# Anatomical-Guided Attention Enhances Unsupervised PET Image Denoising Performance

arXiv: https://arxiv.org/abs/2109.00802 · HTML: https://mlsanity.com/paper/2109.00802/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- Builds on 9 of its 59 references that are in the map.
- First posted on arXiv 2 Sep 2021.

## Builds on (most cited first, 9 of 9)

- [3D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotation](https://mlsanity.com/paper/1606.06650/) (2016)
- [A Survey on Deep Learning in Medical Image Analysis](https://mlsanity.com/paper/1702.05747/) (2017)
- [Deep Image Prior](https://mlsanity.com/paper/1711.10925/) (2017)
- [Noise2Noise: Learning Image Restoration without Clean Data](https://mlsanity.com/paper/1803.04189/) (2018)
- [Attention Gated Networks: Learning to Leverage Salient Regions in Medical Images](https://mlsanity.com/paper/1808.08114/) (2018)
- [Attention Branch Network: Learning of Attention Mechanism for Visual Explanation](https://mlsanity.com/paper/1812.10025/) (2018)
- [DeepPET: A deep encoder-decoder network for directly solving the PET reconstruction inverse problem](https://mlsanity.com/paper/1804.07851/) (2018)
- [Guided Deep Decoder: Unsupervised Image Pair Fusion](https://mlsanity.com/paper/2007.11766/) (2020)
- [Full-count PET Recovery from Low-count Image Using a Dilated Convolutional Neural Network](https://mlsanity.com/paper/1910.11865/) (2019)

## Built on by (most cited first, 10 of 10)

- [List-Mode PET Image Reconstruction Using Deep Image Prior](https://mlsanity.com/paper/2204.13404/) (2022)
- [Fully 3D Implementation of the End-to-end Deep Image Prior-based PET Image Reconstruction Using Block Iterative Algorithm](https://mlsanity.com/paper/2212.11844/) (2022)
- [Dose-aware Diffusion Model for 3D PET Image Denoising: Multi-institutional Validation with Reader Study and Real Low-dose Data](https://mlsanity.com/paper/2405.12996/) (2024)
- [A Noise-level-aware Framework for PET Image Denoising](https://mlsanity.com/paper/2203.08034/) (2022)
- [Self-Supervised Pre-Training for Deep Image Prior-Based Robust PET Image Denoising](https://mlsanity.com/paper/2302.13546/) (2023)
- [Fed-NDIF: A Noise-Embedded Federated Diffusion Model For Low-Count Whole-Body PET Denoising](https://mlsanity.com/paper/2503.16635/) (2025)
- [DULDA: Dual-domain Unsupervised Learned Descent Algorithm for PET image reconstruction](https://mlsanity.com/paper/2303.04661/) (2023)
- [End-to-end Triple-domain PET Enhancement: A Hybrid Denoising-and-reconstruction Framework for Reconstructing Standard-dose PET Images from Low-dose PET Sinograms](https://mlsanity.com/paper/2412.03617/) (2024)
- [Exploiting network optimization stability for enhanced PET image denoising using deep image prior](https://mlsanity.com/paper/2502.11259/) (2025)
- [DECADE: A Temporally-Consistent Unsupervised Diffusion Model for Enhanced Rb-82 Dynamic Cardiac PET Image Denoising](https://mlsanity.com/paper/2603.07759/) (2026)

## Abstract (arXiv)

Although supervised convolutional neural networks (CNNs) often outperform conventional alternatives for denoising positron emission tomography (PET) images, they require many low- and high-quality reference PET image pairs. Herein, we propose an unsupervised 3D PET image denoising method based on an anatomical information-guided attention mechanism. The proposed magnetic resonance-guided deep decoder (MR-GDD) utilizes the spatial details and semantic features of MR-guidance image more effectively by introducing encoder-decoder and deep decoder subnetworks. Moreover, the specific shapes and patterns of the guidance image do not affect the denoised PET image, because the guidance image is input to the network through an attention gate. In a Monte Carlo simulation of [18F]fluoro-2-deoxy-D-glucose (FDG), the proposed method achieved the highest peak signal-to-noise ratio and structural similarity (27.92 ± 0.44 dB/0.886 ± 0.007), as compared with Gaussian filtering (26.68 ± 0.10 dB/0.807 ± 0.004), image guided filtering (27.40 ± 0.11 dB/0.849 ± 0.003), deep image prior (DIP) (24.22 ± 0.43 dB/0.737 ± 0.017), and MR-DIP (27.65 ± 0.42 dB/0.879 ± 0.007). Furthermore, we experimentally visualized the behavior of the optimization process, which is often unknown in unsupervised CNN-based restoration problems. For preclinical (using [18F]FDG and [11C]raclopride) and clinical (using [18F]florbetapir) studies, the proposed method demonstrates state-of-the-art denoising performance while retaining spatial resolution and quantitative accuracy, despite using a common network architecture for various noisy PET images with 1/10th of the full counts. These results suggest that the proposed MR-GDD can reduce PET scan times and PET tracer doses considerably without impacting patients.