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

You might also read

Related Articles

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

Sort by
Same author

The three snake rules: Lessons in leadership from an inside perspective.

Clinical imaging·2026
Same author

Beyond the Baby Blues: Transforming Maternal Mental Health.

Journal of the American College of Radiology : JACR·2026
Same author

How to Rebrand Academic Radiology.

Journal of the American College of Radiology : JACR·2026
Same author

Reply by Authors.

The Journal of urology·2026
Same author

Stability of the translocator protein 18 kDa density within cognitive circuits in brains of virally-suppressed people with HIV.

AIDS (London, England)·2026
Same author

Repeatability of Semi-Quantitative and Volumetric Features from Artificial-Intelligence-Guided Lesion Segmentation on <sup>18</sup>F-DCFPyL PSMA-PET/CT Images: Results from a Test-Retest Cohort.

Tomography (Ann Arbor, Mich.)·2026

Related Experiment Video

Updated: Oct 28, 2025

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

3.1K

SPECTnet: a deep learning neural network for SPECT image reconstruction.

Wenyi Shao1, Steven P Rowe1, Yong Du1

  • 1Department of Radiology and Radiological Science, Johns Hopkins University School of Medicine, Baltimore, MD 21287, USA.

Annals of Translational Medicine
|July 16, 2021
PubMed
Summary

This study introduces SPECTnet, a deep learning method for single photon emission computed tomography (SPECT) image reconstruction. SPECTnet significantly reduces noise and improves spatial resolution in brain disorder imaging.

Keywords:
Deep learningSPECTconvolutional neural networkimage reconstructionquantitative imaging

More Related Videos

Deep Learning-Based Segmentation of Cryo-Electron Tomograms
10:25

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Published on: November 11, 2022

9.7K
Deep Neural Networks for Image-Based Dietary Assessment
13:19

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

9.5K

Related Experiment Videos

Last Updated: Oct 28, 2025

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

3.1K
Deep Learning-Based Segmentation of Cryo-Electron Tomograms
10:25

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Published on: November 11, 2022

9.7K
Deep Neural Networks for Image-Based Dietary Assessment
13:19

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

9.5K

Area of Science:

  • Medical Imaging
  • Artificial Intelligence
  • Neuroscience

Background:

  • Single photon emission computed tomography (SPECT) is crucial for brain disorder diagnosis but limited by low spatial resolution and high noise.
  • Hardware design and imaging physics contribute to limitations in SPECT's diagnostic capabilities.
  • Developing advanced reconstruction techniques is essential to overcome current SPECT limitations.

Purpose of the Study:

  • To develop a deep learning technique for SPECT image reconstruction directly from raw projection data.
  • To achieve high-resolution and low-noise SPECT images using artificial intelligence.
  • To present an efficient training methodology tailored for medical image reconstruction.

Main Methods:

  • A two-step training strategy was employed using a custom software generating 20,000 2-D brain phantoms for training, validation, and testing.
  • An autoencoder (AE) compressed full-size activity images into a lower-dimensional vector, serving as a compact label for a second network.
  • The second network mapped projection data to this compact vector, which was then decompressed by the AE's decoder to reconstruct the image.

Main Results:

  • SPECTnet was validated using 2,000 test examples, including a synthetic brain phantom and de-identified patient data.
  • SPECTnet performance was compared against the clinical OS-EM reconstruction method.
  • Images reconstructed by SPECTnet exhibited lower noise and more accurate information in uptake areas compared to OS-EM.

Conclusions:

  • A novel deep neural network, SPECTnet, was developed by training two separate, connectable compact networks, simplifying the development of complex models.
  • This approach effectively reduces the challenges associated with training intricate deep neural networks for medical imaging.
  • SPECTnet demonstrates the capability to produce more accurate SPECT images, enhancing diagnostic potential.