Related Experiment Videos
Spatial normalization of lesioned HMPAO-SPECT images
E A Stamatakis1, J T Wilson, D J Wyper
1Department of Psychology, University of Stirling, Stirling, FK9 4LA, Scotland. e.a.stamatakis@stir.ac.uk
Neuroimage
|September 14, 2001
Summary
Nonlinear alignment of brain SPECT images can cause distortions, especially with lesions. Linear affine transformations are safest for aligning images with lesions, and masking can mitigate nonlinear distortions.
Area of Science:
- Medical imaging
- Neuroimaging analysis
- Image processing
Background:
- Linear alignment of Single Photon Emission Computed Tomography (SPECT) images is reliable but may miss global brain shape variations.
- Nonlinear alignment methods offer improved matching by considering global brain shape but risk introducing distortions.
- Lesions in brain images can complicate the alignment process, potentially exacerbating distortions.
Purpose of the Study:
- To investigate the impact of nonlinear image alignment on SPECT images containing lesions.
- To quantify potential image distortions introduced by nonlinear alignment methods in the presence of lesions.
- To compare the effectiveness of different nonlinear alignment models and linear transformations for aligning SPECT images with lesions.
Main Methods:
- Artificial lesions were created in SPECT images to control for lesion characteristics and quantify distortions.
- Nonlinear alignment models from SPM96, SPM99, and AIR3.08 were evaluated.
- 12-point linear affine transformations were used as a baseline and comparison.
- Masking techniques were explored as a method to correct distortions.
Main Results:
- High degrees of nonlinearity in alignment models led to significant, unwanted image deformations.
- 12-point linear affine transformations proved to be the safest method for aligning SPECT images with lesions.
- Employing masking significantly improved the accuracy of alignment when nonlinear methods were used.
Conclusions:
- Nonlinear alignment methods, while potentially improving matching, introduce significant risks of image distortion in the presence of brain lesions.
- Linear affine transformations are recommended for robust alignment of SPECT images with lesions.
- Masking is an effective strategy to mitigate distortions caused by nonlinear alignment techniques.