Application Research Of Multimodal Image Fusion Technique Based On Cerebral Cortex Reconstruction In Neurosurgery

Objective To explore the method for reconstruction and parcellation of cerebral cortex with structural MRI with intra cranial lesions under clinical settings.And to campare the dispay effect between the constructed model and 3D volume rendering.The subtraction method for reconstruction of intracranial veins between MRIs was also established and compared with traditoional MRV.The method for reconstruction of multiple arteries and veins in dual phase was established and evaluated.Multimodal image registration method with different algorithms using open source software was evaluated and correlative factors were analyzed,an optimized registration strategy was established.The advantages and application value of multimodal image fusion technique based on cerebral cortex reconstruction in neurosurgery were further explored.Methods Seventy two patients who need DSA and MRI examination for intra operative navigation in our department during January 2014 and December 2016 were included in current study.The diagnosis for the pateients were brain tumor and vascular malformation.All the patients accepted DSA,MRI,MRI with contrast,MRV.The original DICOMs of all the image studys were exported from the PACS and then imported to a personal computer with MacOS system.Postprocess with open source software as Horos,FSL,Freesurfer and 3D Slicer were performed.The entire postprocess pipeline included cropping,modelling,registration,subtraction,fusion and rendering.Series models were acquired through the process including cerebral cortex model,lesion model,intracranial vein model,and multiple vessels model.Two senior neurosurgeon evaluated each model with standard subjective evaluating scales.Two neurosurgeon whom performed the surgery were assess the match degree between the model and intra operative findings.Comparison studys were performed between the models and 3d volume rendering method,MRI subtraction method and MRV,multiple vessels and 3D-DSA.Results 1.Cerebral cortex models could be construted in every patients and fused with multiple vessel models,intracranial vein models,and MRI volumes.During the modelling and parcellating process,the results were satisfactoru even there was huge mass effect.However,once there were lesions invasied to the cerebral coretex of the interface between the white and grey matter,defects on the final models were founded.For the rest part of models,the models and parcellations were still correct.2.The display effect of the cerebral cortex models were significantly better than 3D volume rendering method(p<0.01).The defects on the models could be repaired with registration and remodeling.3.The intra cranial vein models constructed with MRI subtraction method were as good as traditional MRV for the veins with diameter larger than 2mm.For the veins with diameter between 1mm and 2mm,the intra cranial vein models constructed with MRI subtraction method were better than traditional MRV(p<0.01).Both of the two method failed to detect the veins with diameter smaller than 1mm.4.In the comparison between the 3D-DSA and multivessel models,the models were better at displaying the construction of the vascular malformation and the course of the draining veins(p<0.01).For small vessels with diameter smaller than 0.5mm,both of the models and 3D-DSA were inferior to tradional 2D-DSA.5.The multimodality models based on cerebral cortex reconstruction were consistent with intra operative findings and demonstrated the topography of the gyri,sulci from different angel.The construction of vascular malformation and relationship between the lesions and cerebral cortex were also demonstrated clearly,which was very helpful for presurgical planning,intraoperative location and modification of navigation drift.Conclusions Cerebral cortex reconstruction could be performed perfectly once the lesion did not invasive to cerebral cortex or the interface between the gray matter and white matter.For these patients,our modified method could build models matched with reality.MRI subtraction method could display the intracranial veins and lesion effectively.Multiple vessels model could demonstrate the construction of complex vascular malformation precisely and multimodality models based on cerebral cortex reconstruction could reflect the real topography and relationship between the gyri,sulci,artery,vein,and lesions.The method established in current study was usefu for presurgical planning,intraoperative location and modification of navigation drift in neurosurgery.