# 2.5D Deep Learning for CT Image Reconstruction using a Multi-GPU implementation

arXiv: https://arxiv.org/abs/1812.08367 · HTML: https://mlsanity.com/paper/1812.08367/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 20 Dec 2018.

## Built on by (most cited first, 8 of 8)

- [Sparse-view Cone Beam CT Reconstruction using Data-consistent Supervised and Adversarial Learning from Scarce Training Data](https://mlsanity.com/paper/2201.09318/) (2022)
- [Swap-Net: A Memory-Efficient 2.5D Network for Sparse-View 3D Cone Beam CT Reconstruction](https://mlsanity.com/paper/2410.10836/) (2024)
- [Sparse-View CT Reconstruction using Recurrent Stacked Back Projection](https://mlsanity.com/paper/2112.04998/) (2021)
- [A simple and effective approach for body part recognition on CT scans based on projection estimation](https://mlsanity.com/paper/2504.21810/) (2025)
- [Plug-and-Play with 2.5D Artifact Reduction Prior for Fast and Accurate Industrial Computed Tomography Reconstruction](https://mlsanity.com/paper/2506.14719/) (2025)
- [Beam Geometry and Input Dimensionality: Impact on Sparse-Sampling Artifact Correction for Clinical CT with U-Nets](https://mlsanity.com/paper/2508.17961/) (2025)
- [Unsupervised Semantic Segmentation in Synchrotron Computed Tomography with Self-Correcting Pseudo Labels](https://mlsanity.com/paper/2603.00372/) (2026)
- [Attention-Based Segmentation of WMHs and Differentiation of Vascular vs. Demyelinating Lesions](https://mlsanity.com/paper/2607.08171/) (2026)

## Abstract (arXiv)

While Model Based Iterative Reconstruction (MBIR) of CT scans has been shown to have better image quality than Filtered Back Projection (FBP), its use has been limited by its high computational cost. More recently, deep convolutional neural networks (CNN) have shown great promise in both denoising and reconstruction applications. In this research, we propose a fast reconstruction algorithm, which we call Deep Learning MBIR (DL-MBIR), for approximating MBIR using a deep residual neural network. The DL-MBIR method is trained to produce reconstructions that approximate true MBIR images using a 16 layer residual convolutional neural network implemented on multiple GPUs using Google Tensorflow. In addition, we propose 2D, 2.5D and 3D variations on the DL-MBIR method and show that the 2.5D method achieves similar quality to the fully 3D method, but with reduced computational cost.