Warning: Creating default object from empty value in /site/joomla/components/com_k2/models/item.php on line 445
Registration of low-dose and diagnostic chest CT scans based on skeletal and bronchus surface models
PURPOSES
For the postprocessing of the PET data acquired from PET/CT studies of the human chest is in some cases necessary to produce the fusion of the diagnostic CT and PET images.
Due to the different laying protocols different CT scans of the same subject may represent different morphological states of the target area. This effect makes the comparison of diagnostic and low-dose CT scans complicated. The purpose of this work is to develop an intrasubject chest CT registration method based on skeletal and bronchus surface models.
METHODS
Both in case of the diagnostic and low-dose CT scans dedicated algorithms were developed for segmenting the regions of pectoral bone tissue and bronchus. Former problem was solved by a simple intensity-threshold based procedure, latter was performed by a particular adaptive region growing algorithm. Using the surface models and landmark points retrieved from the segmented regions, a complex transformation and deformation method was constructed, which solves the problem of the CT registration accurately in some regions, while in areas being less relevant in the task, are weaker criteria necessitated.
RESULTS
The developed method permits of the local nonlinear registration of the diagnostic and low-dose CT scan of the same subject. Due to the nature of the method, the accuracy of the registration is highest in the featured environment of the skeleton and bronchus.
The PET/CT data of 3 subjects was used for designing and surveying our issue. The clinical validation of the method is in progress within the confines of the institutional virtual bronchoscopy project.
CONCLUSION
One possible application of the elaborated registration method is the PET-assisted virtual bronchoscopy. The method according to the primary registration tests proved to be effective, however the clinical validation is still in progress.