Grabe N, Roth W, Foersch S
Nationales Centrum für Tumorerkrankungen und Medizinische Onkologie, Universitätsklinikum Heidelberg, Heidelberg, Deutschland.
Tissue Imaging & Analysis Center, Universität Heidelberg, Im Neuenheimer Feld 267, 69120, Heidelberg, Deutschland.
Pathologe. 2018 Nov;39(6):539-545. doi: 10.1007/s00292-018-0540-9.
Immuno-oncology requires objective and standardized methods for measuring immune cell infiltrates for therapy selection and clinical trials.
Current approaches in applying digital pathology in immuno-oncology and developments in computational image analysis were analyzed.
Since 2008, digital pathology has had an ever increasing importance in immuno-oncology. It is currently the only technology allowing the systematic and cost-effective quantitative spatial immune-profiling of patients. The analysis of immunological biomarkers requires integrated staining and image analysis strategies from single- to multistain on slide stacks. Statistical limits of the hypothesis to be tested have to be taken into account. Digital image analysis opens a new technological role for pathology in immuno-oncology and thereby serves as a key technological driver.
Digital pathology delivers objective and quantitative data on the tumor microenvironment. But currently, a fully automatic, high-throughput analytics capability is still missing. Deep learning is the remedy for this, as it improves image analysis with increasing data availability. This requires the creation of systematic data collections but will in the end deliver standardized and automatic immunological analyses.
免疫肿瘤学需要客观且标准化的方法来测量免疫细胞浸润情况，以用于治疗选择和临床试验。
分析了当前在免疫肿瘤学中应用数字病理学的方法以及计算图像分析的发展情况。
自2008年以来，数字病理学在免疫肿瘤学中的重要性与日俱增。它是目前唯一能够对患者进行系统且具有成本效益的定量空间免疫分析的技术。免疫生物标志物的分析需要从玻片堆叠上的单染色到多染色的综合染色和图像分析策略。必须考虑待检验假设的统计限度。数字图像分析为免疫肿瘤学中的病理学开启了新的技术角色，从而成为关键的技术驱动力。
数字病理学可提供关于肿瘤微环境的客观定量数据。但目前仍缺乏全自动、高通量的分析能力。深度学习是解决这一问题的方法，因为随着数据可用性的增加，它可改善图像分析。这需要创建系统的数据收集，但最终将实现标准化和自动化的免疫分析。
