# Efficient Folded Attention for 3D Medical Image Reconstruction and Segmentation

arXiv: https://arxiv.org/abs/2009.05576 · HTML: https://mlsanity.com/paper/2009.05576/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 13 Sep 2020.

## Built on by (most cited first, 9 of 9)

- [Spatially Covariant Lesion Segmentation](https://mlsanity.com/paper/2301.07895/) (2023)
- [LARO: Learned Acquisition and Reconstruction Optimization to accelerate Quantitative Susceptibility Mapping](https://mlsanity.com/paper/2211.00725/) (2022)
- [NeRD: Neural Representation of Distribution for Medical Image Segmentation](https://mlsanity.com/paper/2103.04020/) (2021)
- [DeDA: Deep Directed Accumulator](https://mlsanity.com/paper/2303.08434/) (2023)
- [UNISELF: A Unified Network with Instance Normalization and Self-Ensembled Lesion Fusion for Multiple Sclerosis Lesion Segmentation](https://mlsanity.com/paper/2508.03982/) (2025)
- [High-pass filtered fidelity-imposed network edit (HP-FINE) for robust quantitative susceptibility mapping from high-pass filtered phase](https://mlsanity.com/paper/2305.03844/) (2023)
- [DAGrid: Directed Accumulator Grid](https://mlsanity.com/paper/2306.02589/) (2023)
- [Multi-delay arterial spin-labeled perfusion estimation with biophysics simulation and deep learning](https://mlsanity.com/paper/2311.10640/) (2023)
- [RimSet: Quantitatively Identifying and Characterizing Chronic Active Multiple Sclerosis Lesion on Quantitative Susceptibility Maps](https://mlsanity.com/paper/2312.16835/) (2023)

## Abstract (arXiv)

Recently, 3D medical image reconstruction (MIR) and segmentation (MIS) based on deep neural networks have been developed with promising results, and attention mechanism has been further designed to capture global contextual information for performance enhancement. However, the large size of 3D volume images poses a great computational challenge to traditional attention methods. In this paper, we propose a folded attention (FA) approach to improve the computational efficiency of traditional attention methods on 3D medical images. The main idea is that we apply tensor folding and unfolding operations with four permutations to build four small sub-affinity matrices to approximate the original affinity matrix. Through four consecutive sub-attention modules of FA, each element in the feature tensor can aggregate spatial-channel information from all other elements. Compared to traditional attention methods, with moderate improvement of accuracy, FA can substantially reduce the computational complexity and GPU memory consumption. We demonstrate the superiority of our method on two challenging tasks for 3D MIR and MIS, which are quantitative susceptibility mapping and multiple sclerosis lesion segmentation.