一种甲状腺弥漫性疾病超声语义特征量化评分系统的构建与应用

黄璟 谢帅 郭芳琪 徐琪 赵佳琦

引用本文: 黄璟,谢帅,郭芳琪,等. 一种甲状腺弥漫性疾病超声语义特征量化评分系统的构建与应用[J]. 海军军医大学学报,2026,47(6):751-757. DOI: 10.16781/j.CN31-2187/R.20250606.
Citation: HUANG J, XIE S, GUO F, et al. Construction and application of a quantitative scoring system for ultrasonographic semantic features of diffuse thyroid disease[J]. Acad J Naval Med Univ, 2026, 47(6): 751-757. DOI: 10.16781/j.CN31-2187/R.20250606.

一种甲状腺弥漫性疾病超声语义特征量化评分系统的构建与应用

doi: 10.16781/j.CN31-2187/R.20250606
基金项目: 

海军军医大学(第二军医大学)第二附属医院人才建设三年行动计划“金字塔人才工程”军事医学人才项目 1009;

同济大学附属上海市第四人民医院科研启动专项 SYKYQD06101;

上海市虹口区卫生健康委员会临床重点扶持专科建设项目 HKLCFC202404.

详细信息
    作者简介:

    黄璟,硕士生,主治医师. E-mail: jingrady@hotmail.com.

    通讯作者:

    赵佳琦, E-mail: qiqiblue67@163.com.

Construction and application of a quantitative scoring system for ultrasonographic semantic features of diffuse thyroid disease

Funds: 

Military Medical Talent Project of "Pyramid Talent Program" of Three-year Action Plan for Talent Construction of The Second Affiliated Hospital of Naval Medical University 1009;

Science and Technology Initiation Project of Shanghai Fourth People's Hospital Affiliated to Tongji University SYKYQD06101;

Clinical Key Supporting Project of Health Commission of Shanghai Hongkou District HKLCFC202404.

  • 摘要:
    目的 

    基于常规超声语义特征构建一种量化评分系统,并探讨其对甲状腺弥漫性疾病(DTD)的诊断价值。

    方法 

    前瞻性纳入2024年3月至2025年1月同济大学附属上海市第四人民医院、海军军医大学海军特色医学中心的185例临床疑似DTD患者为研究对象。基于干燥综合征唾液腺超声评分方法,构建包含8项核心指标(体积形态、被膜完整性、腺体回声强度、均匀性、结节、条索状高回声、钙化灶及血流模式)的DTD超声语义特征量化评分系统,总分0~19分。由2名高年资(工作经验≥5年)超声医师根据该评分系统独立完成双盲评分,并于初次评分后2周再随机抽取约30%的病例(n=56)进行二次评分,以评估观察者内信度。观察者间、观察者内一致性通过组内相关系数(ICC)评价。采用ROC曲线评估该评分系统的诊断效能。采用多因素二元logistic回归分析筛选超声语义特征中判别DTD的独立预测因素。

    结果 

    根据实验室检查(甲状腺功能、甲状腺过氧化物酶抗体、甲状腺球蛋白抗体和促甲状腺激素受体抗体)、临床病史及影像学随访结果,185例疑似DTD患者中DTD患者128例(69.19%),非DTD患者57例(30.81%)。DTD患者组的超声语义特征量化评分中位数高于非DTD患者组(9分vs 4分,P<0.001)。观察者间及观察者内一致性极佳(ICC=0.97、0.96)。ROC曲线分析显示,超声语义特征量化评分系统诊断DTD效能优异,AUC为0.91(95%CI 0.87~0.96);取最佳截断值5.5分时,灵敏度为93%,特异度为87%。多因素二元logistic回归分析显示,甲状腺腺体均匀性和血流评分是DTD的独立预测因素。

    结论 

    基于甲状腺常规超声语义特征构建的量化评分系统对DTD的诊断性能良好,可重复性高,有助于为建立甲状腺超声智能化、结构化报告提供参考,有望成为临床超声客观定量诊断DTD的辅助工具。

     

    Abstract:
    Objective 

    To develop a quantitative scoring system based on conventional ultrasonographic semantic features and to evaluate its diagnostic value for diffuse thyroid disease (DTD).

    Methods 

    A total of 185 patients with clinically suspected DTD were prospectively enrolled from Shanghai Fourth People's Hospital Affiliated to Tongji University and Naval Medical Center of Naval Medical University between Mar. 2024 and Jan. 2025. Based on the salivary gland ultrasonographic scoring method for Sjögren's syndrome, a quantitative scoring system of ultrasonographic semantic features of DTD was constructed, comprising 8 core indicators: volume and shape, capsular integrity, echogenicity, homogeneity, nodules, linear hyperechogenicity, calcifications, and vascular pattern, with a total score ranging from 0 to 19. Two senior ultrasound physicians (with ≥5 years of experience) independently performed double-blind scoring according to this system. To assess intra-observer reliability, 30% of the cases (n=56) were randomly selected and re-scored after a 2-week interval. Both inter-observer and intra-observer consistency were assessed using the intraclass correlation coefficient (ICC). Receiver operating characteristic (ROC) curve analysis was used to evaluate the diagnostic performance of the scoring system. Multivariate binary logistic regression was applied to identify independent predictive factors among the ultrasonographic semantic features for discriminating DTD.

    Results 

    Based on laboratory tests (thyroid function, thyroid peroxidase antibody, thyroglobulin antibody, and thyroid stimulating hormone receptor antibody), clinical history, and imaging follow-up, 128 of the 185 (69.19%) suspected DTD patients were confirmed as DTD, and 57 (30.81%) as non-DTD. The median score of the quantitative ultrasonographic semantic feature scoring system in the DTD group was significantly higher than that in the non-DTD group (9 vs 4, P < 0.001). Inter-observer and intra-observer agreements were excellent (ICC=0.97 and 0.96, respectively). ROC curve analysis showed that the quantitative ultrasonographic semantic feature scoring system demonstrated excellent diagnostic performance for DTD, with an area under the ROC curve of 0.91 (95% confidence interval 0.87-0.96). At the optimal cut-off value of 5.5, the sensitivity was 93% and the specificity was 87%. Multivariate regression revealed that glandular homogeneity and vascular score were independent predictors for DTD.

    Conclusion 

    The quantitative scoring system based on conventional thyroid ultrasonographic semantic features shows good diagnostic performance and high reproducibility for DTD. It may contribute to the development of an intelligent structured reporting system for thyroid ultrasound, showing promise as an auxiliary tool for objective and quantitative ultrasonographic diagnosis of diffuse thyroid disease.

     

  • 甲状腺弥漫性疾病(diffuse thyroid disease,DTD)是一组以甲状腺实质广泛性改变为特征的疾病,主要包括桥本甲状腺炎(Hashimoto’s thyroiditis,HT)、格雷夫斯病(Graves disease,GD)等。其病理机制复杂,临床表现多样,早期精准诊断直接影响治疗策略和预后管理[1]。超声检查作为DTD的一线影像学诊断手段,因其无创、实时、可重复性高等优势,被广泛应用于临床[2]。然而,传统超声诊断模式的效果取决于医师的主观经验,高度依赖医师对语义特征(如“回声减低”“不均匀”)的视觉解读,存在观察者间差异大、诊断标准不统一以及疾病监测随访能力不足等局限性,易导致诊断信心不足和随访策略不一[3]。近年来,新的超声定量技术(如弹性成像、超声影像组学)的发展为DTD诊断提供了新思路,但在广泛应用方面有局限性[4-5]。在此背景下,基于常规超声构建量化评估体系成为极具临床价值的方案。超声语义特征量化通过将主观的影像学描述转化为客观的数值评分,有望减少诊断的不确定性,提升报告的一致性与可重复性。国外现有的评价系统在国内的适用性及有效性尚需验证,且缺乏对关键语义特征的精细化与量化评估[6]。本研究通过构建一个包含8项核心超声语义特征的量化评分系统,旨在为DTD提供一种客观、准确的超声诊断工具,以辅助临床早期识别、精准诊断及个体化管理策略制定。

    本研究采用前瞻性队列设计,样本量估算基于ROC曲线分析中AUC的统计方法,参考诊断试验样本量估算指南[7],设定α=0.05(双侧)、β=0.2(检验效能80%),预期AUC=0.85。采用R 4.4.0软件计算,最低样本量为110例。研究共初筛2024年3月至2025年1月于同济大学附属上海市第四人民医院、海军军医大学海军特色医学中心就诊的疑似DTD患者共203例,按纳入和排除标准,排除合并较大结节(≥10 mm)患者6例、既往有甲状腺手术史者5例、失访5例、影像图质量不佳者2例,最终纳入185例临床疑似DTD患者。其中同济大学附属上海市第四人民医院106例,海军军医大学海军特色医学中心79例。

    纳入标准:(1)甲状腺超声显示弥漫性改变;(2)实验室检查[甲状腺功能、甲状腺过氧化物酶抗体(thyroid peroxidase antibody,TPOAb)、甲状腺球蛋白抗体(thyroglobulin antibody,TgAb)、促甲状腺激素受体抗体(thyroid stimulating hormone receptor antibody,TSHRAb)]及临床资料完整。排除标准:(1)甲状腺超声显示正常均匀回声(与颌下腺回声一致,结构清晰);(2)局灶性病变为主(如单个结节≥10 mm或占腺体体积50%以上);(3)既往有甲状腺手术或放射性碘治疗史;(4)临床资料不全或无法配合完成随访。

    根据实验室检查指标(甲状腺功能、TPOAb/TgAb/TSHRAb)、临床病史及影像学随访结果,将患者分为DTD组(128例)和非DTD组(57例)。纳入DTD组条件包括:(1)甲状腺超声显示弥漫性改变(全腺体低回声、结构紊乱或血流分布异常);(2)临床确诊为DTD(HT:TPOAb>60 IU/mL且TSH异常;GD:TRAb≥1.5 IU/L且伴甲状腺毒症等);(3)排除甲状腺结节≥10 mm或其他局部病变。非DTD组纳入条件包括:(1)甲状腺超声显示弥漫性改变,但不符合DTD实验室诊断标准;(2)无甲状腺疾病病史及相关症状;(3)随访≥6个月且状态稳定。

    本研究符合《赫尔辛基宣言》及医学伦理学要求,并经两家医院伦理委员会批准(1009和HKLCFC202404)。所有患者均获得本人或其监护人的同意,并签署知情同意书。

    DTD的典型超声图像见图 1。由2名经验丰富的超声医师回顾性分析超声图像并记录超声语义特征[8]。超声图像语义特征包括体积形态、被膜完整性、腺体回声强度、均匀性、结节、条索状高回声、钙化灶及血流模式[9]。基于既往关于干燥综合征唾液腺超声评分方法[10-11],构建DTD超声语义特征量化评分系统,总分范围0~19分,具体评分标准见表 1

    图  1  DTD的典型超声图像
    Fig.  1  Representative ultrasound images of DTD
    A: Markedly decreased echogenicity (lower than that of the adjacent sternocleidomastoid muscle) with diffuse and pronounced heterogeneity; B: Obviously decreased echogenicity (lower than that of normal thyroid parenchyma) with characteristic "mesh-like" changes; C: Color Doppler shows markedly increased vascularity, presenting as a "thyroid inferno" (Alder semi-quantitative blood flow grading: grade Ⅳ); D: Mildly increased vascularity (Alder semi-quantitative blood flow grading: grade Ⅱ). DTD: Diffuse thyroid disease.
    下载: 全尺寸图片
    表  1  DTD超声语义特征量化评分标准
    Table  1  Quantitative scoring criteria of ultrasonographic semantic features for DTD
    Feature Grade
    0 1 2 3
    Volume & shape Normal Plump Atrophic
    Capsular integrity Clear & intact Partially interrupted Irregular/indistinct
    Echogenicity Isoechoic Focal/mildly Moderately Markedly
    Homogeneity Homogeneous Mildly Obviously Markedly, mesh-like
    Nodule Absent Solitary Multiple
    Linear hyperechogenicity Absent Focal scattered Widespread
    Calcification Absent Coarse Microcalcification
    Vascularity patterna Normal, grade Ⅰ Mildly increased, grade Ⅱ Moderately increased, grade Ⅲ Markedly increased, grade Ⅳ
    a: Classification being performed based on Alder semi-quantitative blood flow grading; -: No corresponding score for that feature. DTD: Diffuse thyroid disease.

    使用Philips IU Elite超声诊断仪(L12-5探头,频率范围5~12 MHz),预设甲状腺专用成像条件(机械指数0.6,帧频28 Hz)。患者取仰卧位,颈部过伸,探头轻触皮肤避免加压,采集双侧腺体横切面及纵切面图像(每侧至少3个切面)。所有图像以DICOM格式保存至医学影像存档与通信系统,确保分辨率达标并保证数据完整性。隐匿病例信息(仅显示编号及检查日期)后由2名高年资超声医师(工作经验≥5年)根据超声语义特征量化评分标准独立完成评分。于初次评分后2周,随机抽取30% 病例(n=56)进行二次评分,以评估组内信度。评分差异≥3分的病例由第3名专家判定。

    采用Python 3.13、R 4.4.0软件进行统计学分析。计量资料先经Shapiro-Wilk检验,符合正态分布者以x±s表示,组间比较采用独立样本t检验;不符合正态分布的计量资料以MQ1Q3)表示,组间比较采用Mann-Whitney U检验。分类资料以频数和百分数表示,组间比较采用χ2检验或Fisher确切概率法。采用ROC曲线分析评估该评分系统的诊断效能,根据最大约登指数确定最佳截断值。采用组内相关系数(intraclass correlation coefficient,ICC)评估观察者间一致性[ICC(2, 1)]及观察者内一致性[ICC(1, 1)]。将单因素二元logistic回归分析中P<0.05的变量纳入多因素二元logistic回归模型(逐步向前法),以筛选超声语义特征中判别DTD的独立预测因素。检验水准(α)为0.05。

    185例疑似DTD患者中非DTD患者57例(30.81%),DTD患者128例(69.19%)。DTD患者组与非DTD患者组超声语义特征总分、甲状腺被膜完整性得分、腺体回声强度得分、均匀性得分、结节得分、条索状高回声得分、血流模式得分及性别差异均有统计学意义(均P<0.05),而年龄、甲状腺体积形态得分、甲状腺内钙化灶得分在两组间的差异均无统计学意义(均P>0.05,表 2)。

    表  2  两组患者基线资料及超声语义特征量化评分比较
    Table  2  Comparison of baseline characteristics and quantitative scores of ultrasonographic semantic features between 2 groups
    Variable Total N=185 Non-DTD N=57 DTD N=128 Statistic P value
    Age/year, x±s 53.46±13.72 53.67±13.25 53.38±13.97 t=0.13 0.894
    Gender, n (%) χ2=14.09 <0.01
      Male 70 (37.84) 33 (57.89) 37 (28.91)
      Female 115 (62.16) 24 (42.11) 91 (71.09)
    Ultrasonvgraphic semantic score, M (Q1, Q3)
      Total 7 (5, 10) 4 (3, 5) 9 (7, 11) Z=-9.00 <0.001
      Volume & shape 0 (0, 1) 0 (0, 1) 0 (0, 1) Z=-1.25 0.212
      Capsular integrity 0 (0, 1) 0 (0, 0) 1 (0, 1) Z=-6.04 <0.01
      Echogenicity 1 (1, 2) 1 (0, 1) 2 (1, 2) Z=-6.70 <0.01
      Homogeneity 2 (1, 3) 1 (1, 1) 3 (2, 3) Z=-9.03 <0.01
      Nodule 0 (0, 1) 0 (0, 1) 0 (0, 1) Z=-2.28 0.023
      Linear hyperechogenicity 2 (1, 2) 1 (1, 2) 2 (1, 2) Z=-4.23 <0.01
      Calcification 0 (0, 0) 0 (0, 0) 0 (0, 0) Z=-1.65 0.098
      Vascularity pattern 1 (0, 1) 1 (0, 1) 1 (1, 1) Z=-5.38 <0.01

    2名观察者评分的ICC(2, 1)为0.97(95%CI 0.96~0.98),表明2名观察者的评分一致性极佳。配对t检验结果显示,两位观察者评分差异无统计学意义(P=0.218)。Bland-Altman分析显示,平均差异为-0.07分,95%一致性界限为-1.59~1.45分。所有样本点均落在一致性界限内,无系统性偏差。观察者内一致性分析显示,ICC(1, 1)为0.96(95%CI 0.94~0.98),表明同一位医师前后两次评分的一致性极佳。配对t检验结果显示,两次评分差异无统计学意义(P=0.218)。Bland-Altman分析显示,平均差异为-0.09分,95%一致性界限为-1.52~1.34分,所有样本点均落在一致性界限内,无系统性偏差。

    ROC曲线分析显示,构建的超声语义特征量化评分系统在鉴别DTD与非DTD中表现优异,AUC为0.91(95%CI 0.87~0.96);最佳临界值为5.5分时,评分系统整体准确度为89%(95%CI 83%~ 93%),灵敏度为93%(95%CI 86%~100%),特异度为87%(95%CI 81%~93%),阴性预测值为97%(95%CI 93%~100%),阳性预测值为76%(95%CI 66%~86%)。见图 2

    图  2  超声语义特征量化评分系统预测DTD的ROC曲线
    Fig.  2  ROC curve for predicting DTD using the quantitative scoring system of ultrasonographic semantic features
    DTD: Diffuse thyroid disease; ROC: Receiver operating characteristic; AUC: Area under curve; 95%CI: 95% confidence interval.
    下载: 全尺寸图片

    单因素二元logistic回归分析显示,超声量化评分指标均与DTD相关(均P<0.05)。多因素二元logistic回归分析显示,各超声量化评分指标中腺体均匀性评分和血流模式评分是判别DTD的独立预测因素。甲状腺腺体均匀性评分每增加1分,DTD检出概率增加9.20倍(OR=10.20,95%CI 4.78~21.76);血流模式评分每增加1分,DTD检出概率增加2.27倍(OR=3.27,95%CI 1.32~8.14)。见表 3

    表  3  超声语义特征量化评分与DTD关系的单因素及多因素二元logistic回归分析
    Table  3  Univariate and multivariate binary logistic regression analyses of the association between quantitative scores of ultrasonographic semantic features and DTD
    Variable Univariate Multivariate
    b SE Z value P value OR (95%CI) b SE Z value P value OR (95%CI)
    Capsular integrity 2.75 0.61 4.53 <0.001 15.71 (4.77, 51.77)
    Echogenicity 1.34 0.23 5.93 <0.001 3.82 (2.45, 5.95)
    Homogeneity 2.48 0.38 6.54 <0.001 11.98 (5.69, 25.23) 2.32 0.39 6.01 <0.001 10.20 (4.78, 21.76)
    Nodule 0.75 0.33 2.30 0.021 2.12 (1.12, 4.04)
    Linear hyperechogenicity 1.00 0.25 4.01 <0.001 2.73 (1.67, 4.45)
    Vascularity pattern 1.63 0.32 5.02 <0.001 5.09 (2.70, 9.62) 1.19 0.46 2.55 0.011 3.27 (1.32, 8.14)
    DTD: Diffuse thyroid disease; b: Regression coefficient; SE: Standard error; OR: Odds ratio; 95%CI: 95% confidence interval.

    DTD是一组以甲状腺实质广泛性受累为特征的疾病,主要包括HT、GD等自身免疫性甲状腺疾病[12],其不仅是导致甲状腺功能异常的常见原因,更因病程迁延、需长期管理而严重影响患者生活质量。该类疾病的早期、准确诊断始终是临床面临的挑战[13]。超声检查虽是诊断DTD的一线手段,但日常临床实践中常面临普遍困境,虽能识别出诸如“回声减低”“不均匀”“血流丰富”等典型征象,但受限于对这些临床超声语义特征的主观、定性评估,当合并甲状腺结节时,则更缺乏客观量化的标准来界定其严重程度与综合评估病变可能性,致使诊断报告在不同医师间的解读差异较大,最终影响临床决策的精准性,既可能因诊断偏差导致不必要的焦虑与激进干预,也可能因诊断不足而延误治疗时机[14]。为克服这一局限性,学界尝试了多种量化方法,如韩国学者将甲状腺影像报告和数据系统(thyroid imaging reporting and data system,TI-RADS)的框架拓展并应用于DTD的定量研究,为该领域的标准化评估提供了新思路[15]。但该体系在我国人群中的适用性尚未得到充分验证,更重要的是,该系统对各项超声语义特征所赋予的权重界定仍存在较大争议。这种争议与不确定性不仅限制了DTD TI-RADS在国内的推广应用,也凸显了开发更适合我国人群特征、具有明确客观量化标准的DTD评估体系的迫切需求。针对这一临床挑战,基于韩国学者提出的DTD TI-RADS框架[15]及干燥综合征唾液腺超声量化评分方法[10-11],本研究构建了DTD超声语义特征量化评分系统,将高度依赖医师经验的诊断过程转化为客观、可重复的量化工具,为解决DTD超声诊断一致性的临床痛点提供了新方案。

    本研究构建的临床超声语义特征量化评分系统在区分DTD中表现出优异的诊断效能(AUC=0.91),且灵敏度(93%)与特异度(87%)均处于较高水平,模型的阴性预测值高达97%,提示其在排除诊断方面具有较高价值,而阳性预测值为76%,提示阳性结果需结合其他检查确认。这一结果表明,该模型在临床实践中能够有效辅助医师区分DTD与非DTD患者,有助于减少漏诊和避免过度检查。传统超声诊断依赖医师经验,存在主观性差异,而本系统通过临床超声语义特征量化评估,显著提升了诊断的一致性,提供了可靠的决策支持工具。与韩国学者提出的DTD TI-RADS系统[15]相比,本系统在本研究队列中表现出更高的AUC(0.91 vs 0.867),且指标体系更贴合DTD的疾病特征。

    本研究提出的核心超声指标(腺体均匀性、血流模式)与DTD的病理生理机制及既往研究结论一致:腺体均匀性减低是DTD的典型特征,其病理基础为淋巴细胞浸润或纤维化导致的腺体结构破坏,在HT和GD患者中,甲状腺腺体回声不均匀是炎症活动的一个显著特征[16-17]。血流模式异常可能与疾病活动性密切相关,部分学者的研究进一步支持了血流评分作为动态监测指标的可行性[18-19]。条索状高回声是纤维化标志,回声强度改变对组织水肿或增生具有提示作用[20-21],这两项均在本研究中得到验证。值得注意的是,尽管上述指标在既往研究中已被探讨,但传统定性描述存在主观性差异,而本研究通过量化整合构建了多维评估体系。例如,单纯条索状高回声可能在退行性病变中出现,但联合异常血流及回声不均可显著提升诊断特异性,这一策略不仅避免了单一指标的局限性,更与疾病不同阶段的病理进程(早期炎症充血、晚期纤维化)动态对应,为临床提供了兼具鉴别与分期价值的影像学工具。在本研究中,钙化灶这一特征在两组间差异无统计学意义。分析认为,该特征较多用于甲状腺结节良恶性鉴别,而在弥漫性病变这一良性病变中,钙化多源于滤泡退化、出血、纤维化后的终末改变[22]。患者年龄及腺体形态在DTD患者组与非DTD患者组间差异亦无统计学意义。年龄方面,DTD好发于中青年人群,年龄分布在组间存在重叠,故未呈现差异;DTD早期腺体轮廓多保持规整,形态学指标不敏感,因此形态评分的区分度有限。

    本系统在验证既往研究核心观点的同时,通过标准量化与多指标协同优化了诊断效能,为DTD的精准诊疗提供了更具临床适用性的解决方案。2名超声医师的评分一致性(ICC=0.97)表明,该评分系统具有高度的可重复性。Bland-Altman分析显示平均差异仅为0.07分,且95%一致性界限较窄(-1.59~1.45),说明系统对操作者依赖性较低。这一特性在临床实践中尤为重要,可减少不同医疗机构或医师间的诊断分歧,推动甲状腺超声评估的标准化进程。此外,高一致性为计算机辅助诊断系统的开发奠定了基础,人工评分数据可为后续人工智能定量分析提供一定的参考。二元logistic回归分析结果显示,甲状腺腺体均匀性和血流模式评分是独立预测因素(均P<0.05),提示医师在超声检查中应重点关注这两项指标。既往研究证实,腺体回声均匀性的动态演变与DTD的病程进展密切相关,异常血流模式亦反映腺体功能状态及炎症活动程度[21]。基于此,对于超声表现为回声不均伴异常血流模式的患者,即使其甲状腺功能指标尚未出现异常,也需纳入密切随访,以实现早期干预。此外,该结果可为治疗监测提供新思路,如通过动态评估均匀性和血流变化来追踪疾病进展或疗效。

    本研究样本量较小(n=185),未来需通过多中心、大样本研究验证模型的普适性。由于DTD常规诊疗中极少进行活检导致金标准缺失,采用综合诊断标准可能引入误分类偏倚,非DTD患者组中可能混杂了部分早期或亚临床DTD患者,这将导致两组间的超声特征差异被低估,从而可能使本系统的诊断效能偏向于保守估计;DTD患者组中亦可能纳入部分非典型或不满足严格诊断标准的病例,将进一步增加组间特征的趋同性。上述偏倚可能导致本系统在实际临床应用中的准确性低于本研究结果。纳入疾病类型(HT、GD等)的亚组分析无法展开,不同DTD亚型的超声特征差异可能导致系统在亚型鉴别中存在不足,后续需针对性验证。此外,本研究受到随访时间限制,非DTD患者组的6个月随访可能不足以完全排除亚临床病例,未来可延长随访至12个月以提升非DTD患者定义的准确性。下一步可结合先进的人工智能技术,利用图像分析将主观评分转化为自动化定量参数,进一步提升超声诊断效率。同时,探索该评分系统与甲状腺相关血清学指标的联合应用,可能有助于建立更全面的诊断策略。

    综上所述,本研究通过提取与分析DTD的关键临床超声语义特征,成功构建了一个综合量化评分系统。该模型具有良好的诊断性能与较高的观察者间一致性,为DTD的客观化、定量化超声诊断提供了具有良好临床应用前景的新工具,有望辅助临床医师提升诊断信心、制定个体化的精准随访与管理策略。

  • 图  1   DTD的典型超声图像

    Fig.  1   Representative ultrasound images of DTD

    A: Markedly decreased echogenicity (lower than that of the adjacent sternocleidomastoid muscle) with diffuse and pronounced heterogeneity; B: Obviously decreased echogenicity (lower than that of normal thyroid parenchyma) with characteristic "mesh-like" changes; C: Color Doppler shows markedly increased vascularity, presenting as a "thyroid inferno" (Alder semi-quantitative blood flow grading: grade Ⅳ); D: Mildly increased vascularity (Alder semi-quantitative blood flow grading: grade Ⅱ). DTD: Diffuse thyroid disease.

    下载: 全尺寸图片

    图  2   超声语义特征量化评分系统预测DTD的ROC曲线

    Fig.  2   ROC curve for predicting DTD using the quantitative scoring system of ultrasonographic semantic features

    DTD: Diffuse thyroid disease; ROC: Receiver operating characteristic; AUC: Area under curve; 95%CI: 95% confidence interval.

    下载: 全尺寸图片

    表  1   DTD超声语义特征量化评分标准

    Table  1   Quantitative scoring criteria of ultrasonographic semantic features for DTD

    Feature Grade
    0 1 2 3
    Volume & shape Normal Plump Atrophic
    Capsular integrity Clear & intact Partially interrupted Irregular/indistinct
    Echogenicity Isoechoic Focal/mildly Moderately Markedly
    Homogeneity Homogeneous Mildly Obviously Markedly, mesh-like
    Nodule Absent Solitary Multiple
    Linear hyperechogenicity Absent Focal scattered Widespread
    Calcification Absent Coarse Microcalcification
    Vascularity patterna Normal, grade Ⅰ Mildly increased, grade Ⅱ Moderately increased, grade Ⅲ Markedly increased, grade Ⅳ
    a: Classification being performed based on Alder semi-quantitative blood flow grading; -: No corresponding score for that feature. DTD: Diffuse thyroid disease.

    表  2   两组患者基线资料及超声语义特征量化评分比较

    Table  2   Comparison of baseline characteristics and quantitative scores of ultrasonographic semantic features between 2 groups

    Variable Total N=185 Non-DTD N=57 DTD N=128 Statistic P value
    Age/year, x±s 53.46±13.72 53.67±13.25 53.38±13.97 t=0.13 0.894
    Gender, n (%) χ2=14.09 <0.01
      Male 70 (37.84) 33 (57.89) 37 (28.91)
      Female 115 (62.16) 24 (42.11) 91 (71.09)
    Ultrasonvgraphic semantic score, M (Q1, Q3)
      Total 7 (5, 10) 4 (3, 5) 9 (7, 11) Z=-9.00 <0.001
      Volume & shape 0 (0, 1) 0 (0, 1) 0 (0, 1) Z=-1.25 0.212
      Capsular integrity 0 (0, 1) 0 (0, 0) 1 (0, 1) Z=-6.04 <0.01
      Echogenicity 1 (1, 2) 1 (0, 1) 2 (1, 2) Z=-6.70 <0.01
      Homogeneity 2 (1, 3) 1 (1, 1) 3 (2, 3) Z=-9.03 <0.01
      Nodule 0 (0, 1) 0 (0, 1) 0 (0, 1) Z=-2.28 0.023
      Linear hyperechogenicity 2 (1, 2) 1 (1, 2) 2 (1, 2) Z=-4.23 <0.01
      Calcification 0 (0, 0) 0 (0, 0) 0 (0, 0) Z=-1.65 0.098
      Vascularity pattern 1 (0, 1) 1 (0, 1) 1 (1, 1) Z=-5.38 <0.01

    表  3   超声语义特征量化评分与DTD关系的单因素及多因素二元logistic回归分析

    Table  3   Univariate and multivariate binary logistic regression analyses of the association between quantitative scores of ultrasonographic semantic features and DTD

    Variable Univariate Multivariate
    b SE Z value P value OR (95%CI) b SE Z value P value OR (95%CI)
    Capsular integrity 2.75 0.61 4.53 <0.001 15.71 (4.77, 51.77)
    Echogenicity 1.34 0.23 5.93 <0.001 3.82 (2.45, 5.95)
    Homogeneity 2.48 0.38 6.54 <0.001 11.98 (5.69, 25.23) 2.32 0.39 6.01 <0.001 10.20 (4.78, 21.76)
    Nodule 0.75 0.33 2.30 0.021 2.12 (1.12, 4.04)
    Linear hyperechogenicity 1.00 0.25 4.01 <0.001 2.73 (1.67, 4.45)
    Vascularity pattern 1.63 0.32 5.02 <0.001 5.09 (2.70, 9.62) 1.19 0.46 2.55 0.011 3.27 (1.32, 8.14)
    DTD: Diffuse thyroid disease; b: Regression coefficient; SE: Standard error; OR: Odds ratio; 95%CI: 95% confidence interval.
  • [1] ALEXANDER L F, PATEL N J, CASERTA M P, et al. Thyroid ultrasound: diffuse and nodular disease[J]. Radiol Clin North Am, 2020, 58(6): 1041-1057. DOI: 10.1016/j.rcl.2020.07.003.
    [2] AHN H S, KIM D W, LEE Y J, et al. Diagnostic accuracy of real-time sonography in differentiating diffuse thyroid disease from normal thyroid parenchyma: a multicenter study[J]. Am J Roentgenol, 2018, 211(3): 649-654. DOI: 10.2214/ajr.17.19164.
    [3] DE CARLOS J, GARCIA J, BASTERRA F J, et al. Interobserver variability in thyroid ultrasound[J]. Endocrine, 2024, 85(2): 730-736. DOI: 10.1007/s12020-024-03731-5.
    [4] SEDLACKOVA Z, HERMAN J, FURST T, et al. Shear wave elastography in diffuse thyroid disease[J]. Biomed Pap Med Fac Univ Palacky Olomouc Czech Repub, 2021, 165(2): 152-156. DOI: 10.5507/bp.2020.018.
    [5] STOIAN D, BORLEA A, MOISA-LUCA L, et al. Multiparametric ultrasound-based assessment of overt hyperthyroid diffuse thyroid disease[J]. Front Endocrinol, 2023, 14: 1300447. DOI: 10.3389/fendo.2023.1300447.
    [6] BAEK H J, KIM D W, RYU K H, et al. Thyroid imaging reporting and data system for detecting diffuse thyroid disease on ultrasonography: a single-center study[J]. Front Endocrinol, 2019, 10: 776. DOI: 10.3389/fendo.2019.00776.
    [7] HAJIAN-TILAKI K. Sample size estimation in diagnostic test studies of biomedical informatics[J]. J Biomed Inform. 2014, 48: 193-204. DOI: 10.1016/j.jbi.2014.02.013.
    [8] 赵静, 赵利辉, 张蕾, 等. 超声影像组学联合术前临床超声语义特征预测甲状腺髓样癌侧颈部淋巴结转移的风险[J]. 中华普通外科杂志, 2024, 39(12): 929-934. DOI: 10.3760/cma.j.cn113855-20240908-00576.
    [9] SCHULZ S L, SEEBERGER U, HENGSTMANN J H. Color Doppler sonography in hypothyroidism[J]. Eur J Ultrasound, 2003, 16(3): 183-189. DOI: 10.1016/S0929-8266(02)00072-1.
    [10] HOCEVAR A, AMBROZIC A, ROZMAN B, et al. Ultrasonographic changes of major salivary glands in primary Sjogren's syndrome. Diagnostic value of a novel scoring system[J]. Rheumatology (Oxford), 2005, 44(6): 768-72. DOI: 10.1093/rheumatology/keh588.
    [11] LEE K A, LEE S H, KIM H R. Ultrasonographic changes of major salivary glands in primary sjögren's syndrome[J]. J Clin Med, 2020, 9(3): 803. DOI: 10.3390/jcm9030803.
    [12] VARGAS-URICOECHEA H. Molecular mechanisms in autoimmune thyroid disease[J]. Cells, 2023, 12(6): 918. DOI: 10.3390/cells12060918.
    [13] 黄璟, 赵佳琦. 甲状腺弥漫性病变的超声影像学定量诊断技术研究进展[J]. 海军军医大学学报, 2024, 45(9): 1134-1140. DOI: 10.16781/j.CN31-2187/R.20220515.
    [14] CHOI S H, KIM E K, KWAK J Y, et al. Interobserver and intraobserver variations in ultrasound assessment of thyroid nodules[J]. Thyroid, 2010, 20(2): 167-172. DOI: 10.1089/thy.2008.0354.
    [15] BAEK H J, RYU K H, AN H J, et al. Validation of diagnostic performance and interobserver agreement of DTD-TIRADS for diffuse thyroid disease on ultrasound: a single-center study[J]. Am J Roentgenol, 2021, 216(5): 1329-1334. DOI: 10.2214/ajr.20.23231.
    [16] RALLI M, ANGELETTI D, FIORE M, et al. Hashimoto's thyroiditis: an update on pathogenic mechanisms, diagnostic protocols, therapeutic strategies, and potential malignant transformation[J]. Autoimmun Rev, 2020, 19(10): 102649. DOI: 10.1016/j.autrev.2020.102649.
    [17] ANTONELLI A, FALLAHI P, ELIA G, et al. Graves' disease: clinical manifestations, immune pathogenesis (cytokines and chemokines) and therapy[J]. Best Pract Res Clin Endocrinol Metab, 2020, 34(1): 101388. DOI: 10.1016/j.beem.2020.101388.
    [18] 王花, 梁光文, 李光来, 等. 彩色多普勒超声联合MRI在甲亢伴Graves病的诊断应用分析[J]. 医学影像学杂志, 2023, 33(7): 1274-1277. DOI: 10.20258/j.cnki.1006-9011.2023.07.036.
    [19] LI S, DING Z, LI X, et al. A meta-analysis: elucidating diagnostic thresholds of peak systolic flow velocities in thyroid arteries for the discrimination of Graves' disease and destructive thyrotoxicosis[J]. Front Endocrinol, 2024, 15: 1393126. DOI: 10.3389/fendo.2024.1393126.
    [20] ORSATTI A, DE LEO A, CHIARUCCI F, et al. Multifocal fibrosing thyroiditis: an under-recognized mimicker of papillary thyroid carcinoma[J]. Endocr Pathol, 2022, 33(3): 335-345. DOI: 10.1007/s12022-022-09726-0.
    [21] CORSELLO A, CACCIAMANI FANELLI P M, LENER L, et al. Biochemical and ultrasonographic parameters predicting long-term hypothyroidism after subacute thyroiditis[J]. Int J Mol Sci, 2025, 26(9): 4176. DOI: 10.3390/ijms26094176.
    [22] LACOUT A, CHEVENET C, THARIAT J, et al. Thyroid calcifications: a pictorial essay[J]. J Clin Ultrasound, 2016, 44(4): 245-251. DOI: 10.1002/jcu.22345.
WeChat 点击查看大图
图(2)  /  表(3)
出版历程
  • 收稿日期:  2025-09-05
  • 接受日期:  2026-05-30

目录

    /

    返回文章
    返回