Bazant-Hegemark Florian, Stone Nicholas
Cranfield Health, Cranfield University at Silsoe, Silsoe, Bedfordshire, UK.
Lasers Med Sci. 2009 Jul;24(4):627-38. doi: 10.1007/s10103-008-0615-6. Epub 2008 Oct 21.
The native contrast of optical coherence tomography (OCT) data in dense tissues can pose a challenge for clinical decision making. Automated data evaluation is one way of enhancing the clinical utility of measurements. Methods for extracting information from structural OCT data are appraised here. A-scan analysis allows characterization of layer thickness and scattering parameters, whereas image analysis renders itself to segmentation, texture and speckle analysis. All fully automated approaches combine pre-processing, feature registration, data reduction, and classification. Pre-processing requires de-noising, feature recognition, normalization and refining. In the current literature, image exclusion criteria, initial parameters, or manual input are common requirements. The interest of the presented methods lies in the prospect of objective, quick, and/or post-acquisition processing. There is a potential to improve clinical decision making based on automated processing of OCT data.
在致密组织中，光学相干断层扫描（OCT）数据的固有对比度可能对临床决策构成挑战。自动数据评估是提高测量临床效用的一种方法。本文评估了从结构性OCT数据中提取信息的方法。A扫描分析可用于表征层厚度和散射参数，而图像分析则适用于分割、纹理和散斑分析。所有全自动方法都结合了预处理、特征配准、数据缩减和分类。预处理需要去噪、特征识别、归一化和细化。在当前文献中，图像排除标准、初始参数或手动输入是常见的要求。所提出方法的意义在于客观、快速和/或采集后处理的前景。基于OCT数据的自动处理，有可能改善临床决策。
