CT影像组学在胃肠道间质瘤中的应用进展

Application progress of CT radiomics in gastrointestinal stromal tumor

胃肠道间质瘤(gastrointestinal stromal tumor,GIST)是胃肠道最常见的间叶源性肿瘤,其生物学特征复杂,危险度不等,且不同危险度患者的治疗方法和预后差异较大,因此早期诊断及危险度评估对于该病的精准治疗至关重要。近年来,CT影像组学作为一种新兴的影像学技术,可以将传统的CT图像特征转变为大量数据,从而反映GIST的内在异质性,甚至与其基因表达特征相关联。该文回顾了CT影像组学在机器学习的助力下应用于GIST诊断、预测的研究进展。目前的CT影像组学不仅可用于GIST与其他胃部疾病的鉴别诊断,并且为GIST的危险度评估提供了新的方式,甚至可以根据CT影像图像进行病理分析及基因诊断,进而对其一线治疗效果及远期预后进行预测。目前CT影像组学通过与临床信息结合构建的多种预测模型在不同临床问题的具体实践中得到了良好的验证,展现了广阔的应用前景。但是在具体的临床应用过程中,样本数据采集及处理方式的不同、机器学习算法选择的差异、二维或三维图像的选择等,都影响到CT影像组学的具体效能,因此统一、规范的影像组学应用规则还有待建立。

Gastrointestinal stromal tumor(GIST)is the most common mesenchymal tumor in the gastrointestinal tract,with complex biological characteristics and varying risks,and the treatment methods and prognosis of patients with different risks are quite different;therefore,early diagnosis and risk assessment are crucial for its precision treatment.In recent years,CT radiomics,as an emerging imaging technology,can transform traditional CT image features into a large number of data,thereby reflecting the inherent heterogeneity of GIST and even correlating with its gene expression features.This paper reviews the research progress of CT radiomics in the diagnosis and prediction of GIST with the help of machine learning.The current CT radiomics can not only be used for the differential diagnosis of GIST and other gastric diseases,but also for the risk evaluation of GIST.Furthermore,pathological analysis and gene diagnosis can be performed based on CT images,and then the first-line treatment effect and longterm prognosis can be predicted.At present,various prediction models constructed by combination of CT radiomics and clinical information have been well verified in the specific practice of different clinical problems,showing broad application prospects.However,in the specific clinical application process,different methods of sample data collection and processing,differences in the selection of machine learning algorithms,and the selection of 2D or 3D images all affect the specific effectiveness of CT radiomics.Hence,unified and standardized application rules for radiomics has to be established.