Slice Reconstruction on 3D Medical Image using Optical Flow Approach

Abstract

High-resolution, high-quality imagery forms the basis of accurate 3D reconstructions. 3D reconstruction is obtained from a 3-dimensional stacking array or image sequence. In some cases the resulting slice image has a low resolution so that some parts miss information when reconstructed into 3D. The reconstruction of the new slice carried out by the researcher using the 3D interpolation technique has a disadvantage, namely that when the calculation is carried out using the structure similarity evaluation metrics, it is still unsatisfactory, therefore the following research will try to reconstruct a new slice using the optical flow approach to calculate the displacement vector field between the two adjacent slices and also we will evaluate the comparative suitability of various interpolation techniques using root mean square error (RMSE), peak signal-to-noise ratio (PSNR), structural similarity index (SSIM). experimental results shows that the use of the optical flow method in image reconstruction between I0 and I1 after interpolation has an increase in the similarity structure value of 0.002, a decrease in the NMSE value of 0.04, and an increase in the PSNR value of 2 leading to figure I0 rather than without using optical flow.