Novel Image Texture Analysis for Advanced Structure Characterization

先进结构表征的新型图像纹理分析我的研究旨在开发，扩展和验证用于先进结构表征的新型信号处理和分析算法。目前，我正在开发新技术，以提高对生物主体的磁共振成像（MRI）组织特性的评估。然而，我的方法在不同的领域有更广泛的应用，包括地球科学、物理学和工程学。傅里叶变换构成了许多成像方式的基石，包括核磁共振成像和计算机断层扫描（CT）。MR数据最初在频域获得，然后使用傅里叶反变换重建。因此，磁共振图像可以使用先进的、基于傅里叶的分析方法来处理，在不丢失信息的情况下增强对组织结构的识别。在过去的几年里，我一直致力于开发和评估先进的方法来检测图像强度的局部模式的细微变化，即MR“纹理”。纹理图案通常太微妙，即使是训练有素的观察者也无法察觉，因此需要数学算法。我的试验数据表明，纹理分析是组织损伤和修复的敏感指标。在这个建议中，我的目标是开发新的纹理分析途径，以提高加拿大人的健康结果。特别是，我建议研究纹理光谱成像和纹理各向异性，以提高纹理分析的灵敏度和特异性。这个项目的成果可能会提高我们在不修改现有成像协议的情况下评估组织结构的能力。

Novel Image Texture Analysis for Advanced Structure CharacterizationMy research seeks to develop, extend and validate novel signal processing and analysis algorithms for advanced structure characterization. Currently I am developing new techniques to improve the evaluation of tissue properties from magnetic resonance imaging (MRI) of biological subjects. However, my approaches have broader applications in diverse fields including geoscience, physics, and engineering.The Fourier transform forms the cornerstone of many imaging modalities including MRI and computer tomography (CT). The MR data are initially acquired in the frequency domain and are then reconstructed using the inverse Fourier transform. Therefore, MR images can be processed using advanced, Fourier-based analysis methods to enhance identification of tissue structure without loss of information.Over the past few years I have been engaged in developing and evaluation of advanced methods to detect subtle alterations in the local pattern of image intensity, namely, MR 'texture'. Texture patterns are generally too subtle to be perceived even by trained observers, thus mathematical algorithms are required. My pilot data demonstrate texture analysis to be a sensitive measure of tissue injury and repair.In this proposal, I aim to develop new avenues of texture analysis to enhance health outcomes of Canadians. In particular, I propose to investigate texture spectral imaging and texture anisotropy to improve the sensitivity and specificity of texture analysis. Output from this project may advance our capacity to evaluate tissue structure without modifying existing imaging protocols.