| dc.description.abstract | Image processing has been an area where several algorithms and software have been developed due to its needs in numerous applications in life. One of its main field is medical image processing, which can help doctors and medical staffs to diagnose a patient’s disease more easily and accurately. Another applicable purpose of image processing is 3D parts printing, object inspection in manufacture industrial field. The usefulness of the image processing comes in with the 3D object reconstruction feature by allowing doctors and staffs to see the real shape of the patient’s organisms and by saving time to manufacture machine parts using 3D printing. In this thesis, we will examine some popular reconstruction algorithms in medical image processing field and then propose an improved method to transform 2D image sets to a 3D object by using DICOM image format. The algorithm is designed to read an inputted DICOM image files, extracts the boundary points then performs a filtered process, which removes noise data and adds missing data points for each file. Finally, we construct a 3D object point cloud by stacking each boundary on top of each other and render it by using Ball Pivot algorithm. We also provide a software application that can help medical staffs and researchers to implement the algorithm in real life scenarios and support interactions with the object. Different algorithms are implemented in the application to provide a suitable alternative depending on different scenarios and special data sets.
Keywords: Medical images processing, DICOM datasets, Contour extraction, 3D Reconstruction, Point cloud | en_US |
