Publication

Abstract : The efficient and reliable segmentation of three dimensional (3D) medical images is a critical task in many diagnostic and therapy operations. Traditional segmentation methods frequently entail training sophisticated 3D models, which can be computationally and time-consuming. In this paper, we suggest a novel segmentation approach that involves transforming three- dimensional (3D) images into two-dimensional (2D) slices and applying a two-dimensional (2D) U-Net architecture. This change not only lowers the computing cost but also allows us to use proven 2D segmentation techniques. To generate 2D images, we slice 3D images at defined intervals and use maximum intensity projections. For segmentation, we use a modified 2D U-Net architecture that has been trained on these transformed 2D slices. Our experimental results demonstrate the usefulness of this strategy.We compare our method to traditional 3D segmentation algorithms and talk about the trade-offs between computational efficiency and accuracy. Our findings indicate that the 2D conversion method offers promise for efficient segmentation without sacrificing results quality. This study contributes to the field of medical picture segmentation by providing a fresh look at how to deal with the computational hurdles that come with 3D data.