Evaluation of Extra Pixel Interpolation with Mask Processing for Medical Image Segmentation with Deep Learning
This work addresses a specific technical issue in medical image segmentation for researchers, but it is incremental as it builds on prior interpolation comparisons.
The study evaluated the impact of using bicubic interpolation for both images and masks in medical image segmentation with deep learning, finding that the BIC-BIC model improved performance by up to 8.96% compared to baseline methods at certain image sizes.
Current mask processing operations rely on interpolation algorithms that do not produce extra pixels, such as nearest neighbor (NN) interpolation, as opposed to algorithms that do produce extra pixels, like bicubic (BIC) or bilinear (BIL) interpolation. In our previous study, the author proposed an alternative approach to NN-based mask processing and evaluated its effects on deep learning training outcomes. In this study, the author evaluated the effects of both BIC-based image and mask processing and BIC-and-NN-based image and mask processing versus NN-based image and mask processing. The evaluation revealed that the BIC-BIC model/network was an 8.9578 % (with image size 256 x 256) and a 1.0496 % (with image size 384 x 384) increase of the NN-NN network compared to the NN-BIC network which was an 8.3127 % (with image size 256 x 256) and a 0.2887 % (with image size 384 x 384) increase of the NN-NN network.