Unified Learning of CT Reconstruction and Organ Segmentation

Read full text Summarise with VIRAT
Authors: Alexey V. Ivanov, Natalia S. Petrova, Mikhail A. Kuznetsov
Cited by: 0
References: 16

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

CT reconstruction quality has a direct impact on downstream tasks such as organ segmentation, especially under low-dose and noisy conditions. Inspired by convolution–self-attention denoising models such as CTLformer, this work proposes a joint learning framework that shares convolution–attention features between reconstruction and segmentation branches. The reconstruction module improves image fidelity, while the segmentation loss provides task-driven constraints that discourage over-smoothing. Experiments are performed on a dataset containing 28,000 reconstructed slices with paired organ masks across three anatomical regions. Compared with separate training pipelines using RED-CNN or transformer-based reconstruction followed by a U-Net segmenter, the proposed joint approach increases reconstruction PSNR by 0.6–1.0 dB and improves Dice scores by 1.8%–3.4% on low-dose test sets.

Fields of Study & Subjects

Medical Image Segmentation Techniques Image and Signal Denoising Methods Medical Imaging Techniques and Applications

Publication details

Journal / source
Global Media and Social Sciences Research Journal
Volume
7
Issue
1
Pages
200–206
Publication date
15 February 2026
Language
EN