面向乳腺癌早期诊断的三维多尺度定量光声成像方法的研究

A safe, effective and economic imaging approach is needed for early diagnosis of breast cancer due to the limitations of current clinical imaging techniques. As a promising hybrid imaging modality, photoacoustic imaging which retains the desired high optical contrast and sensitivity while providing much improved spatial resolution, has the potential to match this requirement, since it can provide high resolution and contrast level images for optical absorption property in biomedical tissues. In addition, photoacoustic imaging may provide accurate functional information through a finite element based quantitative reconstruction algorithm combining with the multi-spectral imaging technique. However, so far photoacoustic imaging can only present two dimensional single scale recovered images, which is far from meeting the demand of early diagnosis of breast cancer. Therefore, in this project, by incorporating the computer science into the biomedical photonics, we propose to develop a new three dimensional quantitative photoacoustic reconstruction algorithm based on finite element method and the GPU parallel strategy, as well as the related multiple scale imaging system. Thus, high resolution, high contrast level, and high accuracy three dimensional structural and functional photoacoustic images of both the tumor and the surrounding blood vessels inside human breast will be recovered simultaneously, and a new approach for early diagnosis of breast cancer and examining of the characteristic of the breast tumor will also be established. Testing and evaluating for this approach will be applied via different types of well designed experiments thereafter. This study will lay a solid foundation for the clinic applications of the photoacoustic imaging technique in the diagnosis and treatment in breast cancer.

当前的医学影像方法中缺乏一种安全、有效、经济的技术来对乳腺癌进行早期诊断。作为一种新型的混合成像方法，光声成像融合了光学成像高对比度与超声成像穿透力强的优点，能提供乳房等生物组织的高对比度、高分辨率的光学吸收特性信息，而通过基于有限元方法的定量重建算法和多光谱技术，还可提供组织中准确的功能性信息。但目前的定量光声成像技术还只能提供二维单尺度重建图像，远远无法满足乳腺癌早期诊断方面的需求。为此，本项目通过计算机技术和生物医学光子学的融合，研发出一套基于并行化加速策略的光声成像的新型三维定量重建算法及配套的多尺度成像系统，对人体乳房的肿瘤及其周边血管网络同时进行高分辨率、高对比度、高准确性的三维结构与功能成像，从而建立起一种可用于乳腺癌早期诊断和肿瘤的良恶性甄别的方法，并通过实验研究对该方法进行测试与评估，为光声成像技术在乳腺癌诊断与治疗等临床应用方面奠定基础。

光声成像融合了光学成像高对比度与超声成像分辨率高和穿透力强的优点，能提供乳房等生物组织的高对比度、高分辨率光学吸收特性信息，是一种有望对乳腺癌进行安全、有效、经济的早期诊断技术。本项目首先完成了三维定量光声成像有限元重建算法的编写；同时，完成了高性能数值计算平台的构建(算法加速率＞500)，并对算法性能进行了测试与验证；最后，通过双频聚焦超声换能器的研制和三维光声成像实验系统(横向和轴向分辨率分别为110 μm和800 μm)的研发，进一步通过实验研究对本项目所提技术进行了测试与评估，为光声成像技术在乳腺癌诊断与治疗等临床应用方面奠定了基础。