Àá½Ã¸¸ ±â´Ù·Á ÁÖ¼¼¿ä. ·ÎµùÁßÀÔ´Ï´Ù.
KMID : 1144120140040040414
Biomedical Engineering Letters
2014 Volume.4 No. 4 p.414 ~ p.424
Multimodality medical image fusion based on new features in NSST domain
Ganasala Padma

Kumar Vinod
Abstract
Purpose: Multimodality medical image fusion supports better visualization of complimentary information given by different medical imaging modalities. This helps the radiologist for the precise diagnosis of disease and treatment planning. Main purpose of this research is to design a unified frame work for fusion of different anatomical imaging modalities and fusion of functional image with an anatomical image.

Methods: A novel image fusion framework utilizing new features in Nonsubsampled Shearlet Transform (NSST) domain is proposed for fusion of anatomical images. The source images are represented in low-frequency (LF) and high-frequency (HF) sub-bands using NSST. LF sub-bands are combined by fusion rule based on sum of variation in squares. HF fusion rule is formulated based on two different features. Inverse NSST of fused sub-bands gives the fused image. Further, this framework is utilized for fusion of functional and anatomical images in l-Alpha-Beta color space.

Results: The proposed image fusion framework is validated on nine sets of CT-MRI, and SPECT-MRI images of different diseases. It is compared with state of the art image fusion methods both quantitatively and qualitatively.

Conclusions: Visual analysis of CT-MRI image fusion results reveal that the fused images by proposed method retain the salient information of both CT and MRI images with more contrast than other methods. Fused SPECT-MRI images by proposed method presents anatomical details of MRI images without altering the functional content of SPECT images. However, spectral distortion is present in other methods. Quantitative comparison proved the superiority proposed method compared to other methods.
KEYWORD
Image fusion, Nonsubsampled shearlet transform, l-alpha-beta color space, MRI-PET fusion, CT-MRI fusion
FullTexts / Linksout information
Listed journal information