Multiresolution Autofocusing for Automated Microscopy

用于自动显微镜的多分辨率自动对焦

• 批准号: 6833120
• 负责人: QIANG WU
• 金额: $ 42.78万
• 依托单位: ADVANCED DIGITAL IMAGING RESEARCH, LLC
• 依托单位国家: 美国
• 财政年份: 2002
• 资助国家: 美国
• 起止时间: 2002-05-15 至 2006-08-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/6833120
• 关键词: artificial intelligence; bioimaging /biomedical imaging; biomedical automation; clinical research; computer program /software; computer system design /evaluation; digital imaging; flow cytometry; fluorescence microscopy; fluorescent in situ hybridization; human data; image enhancement; image processing; mathematical model

DESCRIPTION (provided by applicant): This project will further develop and refine an innovative digital auto-focus technology for automated microscopy. Auto-focusing is essential to automated microscope imaging. Currently available techniques rely on various algorithms of focus computation at a single image resolution and suffer from inherent performance limitations, which affect their success and utilization in clinical and research applications. Auto-focusing for fluorescence microscopy, for example, represents a serious challenge to existing methods for desired accuracy, reliability and speed since in this case the images have very low signal-to-noise ratio and narrow depth-of-fields while specimen exposure to fluorescent excitation must be minimized to avoid photo-bleaching and formation of undesirable substances such as free radicals and singlet oxygen. We propose a novel multi-resolution image analysis approach to microscope auto-focusing, based on the recently developed mathematical theory of wavelet transform. The new approach overcomes a number of inherent limitations of currently available techniques, and holds the promise to make the measurement of the microscope focus function and the detection of best-focus imaging position considerably more accurate, reliable, and fast. This innovative technology will significantly increase the ability and efficacy of automated microscope instruments for a wide range of clinical and research applications where a large number of specimens need to be imaged and quantitatively analyzed on a routine basis. During the Phase 1 project we investigated the feasibility of the proposed technology for fluorescence microscopy. We developed software to implement the algorithms for multi-resolution focus function measurements and for in-focus imaging position search. We evaluated the new approach in software simulation on a variety of sample image stacks of cytogenetic FISH specimens, and compared it with all current best-performing methods for microscope auto-focusing using the criteria of (1) accuracy, (2) range, (3) robustness, and (4) speed. The Phase 1 results suggest that, by using a proper wavelet-based auto-focus function, the new multi-resolution method significantly outperforms all competing methods in each of the aforementioned performance categories, and clearly exceeds the Phase 1 feasibility criteria. In the Phase 2 project, we will further develop, refine, integrate, and validate the new technology in real-time operation environment. We plan to build a prototype system with multi-resolution auto-focusing capabilities for both fluorescence and bright-field microscope imaging. We will evaluate the system extensively for a variety of applications including genetics, pathology, and cytology. We will beta test the new system and technology in routine clinical laboratory environment and optimize the technology as end user input and feedbacks are gathered. Once fully developed and qualified, this new technology will be patented and incorporated into future IRIS automated imaging cytometry instruments. It will also be made commercially available to Applied Imaging Corporation and other manufacturers of automated microscope instruments through licensing agreements and partnerships.

描述（由申请人提供）：该项目将进一步开发和完善用于自动显微镜的创新数字自动对焦技术。自动对焦对于自动显微镜成像至关重要。目前可用的技术依赖于单一图像分辨率下的焦点计算的各种算法，并且受到固有的性能限制，这影响了它们在临床和研究应用中的成功和利用。例如，荧光显微镜的自动对焦对现有方法的所需精度、可靠性和速度提出了严峻的挑战，因为在这种情况下，当样本暴露在必须最大限度地减少荧光激发，以避免光漂白和形成不良物质，例如自由基和单线态氧。我们基于最近发展的小波变换数学理论，提出了一种新颖的显微镜自动聚焦多分辨率图像分析方法。新方法克服了当前可用技术的许多固有限制，并有望使显微镜焦点功能的测量和最佳焦点成像位置的检测变得更加准确、可靠和快速。这项创新技术将显着提高自动化显微镜仪器的能力和功效，适用于需要对大量标本进行常规成像和定量分析的广泛临床和研究应用。在第一阶段项目中，我们研究了所提出的荧光显微镜技术的可行性。我们开发了软件来实现多分辨率聚焦函数测量和对焦成像位置搜索的算法。我们在细胞遗传学 FISH 标本的各种样本图像堆栈上评估了软件模拟的新方法，并使用 (1) 准确性、(2) 范围、( 3）稳健性，以及（4）速度。第一阶段的结果表明，通过使用适当的基于小波的自动对焦功能，新的多分辨率方法在上述每个性能类别中都显着优于所有竞争方法，并且明显超出了第一阶段的可行性标准。在第二阶段项目中，我们将进一步开发、完善、集成并在实时操作环境中验证新技术。我们计划构建一个具有多分辨率自动对焦功能的原型系统，用于荧光和明场显微镜成像。我们将广泛评估该系统的各种应用，包括遗传学、病理学和细胞学。我们将在常规临床实验室环境中对新系统和技术进行 Beta 测试，并根据收集的最终用户输入和反馈来优化技术。一旦完全开发并合格，这项新技术将获得专利并纳入未来的 IRIS 自动成像细胞计数仪器中。它还将通过许可协议和合作伙伴关系向应用成像公司和其他自动显微镜仪器制造商提供商业化服务。