第一部分

目的：探讨单指数、双指数、拉伸指数三种IVIM-DWI模型与喉部及下咽鳞状细胞癌放化疗后患者预后及生存风险的关系。

材料与方法：分析2014年12月至2016年3月经我院诊治的45例喉或下咽鳞状细胞癌患者，治疗前行MRI（包括IVIM-DWI序列）检查。IVIM序列设定12个b值0，图像后处理经三种模型生成对应参数图，分别测得单指数模型参数包括表观扩散系数平均值（ADCmean）、最大值（ADCmax）、最小值（ADCmin）和差值（ADCrange），双指数模型参数包括真扩散系数（D）、伪扩散系数（D*）、灌注分数（f），拉伸模型参数包括分布扩散系数（DDC）和扩散非均质性指数（α）。对入组病例随访5年，获得患者肿瘤复发进展、转移及生存情况。

结果：患者治疗失败组31例，肿瘤控制组14例。治疗失败组ADCmean、ADCmax、ADCmin、D、f值显著低于肿瘤控制组，D*值显著高于肿瘤控制组（p＜0.05）。D*的AUC值为0.802，当D*值为38.85×10^-3mm²/s时，敏感度和特异度分别为77.4%和85.7%。单因素Kaplan-Meier生存曲线显示ADCmean、ADCmax、ADCmin、D、D*、f、DDC和α值在高低风险分组有显著性差异，p＜0.05。多因素Cox回归分析显示ADCmean和D*值与无进展生存期（PFS）独立相关（HR=0.125，p=0.001；HR=1.008，p=0.002）。

结论：IVIM-DWI模型的治疗前参数ADC、D、f值和D*值与喉、下咽鳞癌远期预后显著相关，ADC平均值、D*值是喉、下咽癌5年生存风险的独立预测因子。

第二部分

目的：利用体素内不相干运动-扩散加权成像（IVIM-DWI）双指数模型建立喉及下咽鳞状细胞癌患者放化疗后生存风险的预测模型。

材料与方法：回顾性研究45例喉或下咽鳞状细胞癌患者，所有患者均行治疗前IVIM-DWI检查，采用双指数模型测定参数图包括真扩散系数（D）、伪扩散系数（D*）、灌注分数（f）。随访5年，获得每位患者的生存数据。使用ITK-SNAP软件勾画感兴趣区，PyRadiomics软件包获得放射组学特征，并通过最小绝对收缩和选择算子回归建立放射组学评分公式，生成放射组学评分的诺模图，采用Kaplan-Meier法及Cox回归进行生存分析。

结果：患者治疗失败组31例，肿瘤控制组14例。放射组学评分由6个IVIM-DWI双指数模型特征组成，建立模型和诺模图。模型的AUC值为0.804（95%CI：0.647-0.913），模型的敏感性、特异性、准确性分别为93.6%、64.3%、84.4%。N分期和放射组学评分的Kaplan-Meier分析显示风险分层有显著性差异，放射组学评分的风险比（HR）为1.306（95%CI：1.082-1.577），p＜0.05。

结论：IVIM-DWI双指数模型图像的放射组学分析能够建立预测喉及下咽鳞癌患者放化疗后远期预后及生存风险的预测模型及诺模图，用于指导患者的个体化治疗。

第三部分

目的：探讨多期动态增强MRI（DCE-MRI）定量参数与喉及下咽鳞状细胞癌放化疗后患者远期预后及生存风险的关系。

材料与方法：分析2013年1月至2016年3月经我院诊治的88例喉及下咽鳞状细胞癌患者，标准放化疗两周前行DCE-MRI检查，原始图像经后处理测量定量参数包括容积转移常数（Ktrans）、速率常数（Kep）和血管外细胞外容积分数（Ve）值。患者随访5年，获得肿瘤复发进展、转移、生存情况等资料。比较各参数在不同预后组间的差异，绘制ROC曲线评估各定量参数预测喉及下咽癌患者放化疗后远期预后的效能，单因素Kaplan-Meier及多因素Cox回归分析影响无进展生存期（PFS）的因子。

结果：治疗失败组61例，肿瘤控制组27例。治疗失败组的疗前Ktrans值（1.19±1.08/min vs 0.60±0.37/min）及Kep值（2.49±3.43/min vs 1.05±0.56/min）高于肿瘤控制组，Ve值低于肿瘤控制组（0.52±0.27 vs 0.71±0.26），p＜0.05。Kep的曲线下面积（AUC）最大为0.747，当Kep值为1.15/min时，预测远期预后的敏感度和特异度为75.4%和63.0%。单因素Kaplan-Meier生存曲线示Ktrans、Ve、Kep在高低风险分组有显著性差异（p＜0.05），T分期、N分期及肿瘤分期无显著性差异。多因素Cox回归分析显示，Ktrans值和Ve值与无进展生存期（PFS）独立相关（HR=2.912，p=0.000；HR=0.12，p=0.000）。

结论：疗前DCE-MRI定量参数Ktrans、Kep、Ve值能够预测喉、下咽鳞癌放化疗后远期预后，Ktrans、Ve值是喉、下咽鳞癌放化疗后5年生存风险的独立预测因子。 

第四部分

目的：探讨多期动态增强MRI时间-信号强度曲线及半定量参数与喉及下咽鳞状细胞癌放化疗后患者远期预后及生存风险的关系。

材料与方法：分析2013年1月至2016年3月经我院诊治的89例喉及下咽鳞状细胞癌患者，在标准放化疗两周前行DCE-MRI检查，可测量时间-信号强度曲线（TIC）、半定量参数包括达峰时间（TTP）、正性增强积分（PEI）、最大上升斜率（MSI）、最大下降斜率（MSD）、信号增强比率（SER）、最大信号增强比率（SERmax）及注射造影剂后第28s、42s、56s、70s、84s、98s信号增强比率（SERtime）。随访5年，获得每位患者的生存资料。比较各参数在不同预后组间的差异、预测效能及生存分析。

结果：治疗失败组63例，肿瘤控制组26例。两组间T分期及肿瘤分期有统计学差异，N分期及TIC曲线类型无差异。1.5T机器测量的治疗失败组MSI值（106.67±30.39 vs 77.85±31.35）、SER（%）值（48.46±63.70 vs 16.87±19.30）、PEI（%）值（117.61±28.36 vs 94.28±29.83）高于肿瘤控制组，TTP（s）值低于肿瘤控制组（73.52±27.63 vs 91.03±20.49），3.0T机器的治疗失败组的疗前SER（%）值（8.84±4.37 vs 6.04±1.93）高于肿瘤控制组，p＜0.05，SERmax及SERtime无统计学差异。1.5T机器的MSI值的曲线下面积（AUC）为0.758，当MSI阈值为75.53时，预测远期预后的敏感度和特异度为84.2%和64.3%。单因素Kaplan-Meier生存曲线示T分期、1.5T机器的SER、MSI、PEI值及3.0T机器的SER值在高低风险分组有显著性差异（p＜0.05）。多因素Cox回归分析示1.5T机器的MSI值、3.0T机器的SER值在各自组内与无进展生存期（PFS）独立相关（HR=1.012，p=0.012；HR=1.12，p=0.009）。

结论：DCE-MRI半定量参数与喉、下咽癌放化疗远期预后相关，1.5T机器MSI值、3.0T机器SER值是喉、下咽癌放化疗后5年生存风险的预测因子。 

Part 1

Utility of three intravoxel incoherent motion diffusion weighted imaging (IVIM-DWI) models to predict prognosis and survival risk in laryngeal and hypopharyngeal squamous cell carcinoma patients after chemoradiotherapy

Abstract

Purpose: To investigate the predicted prognosis and survival risk power of three IVIM-DWI models in laryngeal and hypopharyngeal squamous cell carcinoma patients after chemoradiotherapy.

Materials and methods: 45 patients with laryngeal or hypopharyngeal squamous cell carcinoma were retrospectively enrolled. All patients had undergone pretreatment IVIM-DWI examination, then mean apparent diffusion coefficient (ADCmean), maximum ADC (ADCmax), minimum ADC (ADCmin) and ADCrange (ADCmax − ADCmean) by mono-exponential model, and true diffusion coefficient (D), pseudo diffusion coefficient (D*), perfusion fraction (f) by bi-exponential model, and distributed diffusion coefficient (DDC), diffusion heterogeneity index (α) by stretched exponential model were all measured. Survival data were followed up for 5 years.

Results: There were 31 cases in treatment failure group and 14 cases in local control group. Significantly lower ADCmean, ADCmax, ADCmin, D, f and higher D* values were presented in the treatment failure group than local control group (p < 0.05). D* got the greatest AUC of 0.802, with the sensitivity 77.4% and specificity 85.7% when D* was 38.85×10^-3 mm²/s. The survival curve of ADCmean, ADCmax, ADCmin, D, D*, f, DDC, and α values were found significant. Cox regression analysis showed ADCmean and D* were independently correlated with progression-free survival (PFS) (HR = 0.125, p = 0.001; HR = 1.008, p = 0.002, respectively).

Conclusion: The pretreatment IVIM-DWI parameters were significantly correlated with prognosis of laryngeal and hypopharyngeal squamous cell carcinoma after chemoradiotherapy, ADCmean and D* values were independent factors for survival risk prediction.

Part 2

Nomogram based on radiomics analysis of intravoxel incoherent motion diffusion weighted imaging (IVIM-DWI) bi-exponential model images: prediction of prognosis and survival risk in primary laryngeal and hypopharyngeal squamous cell carcinoma patients after chemoradiotherapy

Abstract

Purpose: To establish a prediction model by IVIM-DWI bi-exponential model for evaluating the survival risk of laryngeal and hypopharyngeal squamous cell carcinoma patients after chemoradiotherapy.

Material and Methods: 45 patients with laryngeal or hypopharyngeal squamous cell carcinoma were retrospectively enrolled. All patients had undergone pretreatment IVIM-DWI examination with parameters including true diffusion coefficient (D), pseudo diffusion coefficient (D*), perfusion fraction ( f ) measured by bi-exponential model. With 5 years of follow-up, survival data was obtained for patients. Radiomics features were obtained using the PyRadiomics package, and the radiomics score was established by least absolute shrinkage and selection operator regression. A nomogram of radiomics score was generated, and Kaplan-Meier analysis was used for risk stratification.

Results: 31 cases were assigned in the treatment failure group, 14 cases were local control group. The radiomics score consisted of 6 IVIM-DWI bi-exponential model features constructed the model and nomogram. The area under the curve (AUC) of the model was 0.804 (95% CI : 0.647-0.913), and sensitivity, specificity, accuracy of the model were 93.6%, 64.3%, and 84.4%. Kaplan-Meier analysis of N stage and radiomics score showed significant difference in risk stratification. The hazard ratio of radiomics score was 1.306 (95% CI : 1.082-1.577), p < 0.05.

Conclusion: Radiomics analysis of IVIM-DWI bi-exponential model images could establish a prediction model and normogram to predict the long-term prognosis and survival risk of patients with laryngeal and hypopharyngeal squamous cell carcinoma after chemoradiotherapy, which can be used to guide the individualized treatment of patients.

Part 3

Predictive value of dynamic contrast-enhanced MRI (DCE-MRI) quantitative parameters in long-term prognosis and survival risk of laryngeal and hypopharyngeal cancers after chemoradiotherapy

Abstract

Purpose: To investigate the relationship between quantitative parameters of dynamic contrast-enhanced MRI and long-term prognosis of patients with laryngeal and hypopharyngeal squamous cell carcinoma after chemoradiotherapy.

Material and Methods: A total of 88 patients with laryngeal and hypopharyngeal squamous cell carcinoma treated from January 2013 to March 2016 were analyzed. DCE-MRI was performed two weeks before standard chemoradiotherapy. Quantitative parameters including volume transfer constant (Ktrans), rate constant (Kep) and extravascular extracellular volume fraction (Ve) were measured. Patients were followed up 5 years to obtain the survival data.

Results: 61 cases were treatment failure group and 27 cases were local control group. The Ktrans (1.19±1.08/min vs 0.60±0.37/min) and Kep (2.49±3.43/min vs 1.05±0.56/min) in the treatment failure group were higher than those in the tumor control group, Ve was lower than tumor control group (0.52±0.27 vs 0.71±0.26), p<0.05. Kep got the greast AUC of 0.747, and when Kep was 1.15/min, the sensitivity and specificity for predicting long-term prognosis were 75.4% and 63.0%, respectively. Univariate Kaplan-Meier survival curve showed that Ktrans, Ve and Kep had significant differences in high and low risk groups (p<0.05), while there were no significant differences in T stage, N stage and tumor stage. Multivariate Cox regression analysis showed that Ktrans and Ve were independently correlated with progression free survival (PFS) (HR=2.912, p=0.000; HR=0.12, p=0.000).

Conclusion: DCE-MRI quantitative parameters could predict the long-term prognosis of laryngeal and hypopharyngeal carcinoma after chemoradiotherapy. Ktrans and Ve values were independent factors for survival risk prediction.

Part 4

The predictive value of time-signal intensity curve and semi-quantitative parameters of dynamic contrast-enhanced MRI (DCE-MRI) in long-term prognosis of laryngeal and hypopharyngeal carcinomas after chemoradiotherapy

Abstract

Purpose: To investigate the predictive value of time-signal intensity curve (TIC) and semi-quantitative parameters of dynamic contrast-enhanced MRI in long-term prognosis of patients with laryngeal and hypopharyngeal squamous cell carcinoma after chemoradiotherapy.

Material and Methods: A total of 89 patients with laryngeal and hypopharyngeal squamous cell carcinoma treated from January 2013 to March 2016 were analyzed. DCE-MRI was performed two weeks before standard chemoradiotherapy. Semi-quantitative parameters including time to peak (TTP), maximum slope of increase (MSI), maximum slope of decrease (MSD), positive enhancement integra (PEI), signal enhancement ratio (SER), SERmax and SERtime at 28s, 42s, 56s, 70s, 84s, 98s after injection of contrast agent, and TIC types were measured. Patients were followed up 5 years to obtain the survival data.  

Results: There were 63 cases in the treatment failure group and 26 cases in local control group. There were significant differences in T stage and tumor stage, no difference in N stage and TIC type between the two groups. MSI value (106.67±30.39 vs 77.85±31.35), SER (%) value (48.46±63.70 vs 16.87±19.30), PEI (%) value (117.61±28.36 vs 94.28±29.83) of treatment failure group measured by 1.5T machine were higher than those of tumor control group, TTP (s) was lower than that of tumor control group (73.52±27.63 vs 91.03±20.49), SER (%) of 3.0T machine was higher in treatment failure group (8.84±4.37 vs 6.04±1.93), p<0.05. Whereas SERmax and SERtime had no statistical difference. MSI value of 1.5T machine got the greast AUC of 0.758, and when MSI was 75.53, the sensitivity and specificity for predicting long-term prognosis were 84.2% and 64.3%. Univariate Kaplan-Meier survival curve showed that there were significant differences in T stage, SER, MSI, PEI values of 1.5T machine and SER of 3.0T machine (p<0.05). Multivariate Cox regression analysis showed that MSI of 1.5T machine and SER of 3.0T machine were independently correlated with progression free survival (PFS) (HR=1.012, p=0.012; HR=1.12, p=0.009).

Conclusion: DCE-MRI semi-quantitative parameters were correlated with long-term prognosis, and MSI value of 1.5T machine, SER value of 3.0T machine were predictors of 5-year survival risk of laryngeal and hypopharyngeal squamous cell carcinoma after chemoradiotherapy.