基于自适应云模型的多模态脑部图像融合方法

Fusion Method of Multi-model Brain Images Based on Adaptive Cloud Model

多模态医学图像融合通过提取并综合不同模态的医学图像信息,获得对病灶部位更加清晰、全面、准确、可靠的图像描述,为医生对疾病的诊断和合理治疗方案的制定提供可靠的依据。云模型理论是认知科学研究的新成果,具有兼顾随机性和模糊性的优点,在图像融合中的应用较少。借助云模型理论将来自不同模态的MRI(核磁共振成像)脑部图像、MRI与PET(正电子发射断层成像)、MRI与SPECT(单光子发射断层成像)脑部图像进行融合。首先,根据脑部图像自身的灰度直方图特征,对灰度直方图进行拟合;然后,由拟合曲线的谷值点划分区间并通过逆向云发生器自适应地生成云模型;最后,设计云推理规则,得到融合后的图像。实验结果表明,相比传统融合方法,所提方法融合后的图像脑部特征更清晰,激活区域更明显,在主观融合效果与客观评价指标方面均有很大的提高。

Through extracting and combining medical image information from different models of images,multi-model medical image fusion can obtain more clear,comprehensive,accurate and reliable image description on lesion site,and provide reliable basis for doctors to diagnosis of the disease and formulate reasonable treatment plan.Cloud model is a recently proposed theory in cognitive science,it has the advantage of taking the randomness and fuzziness into account,and has less application in image fusion at present.The paper introduced a method to fuse multi-model brain images such as two types MRI（Magnetic Resonance Imaging）brain image,MRI and PET（Positron Emission Tomography）,MRI and SPECT（Single-Photon Emission Computed Tomography）using cloud model theory.Three steps are included in the proposed fusion method.At first,the histogram of input brain image with a smooth curve using the high-order spline function is fitted.Then the intervals in line with the valley point of the fitted curve are divided and cloud model is generated adaptively through reverse cloud generator.At last,cloud reasoning rules are designed and the fused image is gotten.The experimental results show that the characteristics of fused brain images gotten by the proposed method are clearer and the active regions are more significant than existing methods.The proposed method shows great improvement in both subjective effect and objective evaluation.