Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

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

Mitophagy deficiency and calcium dyshomeostasis underlying the pathogenesis and progression of Alzheimer's disease.

Journal of Alzheimer's disease : JAD·2026

Same author

Lodicule Isolation and Morphometric Analysis During Rice Floret Opening.

Bio-protocol·2026

Same author

NADH-Dependent Oxidoreductase Activity Assay of OsAIM1 Using a Microplate Reader.

Bio-protocol·2026

Same author

Beyond Scores: Proximal Diffusion Models.

Advances in neural information processing systems·2026

Same author

Similar Findings on 18F-FDG PET/CT and 68Ga-DOTATATE PET/CT in a Patient With Extraskeletal Ewing Sarcoma.

Clinical nuclear medicine·2026

Same author

18F-FCH PET/CT Localizes Ectopic Parathyroid Adenoma.

Clinical nuclear medicine·2026

Search research articles

Related Experiment Video

Updated: Jul 6, 2025

Brain Infarct Segmentation and Registration on MRI or CT for Lesion-symptom Mapping

Brain Infarct Segmentation and Registration on MRI or CT for Lesion-symptom Mapping

Published on: September 25, 2019

Task-Driven CT Image Quality Optimization for Low-Contrast Lesion Detectability with Tunable Neural Networks.

Matthew Tivnan^1,2, Tzu-Cheng Lee², Ruoqiao Zhang²

¹Department of Biomedical Engineering, Johns Hopkins University. Baltimore, MD, USA.

Proceedings of Spie--The International Society for Optical Engineering

|January 8, 2024

Summary

Tunable neural networks improve low-dose CT image quality for detecting low-contrast lesions. Optimizing networks for bias reduction enhances diagnostic accuracy and patient outcomes.

More Related Videos

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Published on: October 10, 2018

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

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

Published on: November 30, 2022

Related Experiment Videos

Last Updated: Jul 6, 2025

Brain Infarct Segmentation and Registration on MRI or CT for Lesion-symptom Mapping

Brain Infarct Segmentation and Registration on MRI or CT for Lesion-symptom Mapping

Published on: September 25, 2019

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Published on: October 10, 2018

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

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

Published on: November 30, 2022

Area of Science:

Medical Imaging
Artificial Intelligence
Radiology

Background:

Low-dose computed tomography (CT) images often suffer from noise, hindering the detection of low-contrast lesions.
Improved image quality in low-dose CT is crucial for enhancing diagnostic accuracy and patient outcomes.
Current deep learning models for CT reconstruction may not optimally balance image noise and lesion detectability.

Purpose of the Study:

To develop and evaluate tunable neural networks for CT image restoration.
To optimize the variance/bias tradeoff for improved low-contrast lesion detection in low-dose CT.
To identify the optimal denoising level for enhancing lesion detectability in clinical CT imaging.

Main Methods:

Synthesized low-contrast, low-dose CT images from super-high-resolution normal-dose scans for supervised training.
Trained deep learning CT reconstruction models using multiple noise realizations to penalize variance and bias separately.
Employed a training loss function with a 'denoising level' hyperparameter to control the variance/bias tradeoff.
Evaluated CT image quality and low-contrast lesion detectability using a shallow neural network classifier.

Main Results:

Identified an optimal 'denoising level' for tunable neural networks that maximizes low-contrast lesion detectability.
Demonstrated that networks prioritizing bias reduction over mean-squared error yield superior lesion detection performance.
The proposed tunable neural networks show potential for significant clinical benefit in low-dose CT interpretation.

Conclusions:

Tunable neural networks offer a promising approach to enhance low-dose CT image quality for lesion detection.
Optimizing the variance/bias tradeoff, specifically favoring bias reduction, is key for improving diagnostic performance.
This method has the potential to improve diagnostic accuracy and patient outcomes in clinical settings where low-dose CT is utilized.