超声光散射成像对早期乳腺肿瘤的自适应检测方法研究

Ultrasound-guided Diffuse Optical Tomography (US-DOT) has attracted growing interest as an emerging method for the early detection of breast cancer. However, the performance of current approaches has been limited, firstly, since individual differences of mammary tissue have not been considered and, secondly, because noise caused by the human-machine interfacehas not been fully compensated. Thus overall results are subject-dependent and sensitive to user technique. Hither to the adaptive abilities of available systems have not been sufficiently robust, which has limited the development of US-DOT in the field of noninvasive detection of breast tumours. In recognition of the problems mentioned above, an adaptive method for early-stage breast tumour detection will be studied in the project, including the following aspects. Firstly, a theoretical basis for tumour optical detection and imaging will be developed through the creation of a dual-layer model. The correlation between tissue optical properties and geometry of the chest wall-breast interface will be thoroughly researched. An adaptive model of breast structure will also be constructed based on localization provided by ultrasonic phased arrays. We will establish the relationship between probe compression force and the breast tissue optical properties. By means of gradient experiments, an adaptive model of human-machine interface noise will be established. Finally, we will develop and utilise a nonlinear signal reconstruction method based on multifunctional sensing in view of the MIMO character of the adaptive models, and this will be able to boost the detection precision and flexibility as well as to decrease the computational complexity. The results emerging from this project will lead to a deeper understanding of the critical features of the US-DOT approach for adaptive detection and imaging for early-stage breast cancer and will, thereby, strengthen the scientific and technological basis of future research and clinical use of this method.

超声光散射成像在乳腺肿瘤的早期诊断中具有重要的应用价值。但是现有测量方法往往忽略了乳腺组织结构的个体差异及人机接口噪声，导致检测结果对测量环境敏感，检测系统的自适应能力较差，制约了超声光散射成像技术在乳腺肿瘤无创检测领域的深入研究和拓展。针对上述问题，本项目研究一种早期乳腺肿瘤的自适应检测方法，研究内容包括：探索双层模型的肿瘤光学检测和成像原理，研究乳腺-胸壁交界面的几何参数与组织光学特性的相关性，依据超声相控定位技术，构建乳腺组织结构的自适应模型；揭示探头作用力与乳腺组织光学特性的耦合规律，通过梯度实验，建立人机接口噪声的自适应校准模型；针对乳腺肿瘤自适应检测模型的多输入多输出特性，研究基于多功能传感理论的非线性信号重构方法，提高检测的准确度和敏捷度，降低计算复杂度。本项目研究成果有望形成对早期乳腺癌自适应检测和成像技术的有力支撑。

在本项目的资助下，课题组设计并制作了一套近红外肿瘤非侵入检测和成像系统的样机，包括阵列式多波长近红外传感探头和信号处理装置，该样机通过对比正常和癌变乳腺组织的吸收系数，实现肿瘤位置的定位与成像，仿体实验验证了该样机的有效性。此外，为了抑制非目标参量对目标参量的交叉影响，消除或者减弱交叉灵敏特性，课题组进行了多功能传感信号重构算法研究，并将之用于肿瘤检测中血氧浓度的测量环节，进一步提高了检测的精度。本项目在理论和实践方面都取得了一定的成果：其中，理论方面，提出了近红外光谱层析成像方法和实现流程，可用于指导后续研究工作，也可以为同领域其他研究工作者提供一定的理论依据；在实践方面，本项目的研究成果有望转化为乳腺恶性肿瘤检测的医疗设备，从而解决目前乳腺癌普查过程中存在的诊断费用过高、辐射危害大、侵入式检测等问题。总而言之，本项目的研究成果在理论研究和实际应用方面都具有深远的科学意义和广阔的应用前景。