The comparison of single photon emission computed tomography (SPECT) images of inter- and intra-ictal cerebral blood flow (CBF) plays an important role in treating medical refractory epilepsy. The main goal is to localize the focus entertaining the seizures. Problems arise from differing tracer concentrations, patient movements and lack of morphological information in the SPECT images.
As part of a research project, we have developed a workflow that addresses these issues by combining registration techniques, intensity normalization and difference imaging. The steps of the workflow are fast and robust, and the computation takes only up to one minute on current hardware. The developed prototype has been evaluated in a pre-clinical environment. It allows a quick assessment of the data due to the visual focus on interesting image regions and provided good results in comparison with traditional techniques for SPECT epilepsy diagnosis.
The algorithms used in this applications are based on our registration and classification technology. Details can be found in the algorithm section.
Our workflow consists of a rigid registration, an intensity normalization, a non-rigid motion compensation and a subtraction step. At the final stage of the workflow, the result can be used for a fusion with a morphological data set, e.g. an MR or CT image.
Rigid Registration
- Automatic alignment of inter- and intra-ictal images
- State-of-the-art statistical intensity measure
- Support for volume cropping and punching
Non-Rigid Registration
- Compensation of rigid registration bias
- Correction of scanner distortions
- Improved difference image quality
- Substantial artifact reduction
Intensity Normalization
- Automatic background detection
- Detection of brain intensities
- Mapping of affine intensity transform between SPECT images
- Intenstiy correction to diagonal joint histogram
Focal Localization
- Establishment of common coordinate system
- Registration of MR or CT image with SPECT images
- Fusion of difference image with MR or CT
- Quantitative measurements in percentiles of CBF difference
The following example images show the fusion of the colour coded differences in CBF fused with an MRI of the same patient. Here, the colour red indicates a high difference of CBF and blue a low difference. Unlike other techniques, the combined workflow yields a clear visualization with noticeably reduced artifacts from difference imaging.
Focal spot at boundary of resected brain tissue
Focus entertaining seizure
Images have been taken from: "Statistical Medical Image Registration with Applications in Epilepsy Diagnosis and Shape-Based Segmentation", D. A. Hahn, PhD Thesis, University Erlangen-Nuremberg, 2009.
