MRI and PET/SPECT image fusion at feature level using ant colony based segmentation. (January 2019)

• Record Type: Journal Article
• Authors: Shahdoosti, Hamid Reza; Tabatabaei, Zahra
• Is Part Of: Biomedical signal processing and control. Volume 47(2019)
• Page Start: 63
• Page End: 74
• Publication Date: 2019-01
• Subjects: Medical image fusion -- Ant colony algorithm -- Object-based fusion -- Ensemble empirical mode decomposition; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2018.08.017
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: The proposed method uses an ant colony based segmentation map. This paper uses the ensemble empirical mode decomposition to fuse images. The proposed method preserves more functional and anatomical information. Abstract: Extracting salient features from the medical images and combining them by an appropriate algorithm are the key challenges of multimodal image fusion. The commonly used coefficient-wise fusion may also inject noise into the merged images. To tackle the problem, this paper proposes a new method of multimodal image fusion which makes use of a segmentation map given by the ant colony algorithm. Firstly, the proposed method applies the maximum selection rule in ensemble empirical mode decomposition (EEMD) domain to obtain a fusion map. Then, the proposed approach exploits the color information of the pseudo-color image (PET or SPECT) to find spatial regions of pixels belonging to the same object. This step gives the segmentation map. Finally, the proposed method uses the majority voting process to combine the results of the fusion map and the segmentation map. In fact, the majority voting process determines the winner in each region and scale. The EEMD transform is used to decompose images because it is an adaptive and fully data-driven multiscale transform, and the ant colony algorithm is used for segmentation because it can yield a near optimal segmentation solution. Experimental fusion results are presented on three medical image datasets. It is shown in experiments that the proposed scheme improves the fusion results and provides images with more spatial and color information, when compared to state-of-the-art methods.