Simulating medical applications of tissue optical property and shape imaging using open-source ray tracing software
Simulating medical applications of tissue optical property and shape imaging using open-source ray tracing software
复制标题
DOI：
10.1117/12.2576779
复制
发表时间：
2021-03
期刊：
影响因子：
--
通讯作者：
J. Crowley;George S. D. Gordon
中科院分区：
文献类型：
--
作者：
J. Crowley;George S. D. Gordon
食道癌和结肠癌的五年生存率分别为15%和63%。存活率低的部分原因是在内窥镜筛查过程中早期检测不佳，传统的内窥镜提供的组织属性信息不足以发现广泛的潜在肿瘤。改进胃肠道癌症的早期检测将极大地提高他们的五年存活率。空间频域成像(SFDI)是一种低成本的成像技术，可以测量肿瘤的吸收、散射和形状作为潜在的指标。已知特殊的吸收和散射特性与食道恶性肿瘤有关，而形状是结肠癌的重要指标。尽管已经开发了一系列研究和商业SFDI系统，但由于小型化、样品几何形状和照明条件的限制，将这些系统应用于体内临床应用具有挑战性。为了方便在这样的约束下设计新的SFDI系统，我们开发了一个基于开源3D建模软件Blender的SFDI成像系统模型。使用Blender的Cycle光线跟踪引擎，我们能够为许多不同的成像配置、样本几何形状和照明模式模拟一系列不同的散射和吸收系数。使用已建立的处理算法，我们可以在一系列模拟的体外和体内成像几何图形中恢复与临床肿瘤检测相关的吸收、散射和形状的地图。我们的系统能够方便地探索不同的光学配置和现实的照明条件，这将为未来紧凑型、低成本仪器的设计提供信息。
Oesophageal cancer and colon cancer have five year survival rates of 15% and 63% respectively. These low survival rates are due in part to poor early detection during endoscopic screening, with conventional endoscopes providing insufficient information about tissue properties to spot a wide range of potential tumours. Improving early detection of gastrointestinal cancers would dramatically increase their five year survival rates. Spatial Frequency Domain Imaging (SFDI) is a low-cost imaging technique that can measure absorption, scattering and shape as potential indicators of cancer. Specific absorption and scattering properties are known to be linked to malignancy in the oesophagus, and shape is an important indicator in colon cancer. Though a range of research and commercial SFDI systems have been developed, adapting these for in vivo clinical application is challenging due to constraints imposed by miniaturisation, sample geometry and illumination conditions. To facilitate design of novel SFDI systems under such constraints, we have developed a model of an SFDI imaging system built on the open-source 3D modelling software Blender. Using Blender’s Cycles ray-tracing engine, we are able to simulate a range of different scattering and absorption coefficients for a number of different imaging configurations, sample geometries and illumination patterns. Using established processing algorithms, we show we can recover maps of absorption, scattering and shape in a range of simulated ex vivo and in vivo imaging geometries with relevance to clinical detection of tumours. Our system enables accessible exploration of different optical configurations and realistic illumination conditions that will inform future design of compact, low-cost instruments.