目的:设计一款基于图像处理的医学影像处理平台,可对肿瘤影像进行定量分析。方法:系统设计模块包括医学数字成像和通信解读及图像载入模块、图像显示模块、感兴趣区域勾画模块、感兴趣体积三维展示模块、可视化模块共5个模块。系统可自动计算出勾画区域内的定量指标,这些特征可用来进行进一步的肿瘤分化检测、肿瘤特性评估、治疗监控以及评估预后。在完成系统设计后,进行15例宫颈癌病人分化程度的定量分析对比实验。导入扫描得到的磁共振扩散加权成像图以及表观扩散系数图进行分析,得到中高分化和低分化组织的ADC均值ADC5%、、ADC10%、ADC25%、ADC50%、ADC75%、ADC90%、偏度、峰度、熵10个参数,并进行统计学分析,以评估参数对于区分中高分化组织与低分化组织的效能。结果:肿瘤中高分化组织的ADC75%、ADC90%两项参数显著高于低分化组织(P〈0.05)。本次实验还通过ROC分析得到区分中高分化组织和低分化组织各参数的阈值(P〈0.05)。结论:该平台是创新型设计,具有定量挖掘出医学影像中具体参数信息、操作灵活、简单等特点。
Objective To design a medical image processing platform based on image processing for quantitative analysis of tumor images. Methods The designed image processing platform consisted of separate modules for medical digital imaging and communication interpretation, image loading, image display, delineation of the regions of interest (ROI), volume of interest display and visualization. The system automatically calculated the quantitative features from the ROI for further tumor differentiation evaluation, tumor characteristics assessment, treatment monitoring and prognosis evaluation. Using this system, 15 patients with cervical cancer were examined for quantitative assessment of the differentiation of the tumors. The magnetic resonance diffusion weighted images and apparent diffusion coefficient (ADC) images were input into the system to obtain the ADC ADC5%, ADC10%, ADC25%, ADC50%, ADC75%, ADC90%, skewness, kurtosis, and entropy of the tumor tissues. SPSS software was used to evaluate the effectiveness of the parameters to distinguish the poorly differentiated tumors from the well or moderately differentiated ones. Results The ADC75% and ADC90% of well or moderately differentiated tumors were significantly higher than those of poorly differentiated tumors (P〈0.05). The thresholds of the parameters for distinguishing the poorly from well or moderately differentiated tumors were obtained by ROC analysis (P〈0.05). Conclusion With flexible and convenient operation and an innovative design, this medical image processing platform allows for quantitative analysis of specific parameters in medical images.