立预测因子,尽管在单因素分析中,MK在恶性结节中明显高于良性结节。造成这种情况的原因可能是,除了统计方法不同之外,本研究中所采用的DKI扫描方案是基于张量的方法,而SHI等[16]使用了多b值法。这两种不同类型的DKI扫描方案在以往研究中均有应用[21, 22]。张量法的优势在于采用了至少15个弥散方向,可更好地评估弥散的各向异性和峰度特点,同时对运动伪影更不敏感。ADC值在多因素分析中亦未显示独立预测价值,说明经过高斯分布校正后的MD值比ADC值更能体现甲状腺良恶性结节在微观结构上的差异[16]

3.2 边缘不规则和结节大小对鉴别甲状腺良恶性结节的价值

       在本研究中,边缘不规则被证明是诊断甲状腺恶性结节的可靠影像学征象,这与以往的研究一致[23, 24, 25]。边缘不规则的组织病理学机制可能与恶性结节的侵袭性和不同方向上的异质性生长模式有关[26]。本研究结果显示,甲状腺恶性结节的直径比良性结节更小,这也与以往的研究一致。如CAVALLO等[27]的研究显示,甲状腺结节的大小与恶性风险呈负相关,<1 cm的结节中57%为恶性结节,而>6 cm的结节中恶性结节仅占20%。SHAYGANFAR等[28]发现,在超声评价TIRADS 4分和5分的甲状腺结节中,直径小于12 mm的结节需高度提示恶性,建议进行细针穿刺抽吸活检。

3.3 T1WI、T2WI信号及纵横比对诊断甲状腺结节的局限性

       本研究还比较了甲状腺良恶性结节的T1WI及T2WI信号特点,二者的SIR值差异无统计学意义。这与NODA等[29]的研究中认为甲状腺乳头状癌T2WI SIR值低于良性结节的结果有所不同。笔者在收集的病例中发现,部分甲状腺乳头状癌T2WI信号较高,镜下表现为疏松的乳头状结构,并伴有间质水肿,而良性结节中亦存在T2WI低信号,镜下表现为滤泡排列紧密、钙化和纤维组织增生等。因此,笔者认为甲状腺结节的T2WI信号与结节的成分和结构有关,单纯根据T1WI和T2WI信号鉴别甲状腺结节的良恶性存在一定困难。有趣的是,纵横比>1是鉴别甲状腺良恶性最重要的超声特征之一[30],但在本研究中,此征象在甲状腺良恶性结节中并无显著差异。纵横比最早由KIM等[31]提出,其组织病理学机制推测为良性结节倾向于在甲状腺组织中横向生长,而恶性结节倾向于在甲状腺组织中纵向生长,因此纵径大于横径。然而,YOON等[32]认为,当使用超声探头进行压迫时,相比于良性结节,恶性结节由于质地更硬,被压扁的程度更小,即纵径相对更大,因此纵横比>1。与超声相比,MRI检查时甲状腺组织没有受到压迫,这或许可以解释本研究的结果。

3.4 本研究的局限性

       首先,本研究样本量较小,且缺乏外部验证,可能影响研究结果在不同环境和人群中的稳定性和检测效能,未来还需大样本、多中心研究证实该模型的可靠性。其次,由于颈部DKI容易受到磁敏感伪影的影响,且病灶越小越难以显示,因此本研究未纳入<10 mm的甲状腺结节,但根据最新国内外甲状腺结节管理指南[33, 34],最大径≤10 mm的甲状腺结节,对于影像检查发现甲状腺结节有恶性征象但未发现明确的颈侧区淋巴结转移的患者,可在医生充分告知的情况下进行密切随访。未来随着MRI技术的更新和人工智能的辅助,甲状腺MRI的图像质量将更加完善,MRI对甲状腺结节的诊断、分期及侵袭性评估的潜力将日益突显。

4 结论

       综上所述,DKI参数MD值联合常规MRI形态学特征对鉴别诊断甲状腺良恶性结节具有一定价值,有望提高甲状腺癌的无创诊断准确性,为甲状腺结节的临床诊疗决策提供重要依据。

[1]
BOUCAI L, ZAFEREO M, CABANILLAS M E. Thyroid cancer: a review[J]. JAMA, 2024, 331(5): 425-435. DOI: 10.1001/jama.2023.26348.
[2]
LI Y H, JIN C, LI J, et al. Prevalence of thyroid nodules in China: a health examination cohort-based study[J/OL]. Front Endocrinol (Lausanne), 2021, 12: 676144 [2025-02-22]. https://pubmed.ncbi.nlm.nih.gov/34122350/. DOI: 10.3389/fendo.2021.676144.
[3]
ÖZDEMIR M, TÜRK G, BILGILI M, et al. Comparison of diagnostic performances of ATA guidelines, ACR-TIRADS, and EU-TIRADS and modified K-TIRADS: a single center study of 4238 thyroid nodules[J]. Exp Clin Endocrinol Diabetes, 2025, 133(2): 98-104. DOI: 10.1055/a-2498-7952.
[4]
SI C F, YU J, CUI Y Y, et al. Comparison of diagnostic performance of the current score-based ultrasound risk stratification systems according to thyroid nodule size[J]. Quant Imaging Med Surg, 2024, 14(12): 9234-9245. DOI: 10.21037/qims-24-282.
[5]
WATKINS L, O'NEILL G, YOUNG D, et al. Comparison of British thyroid association, American college of radiology TIRADS and artificial intelligence TIRADS with histological correlation: diagnostic performance for predicting thyroid malignancy and unnecessary fine needle aspiration rate[J/OL]. Br J Radiol, 2021, 94(1123): 20201444 [2025-02-22]. https://pubmed.ncbi.nlm.nih.gov/33989038/. DOI: 10.1259/bjr.20201444.
[6]
KIM P H, SUH C H, BAEK J H, et al. Unnecessary thyroid nodule biopsy rates under four ultrasound risk stratification systems: a systematic review and meta-analysis[J]. Eur Radiol, 2021, 31(5): 2877-2885. DOI: 10.1007/s00330-020-07384-6.
[7]
MARTÍN-NOGUEROL T, SANTOS-ARMENTIA E, FERNANDEZ-PALOMINO J, et al. Role of advanced MRI sequences for thyroid lesions assessment. A narrative review[J/OL]. Eur J Radiol, 2024, 176: 111499 [2025-02-22]. https://pubmed.ncbi.nlm.nih.gov/38735157/. DOI: 10.1016/j.ejrad.2024.111499.
[8]
TESSLER F N, MIDDLETON W D, GRANT E G, et al. ACR thyroid imaging, reporting and data system (TI-RADS): white paper of the ACR TI-RADS committee[J]. J Am Coll Radiol, 2017, 14(5): 587-595. DOI: 10.1016/j.jacr.2017.01.046.
[9]
SONG B, CHEN Q H, WANG H, et al. Enhancing diagnostic accuracy of American college of radiology TI-RADS 4 nodules: nomogram models based on MRI morphological features[J]. Quant Imaging Med Surg, 2025, 15(2): 1679-1693. DOI: 10.21037/qims-24-1427.
[10]
CUI Y Y, WANG X H, WANG Y, et al. Restriction spectrum imaging and diffusion kurtosis imaging for assessing proliferation status in rectal carcinoma[J]. Acad Radiol, 2025, 32(1): 201-209. DOI: 10.1016/j.acra.2024.08.021.
[11]
JENSEN J H, HELPERN J A. MRI quantification of non-Gaussian water diffusion by kurtosis analysis[J]. NMR Biomed, 2010, 23(7): 698-710. DOI: 10.1002/nbm.1518.
[12]
ZHANG N G, ZHAO W, SHANG S A, et al. Diffusion kurtosis imaging in diagnosing Parkinson's disease: a preliminary comparison study between kurtosis metric and radiomic features[J]. Acad Radiol, 2025, 32(2): 922-929. DOI: 10.1016/j.acra.2024.07.001.
[13]
CHEN Y, HUANG N, ZHENG Y Y, et al. Characterization of parotid gland tumors: whole-tumor histogram analysis of diffusion weighted imaging, diffusion kurtosis imaging, and intravoxel incoherent motion-A pilot study[J/OL]. Eur J Radiol, 2024, 170: 111199 [2025-02-22]. https://pubmed.ncbi.nlm.nih.gov/38104494/. DOI: 10.1016/j.ejrad.2023.111199.
[14]
KAUR G, MANCHANDA S, SHARMA R, et al. Comparison of conventional diffusion-weighted imaging, diffusion kurtosis imaging and intravoxel incoherent motion in characterization of sonographically indeterminate adnexal masses[J]. Abdom Radiol (NY), 2024, 49(5): 1512-1521. DOI: 10.1007/s00261-024-04292-x.
[15]
刘诗晗, 邵硕, 魏坤杰, 等. 动态对比增强MRI联合扩散峰度成像对预测三阴性乳腺癌的诊断价值[J]. 磁共振成像, 2023, 14(5): 110-115. DOI: 10.12015/issn.1674-8034.2023.05.020.
LIU S H, SHAO S, WEI K J, et al. Diagnostic value of DCE-MRI combined with DKI in predicting the triple negative breast cancer[J]. Chin J Magn Reson Imag, 2023, 14(5): 110-115. DOI: 10.12015/issn.1674-8034.2023.05.020.
[16]
SHI R Y, YAO Q Y, ZHOU Q Y, et al. Preliminary study of diffusion kurtosis imaging in thyroid nodules and its histopathologic correlation[J]. Eur Radiol, 2017, 27(11): 4710-4720. DOI: 10.1007/s00330-017-4874-0.
[17]
JIANG L L, CHEN J, HUANG H P, et al. Comparison of the differential diagnostic performance of intravoxel incoherent motion imaging and diffusion kurtosis imaging in malignant and benign thyroid nodules[J/OL]. Front Oncol, 2022, 12: 895972 [2025-02-22]. https://pubmed.ncbi.nlm.nih.gov/35936691/. DOI: 10.3389/fonc.2022.895972.
[18]
HUANG J, DUAN Z Q, CHENG Y, et al. Advanced diffusion-weighted imaging-derived quantitative parameters as biomarkers of fibrosarcoma-cell proliferation in nude mice: a study based on precise imaging-pathology correlation[J/OL]. Magn Reson Imaging, 2025, 118: 110345 [2025-02-22]. https://pubmed.ncbi.nlm.nih.gov/39892483/. DOI: 10.1016/j.mri.2025.110345.
[19]
SONG Q L, DONG W, TIAN S F, et al. Diffusion kurtosis imaging with multiple quantitative parameters for predicting microsatellite instability status of endometrial carcinoma[J]. Abdom Radiol (NY), 2023, 48(12): 3746-3756. DOI: 10.1007/s00261-023-04041-6.
[20]
LE B D. Apparent diffusion coefficient and beyond: what diffusion MR imaging can tell us about tissue structure[J]. Radiology, 2013, 268(2): 318-322. DOI: 10.1148/radiol.13130420.
[21]
CHO E, BAEK H J, SZCZEPANKIEWICZ F, et al. Clinical experience of tensor-valued diffusion encoding for microstructure imaging by diffusional variance decomposition in patients with breast cancer[J]. Quant Imaging Med Surg, 2022, 12(3): 2002-2017. DOI: 10.21037/qims-21-870.
[22]
WANG F, CHEN H G, ZHANG R Y, et al. Diffusion kurtosis imaging to assess correlations with clinicopathologic factors for bladder cancer: a comparison between the multi-b value method and the tensor method[J]. Eur Radiol, 2019, 29(8): 4447-4455. DOI: 10.1007/s00330-018-5977-y.
[23]
LIU Y C, LIU H, ZHAN J, et al. Contrast-enhanced ultrasound for diagnosing thyroid nodules with indeterminate cytology: a retrospective study[J]. Clin Endocrinol, 2025, 102(2): 223-231. DOI: 10.1111/cen.15160.
[24]
SHARAFELDEEN A, ELSHARKAWY M, KHALED R, et al. Texture and shape analysis of diffusion-weighted imaging for thyroid nodules classification using machine learning[J]. Med Phys, 2022, 49(2): 988-999. DOI: 10.1002/mp.15399.
[25]
ZHANG G L, YANG Y P, ZHOU H L, et al. Diagnostic efficacy of CEUS TI-RADS classification for benign and malignant thyroid nodules[J]. Clin Hemorheol Microcirc, 2025, 89(1): 27-41. DOI: 10.3233/CH-232080.
[26]
HAUGEN B R, ALEXANDER E K, BIBLE K C, et al. 2015 American thyroid association management guidelines for adult patients with thyroid nodules and differentiated thyroid cancer: the American thyroid association guidelines task force on thyroid nodules and differentiated thyroid cancer[J]. Thyroid, 2016, 26(1): 1-133. DOI: 10.1089/thy.2015.0020.
[27]
CAVALLO A, JOHNSON D N, WHITE M G, et al. Thyroid nodule size at ultrasound as a predictor of malignancy and final pathologic size[J]. Thyroid, 2017, 27(5): 641-650. DOI: 10.1089/thy.2016.0336.
[28]
SHAYGANFAR A, HASHEMI P, ESFAHANI M M, et al. Prediction of thyroid nodule malignancy using thyroid imaging reporting and data system (TIRADS) and nodule size[J]. Clin Imag, 2020, 60(2): 222-227. DOI: 10.1016/j.clinimag.2019.10.004.
[29]
NODA Y, KANEMATSU M, GOSHIMA S, et al. MRI of the thyroid for differential diagnosis of benign thyroid nodules and papillary carcinomas[J/OL]. AJR Am J Roentgenol, 2015, 204(3): W332-W335 [2025-02-22]. https://pubmed.ncbi.nlm.nih.gov/25714319/. DOI: 10.2214/AJR.14.13344.
[30]
FUKUSHIMA M, FUKUNARI N, MURAKAMI T, et al. Reconfirmation of the accuracy of the taller-than-wide sign in multicenter collaborative research in Japan[J]. Endocr J, 2021, 68(8): 897-904. DOI: 10.1507/endocrj.EJ20-0379.
[31]
KIM E K, PARK C S, CHUNG W Y, et al. New sonographic criteria for recommending fine-needle aspiration biopsy of nonpalpable solid nodules of the thyroid[J]. AJR Am J Roentgenol, 2002, 178(3): 687-691. DOI: 10.2214/ajr.178.3.1780687.
[32]
YOON S J, YOON D Y, CHANG S K, et al. "Taller-than-wide sign" of thyroid malignancy: comparison between ultrasound and CT[J/OL]. AJR Am J Roentgenol, 2010, 194(5): W420-W424 [2025-02-22]. https://pubmed.ncbi.nlm.nih.gov/20410388/. DOI: 10.2214/AJR.09.3376.
[33]
DURANTE C, HEGEDÜS L, CZARNIECKA A, et al. 2023 European Thyroid Association Clinical Practice Guidelines for thyroid nodule management[J/OL]. Eur Thyroid J, 2023, 12(5): e230067 [2025-02-22]. https://pubmed.ncbi.nlm.nih.gov/37358008/. DOI: 10.1530/ETJ-23-0067.
[34]
中国医师协会介入医师分会超声介入专业委员会, 徐栋, 梁萍, 等. 超声引导下甲状腺结节及颈部淋巴结穿刺专家共识(2023版)[J]. 中华内科杂志, 2024, 63(6): 550-559. DOI: 10.3760/cma.j.cn112138-20240426-00122.
Interventional Ultrasound Committee of Chinese College of Interventionalists, XU D, LIANG P, et al. Expert consensus on ultrasound guided thyroid and neck lymph node puncture(2023 edition)[J]. Chin J Intern Med, 2024, 63(6): 550-559. DOI: 10.3760/cma.j.cn112138-20240426-00122.

上一篇 基于多模态MRI的影像组学模型对自身免疫性与单纯疱疹病毒性脑炎的鉴别诊断研究
下一篇 心脏磁共振特征追踪技术对非扩张型左室心肌病的诊断价值研究
  
诚聘英才 | 广告合作 | 免责声明 | 版权声明
联系电话:010-67113815
京ICP备19028836号-2