基于非下采样剪切波变换的PET/SPECT和MR图像融合

PET/SPECT and MR Image Fusion Based on Nonsubsampled Shearlet Transform

目的针对PET/SPECT伪彩色图像空间分辨率低而MR灰度图像空间分辨率高的融合问题,探究一种新方法来提升PET/SPECT和MR图像的融合质量。方法对待融合图像进行非下采样剪切波变换(nonsubsampled shearlet transform,NSST),得到各自的低频子带和高频子带,然后采用绝对值取大规则去融合高频子带,采用本文提出的基于Haar小波的能量(Haar wavelet based energy,HWE)规则去融合低频子带,最后对融合后的高、低频分量进行逆NSST从而获得融合图像。采用40组脑PET/SPECT和MR图像对本文所提出方法进行测试,并与其他典型的融合方法进行比较。结果本文方法与基于DTCWT,NSCT,NSCT-SR和NSST的融合方法进行测试,结果表明,本文方法在4个指标上最优,1个指标上次优。结论本文方法在PET/SPECT和MR图像的融合中具有优越性,可提升此类医学影像在辅助临床诊断和放疗手术计划中的精准性。

Objective Since PET/SPECT images are pseudo-color and have low spatial resolution while MR images are gray and have high spatial resolution,it is difficult but necessary to study a novel fusion method to improve the fusion quality of PET/SPECT and MR images. Methods The nonsubsampled shearlet transform（ NSST） was performed on the input images to acquire the low frequency subbands and the high frequency subbands. Then the high frequency subbands were merged by the absolute-maximum rule while the low frequency subbands were merged by the proposed Haar wavelet-based energy（ HWE） rule. Finally the fused image was obtained by performing the inverse NSST on the merged subbands. The proposed method was tested on forty pairs of PET/SPECT and MR images and compared with several popular fusion methods. Results The proposed method is tested with the fusion method of other DTCWT,NSCT,NSCT-SR and NSST. The results show that the proposed method is optimal on 4 indices and the 1 index is second. Conclusion The proposed method has advantage in fusing PET/SPECT and MR images,which can improve the precision of clinical diagnosis and radiotherapy treatment planning aided by these medical images.