A comparative analysis of several transformations for enhancement and segmentation of magnetic resonance images

The authors compare performance of several transformations, applied to magnetic resonance image (MRI) scene sequences for image enhancement and segmentation. Comparisons are made for principal component analysis, matched, modified-matched, maximum contrast, eigenimage, target point, ratio, log-ratio, and angle-image transformations. Signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), segmentation of a desired feature (SDF), and correction for partial volume averaging effects (CPV) are used as performance measures. For comparison, simplified analytical expressions for SNRs and CNRs of transformed images are derived, and CPV by a transformation is studied. Properties of transformations are investigated through their applications to simulation, phantom, and brain studies; advantages and weaknesses are discussed. The authors´ conclusions are: (i) principal component analysis, matched, and angle-image are best for SNR. Improvement; (ii) maximum contrast is best for CNR improvement; (iii) eigenimage and target point are best for SDF; and (iv) eigenimage is best for simultaneous SNR and CNR improvements, SDF, and CPV