Related Experiment Video
Updated: Jul 16, 2026
Mouse Models of Periventricular Leukomalacia
Published on: May 18, 2010
Linear normalization of MR brain images in pediatric patients with periventricular leukomalacia
Bart Machilsen1, Emiliano d'Agostino, Frederik Maes
1Laboratorium voor Neuropsychologie, K.U.Leuven Medical School, Leuven, Belgium.
Insights
Linear normalization of child brain images affected by periventricular leukomalacia (PVL) is feasible. Modifying initial parameters and removing background noise improved normalization success rates for these pediatric brain scans.
Area of Science:
- Neuroimaging
- Pediatric Radiology
- Medical Image Analysis
Background:
- Periventricular leukomalacia (PVL) is a common brain injury in premature infants, potentially causing structural abnormalities.
- Accurate spatial normalization of pediatric brain images is crucial for quantitative analysis and group comparisons.
- Linear normalization methods may face challenges with atypical brain structures common in conditions like PVL.
Purpose of the Study:
- To assess the feasibility of linear spatial normalization for pediatric brain images with structural abnormalities due to PVL.
- To evaluate the impact of modifications to the linear normalization process on success rates and accuracy.
- To determine if linear normalization can be reliably applied to brains affected by PVL.
Main Methods:
- Linear spatial transformation of T1-weighted MRI scans from 51 children (4-11 years) to the MNI template.
- Inclusion of control subjects and children with varying degrees of PVL and ventricular enlargement.
- Systematic evaluation of parameter adjustments: initial orientation, zoom, background masking, image smoothing, and template choice (MNI vs. pediatric).
Main Results:
- Normalization failure rates were reduced by adjusting initial zoom parameters and removing background noise.
- Overall algorithm performance significantly improved solely with the removal of background noise.
- Despite structural abnormalities, linear normalization proved feasible for PVL-affected pediatric brains.
Conclusions:
- Linear normalization is a viable technique for pediatric brain images exhibiting structural changes from PVL.
- Image preprocessing steps, particularly background noise removal, are critical for optimizing linear normalization accuracy in this population.
- These findings support the use of linear normalization in research involving children with PVL, facilitating further neuroimaging studies.
Abstract:
The feasibility of linear normalization of child brain images with structural abnormalities due to periventricular leukomalacia (PVL) was assessed in terms of success rate and accuracy of the normalization algorithm. Ten T1-weighted brain images from healthy adult subject and 51 from children (4-11 years of age) were linearly transformed to achieve spatial registration with the standard MNI brain template. Twelve of the child brain images were radiologically normal, 22 showed PVL and 17 showed PVL with additional enlargement of the lateral ventricles. The effects of simple modifications to the normalization process were evaluated: changing the initial orientation and zoom parameters, masking non-brain areas, smoothing the images and using a pediatric template instead of the MNI template. Normalization failure was reduced by changing the initial zoom parameters and by removing background noise. The overall performance of the normalization algorithm was only improved when background noise was removed from the images. The results show that linear normalization of PVL affected brain images is feasible.