4 results about "Virtual endoscopy" patented technology

The application provides a virtual endoscopy imaging method, system, device and medium. The method comprises: acquiring multi-energy level imaging body data of a target object; performing base substance decomposition processing on the multi-energy level imaging body data based on a preset base substance combination to generate base substance images corresponding to each base substance; reconstructing a luminal three-dimensional structure model representing a luminal region of the target object based on the multi-energy level imaging body data; performing lesion detection according to the base substance images and the multi-energy level imaging body data to obtain lesion information and associating the lesion information with the luminal three-dimensional structure model; performing endoscopy in the luminal three-dimensional structure model through a virtual endoscope and generating visualization information of a check region of the virtual endoscope in the check region; wherein the visualization information comprises the check region and / or lesion information in the check region. The application can intuitively observe the lesion position and component information in a dynamic check process.

The three-dimensional virtual endoscopy rendering of the lumen of a tubular structure is based on non-spectral and spectral volume imaging data. The three-dimensional virtual endoscopy rendering includes 2D images of the lumen of the tubular structure from the viewpoint of a virtual camera of a virtual endoscope passing through the lumen. In one case, the three-dimensional virtual endoscopy rendering is similar to the view provided by a physical endoscope camera inserted into and positioned at the actual tubular structure. The non-spectral volume image data is used to determine the opacity and shading of the three-dimensional virtual endoscopy rendering. The spectral volume image data is used to visually encode the three-dimensional virtual endoscopy rendering to visually distinguish the inner wall of the tubular structure and the structures of interest on the wall.

A three-dimensional virtual endoscopy rendering of a lumen of a tubular structure is based on both non-spectral and spectral volumetric imaging data. The three-dimensional virtual endoscopy rendering includes a 2-D image of a lumen of the tubular structure from a viewpoint of a virtual camera of a virtual endoscope passing through the lumen. In one instance, the three-dimensional virtual endoscopy rendering is similar to the view which is provided by a physical endoscopic video camera inserted into the actual tubular structure and positioned at that location. The non-spectral volumetric image data is used to determine an opacity and shading of the three-dimensional virtual endoscopy rendering. The spectral volumetric image data is used to visually encode the three-dimensional virtual endoscopy rendering to visually distinguish an inner wall of the tubular structure and structure of interest on the wall.

The invention relates to a three-dimensional medical image segmentation method based on multi-modal data fusion, which comprises the following steps of: firstly, processing abdominal CT (Computed Tomography) data by using medical image three-dimensional reconstruction software to generate an accurate three-dimensional model; the method comprises the steps that firstly, a three-dimensional model is adopted, virtual endoscopy is conducted on the colon on the three-dimensional model, then a 3D-UNet model is adopted as a frame for segmenting the model, a picture is intercepted from the three-dimensional model, a virtual endoscope picture with polyps and sagittal view pictures, coronal view pictures and axial view pictures corresponding to the virtual endoscope picture with the polyps are intercepted, and a multi-modal data set is formed; and finally, fusing the DICOM-format image data of the abdomen CT with the corresponding three-dimensional model intercepted image, introducing the multi-modal data characteristics of the intercepted image, and adjusting and training the 3D-UNet model to enhance the recognition capability of the 3D-UNet model for a complex structure. Therefore, the accuracy of medical image segmentation is realized, and more reliable technical support is provided for clinical diagnosis.