China
Africa
Explore chapters and articles related to this topic
Feature extraction and qualification
Published in Ruijiang Li, Lei Xing, Sandy Napel, Daniel L. Rubin, Radiomics and Radiogenomics, 2019
One important prerequisite procedure is the pre-processing of imaging data before the extraction of features, which aims at enhancing image quality, obtaining region of interest, thus enabling the repeatable and comparable radiomic analysis. The starting point is the image stack (e.g., readouts from Digital Imaging and Communications in Medicine (DICOM) files). For some imaging modalities, such as PET, the images should be converted to a more meaningful representation (standardized uptake value, SUV) to take into account varying biodistributions related to the injected amount of the tracer, its activity, and patient’s weight. Voxel size resampling is a vital pre-processing step for datasets that have variable voxel sizes (Thibault et al. 2014). Specifically, isotropic voxel size is required for some texture feature extraction. Down-sampling to larger dimension will lead to information loss. While, up-sampling may add artificial information. It is still unclear which one is better and perhaps the best way to evaluate this is to base the judgement on the model’s performance. There are two main categories of interpolation algorithms: polynomial and spline interpolation. Nearest neighbor is a zero-order polynomial method that assigns gray-level values of the nearest neighbor to the interpolated point. Bilinear or trilinear interpolation and bicubic or tricubic interpolation are often used for 2D in-plane interpolation or 3D cases. Cubic spline and convolution interpolation are third-order polynomial methods that interpolate smoother surface than linear methods, while being slower in implementation. Linear interpolation is a rather commonly used algorithm, since it neither leads to the rough blocking artifacts images that are generated by nearest neighbors, nor will it cause out-of-range gray levels that might be produced by higher order interpolation (Zwanenburg et al. 2016).
Medical image interpolation based on 3D Lanczos filtering
Published in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2020
Thiago Moraes, Paulo Amorim, Jorge Vicente Da Silva, Helio Pedrini
The tricubic interpolation (Lekien and Marsden 2005) generates values at arbitrary points on a three-dimensional regular grid. The tricubic interpolation preserves fine detail in the output image, however, it demands high computational cost. The method requires the approximation of a function expressed as