A Research On Virtual Bronchoscopy

Due to the haze and other severe environmental problems,the incidence of lung cancer has been increasing rapidly.The most effective means for lung cancer treatment is early diagnosis.The traditional diagnosis methods include CT imaging diagnosis and endoscopic examination.However,with the continuous progress in medical imaging technology,computer graphics and virtual reality,a new diagnosis technology Virtual Endoscopy(VE)was proposed and developed,and Virtual Bronchoscopy(VB)is one of the most complex and important branches of VE systems.In order to decrease thedependence on hardware performance in tradition VE systems,we aim to optimize the key functional module in VB system and make it possible to work stably and efficiently on normal PCs.In our research,we firstly made a revision of the important existing technologies in different module of VB system,which includes(1)collection of the original data(2)segmentation of the human airways(3)extraction of the skeleton(4)3-dimensional airway surface reconstruction(5)virtual navigation guided by skeleton and human machine interaction module.And then,we made further research on some of the above listed step,and made following achievements while studying the traditional resolutions:(1)For the segmentation of the airways,we optimized the existing algorithm.Instead of the reclusively data set for region grow,we introduced queue data structure,which made the procedure a serial operation and significantly improved its performance.After the rough growth,we used hybrid fuzzy model to make further grow,in order to make the airway more complete.(2)For the extraction of airway skeleton,we add LUT(look-up-table)into traditional topological thinning method.By using region grow method for thinning model construction,the time complexity has been decreased.And due to the usage of LUT,the thinning procedure in the tradition algorithm has been literally augmented,and thus the performance of the whole algorithm.Related experiments demonstrated that the proposed algorithm is about 23 times faster than the traditional topological method and 220 times better than the distance fields based algorithms.(3)For the virtual navigation and human machine interaction part,we fully considered the various possibilities of interaction well designed the user experience module,which are realized based on the core of software engineering.The system provides different modes of navigation,such as purely manual,machine automatic and semi-automatic.In the meanwhile,the other operations such camera control,light control,color mapping control,transparency control,distance measure and photo capture are also available.