第一部分

集成MRI直方图分析在鉴别头颈部鳞状细胞癌颈部转移淋巴结中的价值

【目的】 探讨集成磁共振成像（SyMRI）直方图参数在区分头颈部鳞状细胞癌（HNSCC）患者颈部转移与非转移淋巴结中的作用，并构建预测模型。

【方法】 本研究前瞻性纳入149例经病理学证实的HNSCC患者的颈部淋巴结，其中58个转移性淋巴结，91个非转移性淋巴结。将淋巴结分层、按7:3的比例随机分为训练集和测试集。勾画所有淋巴结的感兴趣区获得SyMRI定量图（T1、T2和PD map）的直方图参数，测量每个淋巴结的表观扩散系数（ADC）值、短径和长径。在训练集中筛选出差异显著的参数，使用logistic回归分析构建模型，并绘制列线图。3名放射科医生参与淋巴结的主观评价。使用受试者工作特征曲线、临床决策曲线比较模型的诊断性能。

【结果】 3个直方图参数纳入构建SyMRI模型，直方图参数、ADC值、短径纳入构建联合模型。在测试集中，各模型的AUC值分别为0.882（SyMRI模型），0.755（DWI模型），0.952（联合模型）。联合模型具有最高的诊断性能，在训练集和测试集中的准确性分别为0.905和0.864。联合模型在所有淋巴结（AUC：0.956 vs. 0.708-0.811，P 值均< 0.001）及亚厘米淋巴结中（AUC：0.909 vs. 0.568-0.684，P 值均< 0.001）的诊断性能均显著优于放射科医师的主观评价。

【结论】 SyMRI直方图参数在区分HNSCC患者转移性和非转移性颈部淋巴结中具有一定价值，并且与DWI和淋巴结短径联合时，诊断效能最佳。

第二部分

集成MRI直方图分析在预测头颈部鳞状细胞癌生物标志物表达中的价值

【目的】 探讨集成磁共振成像（SyMRI）直方图参数与头颈部鳞状细胞癌（HNSCC）生物标志物表达水平之间的联系。

【方法】 本研究前瞻性纳入87例经病理学检查确认为HNSCC的患者。提取原发肿瘤SyMRI定量图（T1、T2和PD map）的直方图参数。使用独立样本t检验或Mann-Whitney U检验比较两组参数的之间的差异，将差异显著的参数纳入多因素回归分析。使用Spearman相关性分析评估直方图参数与生物标志物之间的相关性。使用受试者工作特征曲线进行分析并计算曲线下面积（AUC）。

【结果】 T1及PD参数与肿瘤分化程度呈正相关，PD第90个百分位数诊断的AUC为0.736。PD最小值与EGFR表达呈负相关（r = -0.258），其余参数呈正相关（r = 0.261～0.336）。PD参数与Ki-67表达水平呈正相关（r = 0.344～0.386），PD中位数相关性最强（r = 0.319）。T1、T2及PD map中熵及平均绝对偏差均与p16状态具有中度相关性（r = -0.403～-0.508），且在HPV感染组及非感染组间差异显著。T2熵、PD 偏度为预测HPV感染的独立预后因子，T2熵具有最高的诊断性能（AUC = 0.753）。

【结论】 SyMRI直方图参数能够反映肿瘤组织的内在生物学特征，在评估头颈部鳞状细胞癌肿瘤分化等级、p16状态、Ki-67和EGFR表达水平方面具有一定价值。

Part 1

The value of Synthetic MRI based histogram analysis in diagnosis of cervical lymph node metastasis in head and neck squamous cell carcinoma

Objectives: To explore the role of Synthetic MRI (SyMRI) histogram parameters in differentiating metastatic from non-metastatic cervical lymph nodes (LNs) in head and neck squamous cell carcinoma (HNSCC) patients, and construct a practical model.

Methods: A total of 149 pathologically confirmed LNs (metastatic LNs: 58, non-metastatic LNs: 91) were prospectively included in the study. LNs were stratified and randomly divided into a training set and an independent test set in a ratio of 7:3. Volumes of interest were delineated for all lymph nodes to obtain histogram parameters from SyMRI quantitative maps (T1, T2, and PD maps). The apparent diffusion coefficient (ADC) values, short- and long-axis diameter of each LN were measured. Parameters showing significant differences in the training set were selected, and a logistic regression analysis was performed to construct the models, followed by the creation of a nomogram. Three radiologists participated in the subjective evaluation of the lymph nodes. The diagnostic performance of different models was assessed using receiver operating characteristic curves and decision curve analysis.

Results: Three histogram parameters were included to construct the SyMRI model, while histogram parameters, ADC values, and short-axis diameter were incorporated into the combined model. In the test set, the AUC values of the models were 0.882 (SyMRI model), 0.755 (DWI model), and 0.952 (combined model), respectively. The combined model demonstrated the highest diagnostic performance, with accuracies of 0.905 and 0.864 in the training and test sets, respectively. The diagnostic performance of the combined model was significantly superior to the subjective evaluations by radiologists for all lymph nodes (AUC: 0.956 vs. 0.708–0.811, all P < 0.001) as well as for sub-centimeter lymph nodes (AUC: 0.909 vs. 0.568–0.684, all P < 0.001).

Conclusion: Histogram parameters derived from SyMRI are feasible in discriminating metastatic from non-metastatic cervical LNs in HNSCC, and the diagnostic efficacy is optimal when combined with DWI and size.

Part 2

The value of Synthetic MRI based histogram analysis in predicting biomarker expression in head and neck squamous cell carcinoma

Objective: To investigate the association between synthetic MRI (SyMRI) histogram parameters and expression of biomarkers in head and neck squamous cell carcinoma (HNSCC).

Methods: This study prospectively enrolled 87 patients with pathologically confirmed HNSCC. Histogram parameters were extracted from SyMRI quantitative maps (T1, T2, and PD maps) of primary tumor. Independent samples t-tests or Mann-Whitney U tests were employed to compare differences in parameters between the groups. Statistically significant parameters were included in multivariate regression analysis. Spearman correlation analysis was used to evaluate the relationship between histogram parameters and biomarkers. Receiver operating characteristic curve analysis was performed, and the area under the curve (AUC) was calculated.

Results: T1 and PD parameters showed a positive correlation with tumor differentiation grade, with the PD 90th percentile exhibiting an AUC of 0.736 for diagnosis. Minimum from PD map was negatively correlated with EGFR expression (r = -0.258), while other parameters showed positive correlations (r = 0.261-0.336). PD parameters were positively correlated with Ki-67 expression levels (r = 0.344~0.386), with the PD median demonstrating the strongest correlation (r = 0.319). Entropy and mean absolute deviation from T1, T2, and PD maps were moderately correlated with p16 status (r = -0.403-0.508) and showed significant differences between HPV-positive and HPV-negative groups. T2 entropy and PD skewness were independent predictive factors for HPV infection, with T2 entropy demonstrating the highest diagnostic performance (AUC = 0.753).

Conclusion: SyMRI histogram parameters can reflect intrinsic characteristics of tissue and show value in assessing differentiation grade, p16 status, Ki-67 and EGFR expression in HNSCC.