McConnell Brain Imaging Center, Montreal Neurological Institute and Hospital McGill University, McGill University, Montreal, Quebec, Canada.
PLoS One. 2018 Mar 13;13(3):e0193636. doi: 10.1371/journal.pone.0193636. eCollection 2018.
Direct volume rendering has become an essential tool to explore and analyse 3D medical images. Despite several advances in the field, it remains a challenge to produce an image that highlights the anatomy of interest, avoids occlusion of important structures, provides an intuitive perception of shape and depth while retaining sufficient contextual information. Although the computer graphics community has proposed several solutions to address specific visualization problems, the medical imaging community still lacks a general volume rendering implementation that can address a wide variety of visualization use cases while avoiding complexity. In this paper, we propose a new open source framework called the Programmable Ray Integration Shading Model, or PRISM, that implements a complete GPU ray-casting solution where critical parts of the ray integration algorithm can be replaced to produce new volume rendering effects. A graphical user interface allows clinical users to easily experiment with pre-existing rendering effect building blocks drawn from an open database. For programmers, the interface enables real-time editing of the code inside the blocks. We show that in its default mode, the PRISM framework produces images very similar to those produced by a widely-adopted direct volume rendering implementation in VTK at comparable frame rates. More importantly, we demonstrate the flexibility of the framework by showing how several volume rendering techniques can be implemented in PRISM with no more than a few lines of code. Finally, we demonstrate the simplicity of our system in a usability study with 5 medical imaging expert subjects who have none or little experience with volume rendering. The PRISM framework has the potential to greatly accelerate development of volume rendering for medical applications by promoting sharing and enabling faster development iterations and easier collaboration between engineers and clinical personnel.
直接体绘制已经成为探索和分析 3D 医学图像的重要工具。尽管该领域已经取得了多项进展，但仍然存在一些挑战，例如生成一张突出感兴趣解剖结构、避免重要结构遮挡、提供直观的形状和深度感知，同时保留足够上下文信息的图像。尽管计算机图形学社区已经提出了几种解决方案来解决特定的可视化问题，但医学成像社区仍然缺乏一种通用的体绘制实现，可以解决广泛的可视化用例，同时避免复杂性。在本文中，我们提出了一个名为可编程光线积分着色模型（Programmable Ray Integration Shading Model，简称 PRISM）的新开源框架，该框架实现了一种完整的 GPU 光线投射解决方案，其中光线积分算法的关键部分可以被替换，以产生新的体绘制效果。一个图形用户界面允许临床用户轻松地尝试来自开放数据库的现有渲染效果构建块。对于程序员，该界面允许实时编辑块内的代码。我们表明，在其默认模式下，PRISM 框架生成的图像与 VTK 中广泛采用的直接体绘制实现生成的图像非常相似，帧率相当。更重要的是，我们通过展示如何在 PRISM 中实现几种体绘制技术，只需几行代码，证明了该框架的灵活性。最后，我们通过 5 名具有或几乎没有体绘制经验的医学成像专家进行的可用性研究，展示了我们系统的简单性。PRISM 框架通过促进共享、实现更快的开发迭代以及工程师和临床人员之间更容易的协作，有可能极大地加速医学应用中的体绘制开发。
