Please use this identifier to cite or link to this item:
Title: Partial k-space MRI reconstruction using a modified Homodyne approach
Authors: Abche, Antoine 
Yaacoub, Fadi
Karam, Elie 
Affiliations: Department of Electrical Engineering 
Department of Electrical Engineering 
Keywords: Medical image processing
Biomedical MRI
Image reconstruction
Mean square error methods
Issue Date: 2011
Publisher: IEEE
Part of: Signal Processing Algorithms, Architectures, Arrangements, and Applications Conference Proceedings
Start page: 1
End page: 6
Conference: Signal Processing Algorithms, Architectures, Arrangements, and Applications SPA 2010 (23-25 Sept. 2010 : Poznan, Poland) 
A drawback of the conventional Homodyne approach which is originated from the communication theory, is that the reconstructed image acquired by the Magnetic Resonance Imaging (MRI) modality is blurred. Thus, a new homodyne based-approach is proposed in order to improve the quality of the MRI image. Consequently, the segmentation of boundaries of a particular organ (or any structure), a crucial step for medical diagnosis in the estimation of its size as well as the computation of various functional parameters is improved. The proposed approach is evaluated visually by segmenting the MR brain images using an active contour "snake" model as well as by subtracting the reconstructed images from the original image. Also, the evaluation is performed quantitatively using different criteria (the Performance Test and the Root Mean Square) for various percentage of the collected k-space. The proposed approach is compared with other MRI Reconstruction techniques such as the Homodyne approach, the Zero-filling approach, the POCS approach and the conjugate Synthesis approach. Consequently, its performance is much better than the indicated conventional techniques.
Ezproxy URL: Link to full text
Type: Conference Paper
Appears in Collections:Department of Electrical Engineering

Show full item record

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.