Multi-Modality Medical Image Fusion using Discrete Wavelet Transform

Diagnosis and treatment of ailments require that precise information be obtained through various modalities of medical images such as Computed Tomography (CT), Positron Emission Tomography (PET), and Magnetic Resonance Imaging (MRI) etc. Often these techniques give some information regarding the ailment which is incomplete and ambiguous. In this scenario, image fusion gains utmost importance as the overall quality of scans can be improved. Thus, fusing various multi – modality medical images into a distinct image with more detailed anatomical information and high spectral information is highly desired in clinical diagnosis. In this work, MRI and PET images are preprocessed along with enhancing the quality of the input images which are degraded and non-readable due to various factors by using spatial filtering techniques like Gaussian filters. The enhanced image is then fused based on Discrete Wavelet Transform (DWT) for brain regions with different activity levels. The system showed around 80-90% more accurate results with reduced color distortion and without losing any anatomical information in comparison with the existing techniques in terms of performance indices including Average Gradient (AG) and Spectral Discrepancy (SD), when tested on three datasets - normal axial, normal coronal and Alzheimer's brain disease images.