Edge Retention based CT and MRI fusion using decomposition techniques NSST and SWT

A hybrid fusion technique is proposed based on NSST and SWT decomposition techniques for MRI and CT images. Firstly, the medical images undergo decomposition of low-frequency and high-frequency components using Non-subsampled shearlet transform (NSST). Then SWT is implemented for further decomposition of sub-bands obtained. Edge preservation techniques such as phase congruency and guided filter are used to extract the information related to the edges present in the input source images into fused images. The evaluation of the experimental results exhibits that the resolution, clarity and sharpness of the obtained fused image has been improved. This is in turn, aids the doctors in locating the volume of tumour and affected portions effectively. This hybrid fusion method is compared with the existing method SIFT-DCNN-SIST for two brain tumour diseases: Fatal Stroke and Metastatic Bronchogenic. It is also compared with existing approaches such as NSST and SWT for the different brain tumour diseases such as hypertensive encephalopathy, multiple embolic infarctions and meningioma. Seven metrics such as MI, EN, VIFF, AG, Q
^AB/F
, SD, SF and Q
cb
are considered for the objective assessments. The experimental results exhibit a maximum increase of VIFF by 49%, AG by 66%, SD by 17%, SF by 83% and Q
cb
by 49% in the case of Hypertensive Encephalopathy disease, Q
^AB/F
by 25% in the case of Fatal Stroke disease when compared with existing techniques. These results show the fused image obtained is good by both subjective wise and objective wise.