Post-processing Techniques To Aid Cranial CT Images Identification In The Same Radiographic Positioning

ObjectiveTo explore the applying of post-processing techniques and sectional anatomy knowledge in reconstructing any oblique slice image, and attempt to provide a new technology advantage for radiographic comparison between two group images of CT.Methods26cases with twice cranial CT scans were selected from the patient image data of the Huashan Hospital and Ruijin Hospital over a period of two years (May2008-October2009). CT scans of10mm thickness of26patients were selected as’Primary group’, the multislice computed tomography (MSCT) data with0.625mm thickness taken in different time of those26patients were selected as ’After group’. MSCT data with0.625mm thickness of patients whose value of the maximum cranial length(g-op)and the gender were similar with those26patients were selected as’Control group’. After importing three group data, observing anatomical structures, and searching for signs of landmark, we used ’After group ’or’ Control group’data to reconstruct any oblique slice with’ObliqueSlice’module for duplicating’Primary group’image in the same radiographic positioning, then used ’fit to points’ toggle and ’setplane’command to minute adjust the position, and applied ’AlignSlices’ module and’Manipulator’ module to match twice CT images. Loading ’After group’or’Control group’ data and’Primary group’ in two viewers, set landmarks on their surfaces by locating corresponding features in the two objects and to select reasonable positions for landmarks. Connect the LandmarkWarp module to MSCT data, and then make a rigid transformation to match the corresponding landmarks as well as possible by performing rotations and translations of MSCT data.ResultsThe difference value of the maximum cranial length(g-op) in3D image between’After group’and’Control group’was0-0.11cm(0.0381±0.03359cm). We observed both’Primary group’ and ’After group’ data, translated the slices along its direction and defined corresponding landmarks, numerated the number of center of rotate handle, slice distance, the number of between the sections, and the distance to the center of rotation. After selecting rotation center slice with translates slider, changing the slice orientation slowly by rotate toggle, we can reconstruct ’After group’ radiographic positioning in accordance with’Primary group’. In the ’Manipulator’ module registration, different map modes were used to distinguish the differences, the twice CT group could be matched to the smallest detail, and achieved an effective comparison. We also did similar operation on’Control group’data, the fitting degree with’Primary group’ is low. No meaningful reconstruction slice can be obtained to simultaneously reproduce the’Primary group’orientation and position. Displaying ’Primary group’ and ’After group’data in two viewers, setting landmarks on their surfaces by locating corresponding features in the two objects and to select reasonable positions for landmarks. Connect the LandmarkWarp module to MSCT data, and then make a rigid transformation to match the corresponding landmarks as well as possible by performing rotations and translations of MSCT data. After three-dimensional image registration data, the CT slices and CT volume rendering can be compared in a three-dimensional space, With an alignment of the MSCT object to’Primary group’ data performed in3D viewer, we can see that the result of the’After group’ transformation fits the’Primary group’quite well. The outlines and edges of the zygomatic arches, the mandibles, the nasal septum deviation, as well as the occipital bone, were a positive match. The correspondence of all the morphologic features observed in the comparison was sufficient to yield a positive identification.ConclusionsBased on post-processing techniques, MSCT data set of the same person can be used in reconstructing any oblique slices to duplicating primary radiographic positioning in different time. The created slices can be used to radiographic superimposition, and achieved an effective comparison. Among these module, the different map modes of’Manipulator’ module registration and CT volume rendering registration which could be matched to the smallest detail may be better methods. ObjectiveAfter using post-processing techniques, the oblique slices, whose radiographic positioning is consistent with primary images, can be obtained. Then to explore equations for identification of the craniofacial anthropometric measurement in CT images by binary Logistic regression analysis. These measurement are differences of indexes of the different cranial CT examinations on the same people, or indexes of the cranial CT examinations between the different individuals.MethodsUsing post-processing techniques to deal with ’After group’ and ’Control group’ data to obtain new oblique slices, whose radiographic positioning were consistent with ’Primary group’.34landmarks were identified in the sphenoid plane and other planes for craniofacial anthropometric measurement. The differences of indexes of the same and different individuals were calculated, and the results were descriptive statistics. Set the discrepancies between ’After group’ and’Primary group’ as the covariates X, the corresponding dependent variable Y set to be0. Set the discrepancies between’Control group’ and ’Primary group’ as the covariates X, the corresponding dependent variable Y is set to be1. Enter method of Binary Logistic regression was used to analysis16groups of indicators discrepancies. Then indicators were selected according to the binary value and score and P value, and taken to Backward Stepwise Conditional.ResultsWe identified34landmarks in the sphenoid plane and other planes according to previous studies for craniofacial anthropometric measurement. In this study,16parameters of the skull CT were designed to craniofacial anthropometric measurement. Some results are as follows:no significant difference among the three groups. After the ’After group’ and the’ Control group’ were subtracted by the ’Primary group’, the discrepancies of parameters between the same individuals were significantly less than the parameters between different individuals. Enter method of Binary Logistic regression was used to analysis16groups of indicators discrepancies, and16equations for identification of cranial CT images were established. The correct rate of index of maxillary sinus width or bizygomatic breadth, was the highest,90.4%, and other indexes, such as angle of petrous bone, frontal sinus width, maximum cranial width in frontal sinus plane, their identification correct rates were more than80%and had practical value. With the method of Backward Stepwise Conditional, the following formula can be used for identity determination from measurements:Y=X3×89.716+X4×21.186+-X9×3.185+X16×26.174-53.098, error rates of0%, OR of1.1489732385186304E23.ConclusionsAfter MSCT data can be reconstructed to be’Primary group’CT position, it could provide accurate measurement of objective data, the discrepancies of parameters between the same individuals were significantly less than the parameters between different individuals.The Binary Logistic Regression, access to bizygomatic breadth, maximum breadth between the sigmoid sinuses, angle of greater wing and maximum cranial width in frontal sinus plane, could be used to identify the same head CT films, and provide more objective error rate which consistent with rules of evidence.