An Automatic Image Sorting Algorithm For4D-CT Based On Spatial Continuity

The movement of the tumor can be positioned within the whole respiratory cycle in the4D-CT images of radiation therapy in cancer of the spatial form to eliminate image artifacts caused by respiratory movement, not only to the real display of tumors and organs, but also can reflect the law of motion of the tumor and organs to help narrow the target of putting the border, assisted precise4D image guided radiotherapy to improve the target exposure doses at the same time, reduce the toxicity to normal tissue, helping the doctor according to the patient target movement characteristics of individual radiotherapy plan designed to provide a strong image support for the realization of precise radiotherapy.At present, the4D-CT in the clinical application need to rely on external respiratory signal detection device to provide respiratory signal motion information. The usual practice is to scan the patient at the same time, the recording device in the patient’s body surface plus the respiratory cycle, which requires that the external device to the matching effect is still poor, but in some cases. This is not the original data set, but the breathing phase of the axial position of the mismatch caused. This paper presents a novel axial position between the image phase automatic matching algorithm does not require external respiratory monitoring devices. Continuous feature which is based on the axial images of natural space to find the image of the same respiratory phase between adjacent bed spaces be included in the corresponding images in the collection, and so on, until the image of all the beds are classified into corresponding phase image collection. Image phase matching, we improved the traditional "daisy chain" Sort of a single image, the first time the loop Sort strategy to achieve a fast matching of the image phase. As for sorting algorithm of image phase matching, this paper completed the following work:(1) Calculate and determine whether the two adjacent beds image phase matching cross-correlation coefficient between the first quantify and calculate the image of the same respiratory phase of the reference sort beds, and cross-correlation coefficient as the respiratory phase spatial continuity of indicators, and then calculate the cross-correlation coefficient between images of adjacent beds, and finally according to the spatial continuity of indicators to determine the respiratory phase of the two sets of images to match.(2) In the process of image phase matching, we improve the traditional "daisy chain" sort method of a single image, and propose loop sort strategy to achieve a fast matching of the image phase.(3) This paper uses a large number of experimental data to verify the algorithm, including the breathing phantom data and clinical oncology patients’ scan data.After multiple breathing phantom and clinical image data verification, the proposed algorithm can match all the axial position of the data sets, facilitate doctors in spatial orientation and accurate outline of the point in time the lesion outline, develop sophisticated radiation treatment planning to achieve accurate treatment and reduction of the purpose of patient dose.