Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Concept Videos

Downsampling01:20

Downsampling

250
When considering a sampled sequence with zero values between sampling instants, one can replace it by taking every N-th value of the sequence. At these integer multiples of N, the original and sampled sequences coincide. This process, known as decimation, involves extracting every N-th sample from a sequence, thereby creating a more efficient sequence.
The Fourier transform of the decimated sequence reveals a combination of scaled and shifted versions of the original spectrum. This...
250
Deconvolution01:20

Deconvolution

246
Deconvolution, also known as inverse filtering, is the process of extracting the impulse response from known input and output signals. This technique is vital in scenarios where the system's characteristics are unknown, and they must be inferred from the observable signals.
Deconvolution involves several mathematical techniques to derive the impulse response. One common approach is polynomial division. In this method, the input and output sequences are treated as coefficients of...
246
Upsampling01:22

Upsampling

309
Managing signal sampling rates is essential in digital signal processing to maintain signal integrity. A decimated signal, characterized by a reduced frequency range due to its lower sampling rate, can be upsampled by inserting zeros between each sample. This upsampling process expands the original spectrum and introduces repeated spectral replicas at intervals dictated by the new Nyquist frequency. To refine this zero-inserted sequence, it is passed through a lowpass filter with a cutoff...
309
Difference from Background: Limit of Detection01:05

Difference from Background: Limit of Detection

7.1K
The limit of detection (LOD) is the smallest amount of analyte that can be distinguished from the background noise. The LOD value corresponds to the concentration at which the analyte signal is three times larger than the standard deviation of the blank signal. Below this value, the analyte signal cannot be differentiated from the background noise. It is calculated by dividing the calibration slope by 3 times the standard deviation of the blank signals.
The LOD indicates the presence or absence...
7.1K

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Immune mechanisms in chronic kidney disease-mineral and bone disorder: current insights and therapeutic implications.

Frontiers in medicine·2025
Same author

Anatomically and metabolically informed diffusion for unified denoising and segmentation in low-count PET imaging.

Medical image analysis·2025
Same author

Automated Quantification of Lens Cortex and Nuclear Opacity Based on Swept-Source Anterior Segment Optical Coherence Tomography.

Journal of refractive surgery (Thorofare, N.J. : 1995)·2025
Same author

LeqMod: Adaptable Lesion-Quantification-Consistent Modulation for Deep Learning Low-Count PET Image Denoising.

IEEE transactions on medical imaging·2025
Same author

Recent progress in the patterning of perovskite films for photodetector applications.

Light, science & applications·2025
Same author

Clinical features and prognosis analysis of patients with follicular lymphoma: a real-world study in China.

Annals of hematology·2025

Related Experiment Video

Updated: Sep 8, 2025

DNA Virus Detection System Based on RPA-CRISPR/Cas12a-SPM and Deep Learning
04:17

DNA Virus Detection System Based on RPA-CRISPR/Cas12a-SPM and Deep Learning

Published on: May 10, 2024

867

A personalized deep learning denoising strategy for low-count PET images.

Qiong Liu1, Hui Liu2,3,4, Niloufar Mirian2

  • 1Department of Biomedical Engineering, Yale University, United States of America.

Physics in Medicine and Biology
|June 13, 2022
PubMed
Summary

Personalized deep learning effectively denoises low-count positron emission tomography (PET) images by adapting to varying noise levels. This strategy improves image quality and lesion detectability compared to general models.

Keywords:
deep learninglow count PETnoise level disparitypersonalized denoising

More Related Videos

Deep Neural Networks for Image-Based Dietary Assessment
13:19

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

9.3K
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
04:23

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

2.0K

Related Experiment Videos

Last Updated: Sep 8, 2025

DNA Virus Detection System Based on RPA-CRISPR/Cas12a-SPM and Deep Learning
04:17

DNA Virus Detection System Based on RPA-CRISPR/Cas12a-SPM and Deep Learning

Published on: May 10, 2024

867
Deep Neural Networks for Image-Based Dietary Assessment
13:19

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

9.3K
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
04:23

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

2.0K

Area of Science:

  • Medical Imaging
  • Artificial Intelligence
  • Radiology

Background:

  • Positron emission tomography (PET) images often have variable noise levels, making standard deep learning denoising difficult.
  • Current deep learning models struggle with generalizability due to the wide range of noise in clinical PET scans.

Purpose of the Study:

  • To develop a personalized deep learning denoising strategy for low-count PET images.
  • To address the challenge of varying noise levels in PET imaging for improved clinical use.

Main Methods:

  • Trained multiple 3D U-Net models on PET images with different noise levels.
  • Developed a personalized weighting method by blending models trained on 20% and 60% count-level images.
  • Investigated the impact of training noise levels on model performance and spatial blurring.

Main Results:

  • Models trained on noisier data offered better denoising but increased spatial blurring.
  • A one-size-fits-all model showed poor generalization across diverse noise levels.
  • Personalized denoising optimized for structural similarity, mean squared error, and liver lesion detectability.

Conclusions:

  • Training data noise levels significantly impact deep learning-based PET denoising performance.
  • The proposed personalized strategy overcomes individual network limitations, offering adaptable denoising for clinical reading.
  • A personalized approach enhances PET image quality and diagnostic accuracy.