Deep Learning Image Reconstruction for Gemstone Spectral Imaging

by GE HealthCare  · Based in United States → — Elevating CT Image Quality and Diagnostic Confidence with Deep Learning.
Interventional Radiology Radiology

Contact vendor for pricing
Regulatory Status Disclosed

Overview

Deep Learning Image Reconstruction for Gemstone Spectral Imaging (DLIR-GSI), often branded as TrueFidelity™ for GSI, is an advanced image reconstruction solution developed by GE HealthCare for its Computed Tomography (CT) systems equipped with Gemstone Spectral Imaging (GSI) technology. This innovative tool leverages a dedicated deep neural network (DNN) to significantly enhance image quality by intelligently differentiating and suppressing noise while preserving intricate anatomical and pathological structures.

DLIR-GSI processes material decomposed sinograms from dual-energy CT acquisitions, learning unique noise characteristics in material bases like iodine and water. The resulting TrueFidelity CT images offer outstanding image quality with preferred noise texture, improved spatial resolution, and reduced artifacts compared to traditional filtered back-projection (FBP) and iterative reconstruction (IR) methods.

Designed for fast reconstruction speed, DLIR-GSI supports routine CT use, even in acute care settings, and has the potential to enable lower radiation doses without compromising diagnostic performance. It is intended for whole body, vascular, and contrast-enhanced head CT applications across all ages, improving diagnostic confidence in a wide range of clinical scenarios.

Reviewed by Pouyan Golshani, MD — Interventional Radiologist

Key Features

  • Deep Neural Network (DNN) based image reconstruction
  • Intelligent noise suppression while preserving natural image texture
  • Enhanced spatial resolution and artifact reduction
  • High quantitative accuracy in spectral imaging
  • Fast reconstruction speed for routine CT workflows
  • Supports various spectral image types (monochromatic, material basis pairs, VUE)
  • Applicable for whole body, vascular, and contrast-enhanced head CT
  • Potential for lower radiation doses
  • Improved diagnostic confidence and lesion detectability
  • Compatible with GE's Gemstone Spectral Imaging (GSI) CT systems

Use Cases

  • Head imaging (e.g., contrast-enhanced head CT)
  • Whole body imaging
  • Cardiovascular imaging (e.g., CT angiography, arterial plaque discernment)
  • Pulmonary imaging (e.g., small airways, pulmonary nodules, interstitial lung disease)
  • Musculoskeletal imaging (e.g., trabecular bone patterns, joint replacement)
  • Abdominal imaging (e.g., small lesion detection, kidney stone characterization)

What Physicians Need to Know

DICOM Support & Standards
GE HealthCare leverages DICOM data directly from imaging devices for operational insights, indicating adherence to DICOM standards for image data handling and workflow integration.
PACS Integration Method
The solution integrates seamlessly into existing workflows, suggesting standard PACS integration, likely via DICOM. GE HealthCare also offers 'PACS and AI Orchestration' as part of its IT solutions.
Reading Room Workflow Impact
Designed to deliver outstanding detail, clarity, and texture, TrueFidelity for GSI aims to improve image quality not easily achievable by conventional methods. It enables radiologists to make faster, more confident diagnoses due to clearer, sharper images, potentially reducing reading times and radiologist fatigue. The flexible workflow allows for effortless generation of various image types directly from the scanner.
AI Model Architecture (deep learning approach)
Utilizes a deep neural network (DNN) trained on thousands of high-dose ground truth filtered back projection (FBP) images. The neural network learns desired image characteristics and is specifically trained to reduce noise while preserving details in spectral material basis (e.g., iodine and water). It employs convolutional neural networks (CNNs) and incorporates model compression techniques for computational efficiency.
Processing Speed (per study)
Offers routinely fast reconstruction speeds for daily needs, even for spectral imaging. For instance, it can reconstruct 585 0.625 mm images of a typical chest PE scan in less than 2 minutes on Revolution CT and Revolution Apex scanners. Other related DL tools like True Definition DL can perform chest imaging in under one second.
FDA Clearance Pathway (510k/De Novo)
GE HealthCare's deep learning reconstruction solutions, including those building on TrueFidelity DL (like True Definition DL), have received 510(k) clearance from the U.S. FDA. The underlying GSI Xtream technology also received 510(k) clearance.
Supported Modalities (CT/MRI/X-ray/US)
Primarily supports **Computed Tomography (CT)**, specifically with **Gemstone Spectral Imaging (GSI)**. It is available on GE HealthCare's Revolution CT and Revolution Apex scanners.
Sensitivity & Specificity Data
The focus is on improving image quality metrics. TrueFidelity for GSI significantly reduces image noise (e.g., 27u00b13% reduction compared to IR 50%) and enhances contrast-to-noise ratio. It has shown improved lesion detectability, increasing it by an average of 64% for hepatocellular carcinomas and 60% for hypervascularized metastasis between 40 and 80 keV. It also improves visualization of small structures like pulmonary nodules and airways.
RSNA/ACR Validation
GE HealthCare regularly presents and details its AI advancements, including deep learning reconstruction for GSI, at the Radiological Society of North America (RSNA) annual meetings. Peer-reviewed evidence supporting its image quality and diagnostic performance benefits has also been published.
Physician Tip

Leverage the enhanced image quality, including reduced noise, improved contrast, and sharper details, for more confident and potentially faster diagnoses, especially for subtle lesions and complex anatomies. Utilize the flexibility to generate various spectral image types (e.g., monochromatic, iodine-specific) to optimize visualization for specific clinical questions. The deep learning reconstruction can enable dose reduction while maintaining or improving diagnostic image quality. The seamless integration into existing workflows should facilitate quick adoption and minimal disruption.

The Deep Learning Image Reconstruction for Gemstone Spectral Imaging (TrueFidelity for GSI) is integrated directly on compatible GE HealthCare CT scanners (Revolution CT, Revolution Apex) and is part of the broader 'Effortless Recon DL' portfolio. It is designed for seamless integration into existing radiology workflows, utilizing DICOM data for operational insights and supporting both new system installations and upgrades.

Details

Category Radiology & Imaging AI
Pricing Contact vendor for pricing — Not publicly available
DeploymentIntegrated into GE HealthCare CT scanners with Gemstone Spectral Imaging (GSI) technology (e.g., Revolution CT, Revolution HD CT, Discovery CT750 HD).
Compliance
BAA AvailableUnknown AI-estimated
HIPAA CompliantUnknown AI-estimated
FDA Status 1 AI-estimated

The Deep Learning Image Reconstruction for Gemstone Spectral Imaging (DLIR-GSI) option received FDA 510(k) clearance (K201745) on December 10, 2020. It is intended to produce cross-sectional images by computer reconstruction of dual energy X-ray transmission data acquired with Gemstone Spectral Imaging, for all ages, and can be used for whole body, vascular, and contrast enhanced head CT applications.

Integrations
EHR Not specified
Specialties Interventional Radiology, Radiology

What the Web Says

GE Healthcare's Deep Learning Image Reconstruction (DLIR) with Gemstone Spectral Imaging (GSI), branded as TrueFidelity for GSI, is a significant advancement in CT imaging. It aims to improve image quality, reduce noise, and enhance diagnostic confidence across various clinical applications. The technology leverages deep neural networks trained on high-quality data to produce images with natural texture and superior detail compared to traditional iterative reconstruction methods.

Overall: Positive

Strengths

  • Superior image quality with outstanding detail, clarity, and natural texture.
  • Significant noise reduction, especially in low-energy keV and material-selective images.
  • Improved lesion detectability and characterization.
  • Enhanced contrast-to-noise ratio (CNR) and signal-to-noise ratio (SNR).
  • Effective in reducing metal artifacts when combined with Metal Artifact Reduction (MAR) algorithms.
  • Potential for radiation dose optimization while maintaining high image quality.

Limitations

  • Deep learning techniques can be computationally expensive.
  • Requires large amounts of training datasets.
  • Lack of decent theory to fully explain why the algorithms work.
  • Potential issues with generalization and robustness in varied clinical scenarios.
  • The reliance on large labeled datasets may limit broader application in medical image reconstruction.
  • Some deep learning reconstruction algorithms for spectral images are not yet widely available on commercial DECT platforms.

Based on reviews from: GE Healthcare, CT Theory and Applications, PMC, Intelligent Medicine, arXiv, AAPM, Frontiers, OUCI

Last updated: 2026-07-18

Ratings & Reviews

No reviews yet. Be the first to review this tool!

Rate Deep Learning Image Reconstruction for Gemstone Spectral Imaging

Clinical Value
Ease of Use
Integration
Support & Docs
Value for Money

Press & Coverage

FDA
Deep Learning Image Reconstruction for Gemstone Spectral Imaging (DLIR-GSI) 510(k) Premarket Notification
This regulatory announcement details the FDA's clearance of GE Medical Systems' Deep Learning Image Reconstruction for Gemstone Spectral Imaging (DLIR-GSI) option, intended for whole body, vascular, and contrast-enhanced head CT applications. The document confirms that DLIR-GSI successfully completed design control testing and was deemed substantially equivalent to its predicate device.
2020-12
GE Healthcare
TrueFidelityu2122 for Gemstoneu2122 Spectral Imaging - GE Healthcare
This GE Healthcare white paper introduces TrueFidelityu2122 for Gemstoneu2122 Spectral Imaging (GSI), highlighting it as a new generation of spectral imaging powered by deep learning. It emphasizes the technology's ability to transform image quality for dual-energy spectral CT by reducing noise and preserving natural image texture.
2019-07
Physician AI Tools
Deep Learning Image Reconstruction for Gemstone Spectral Imaging (DLIR-GSI)
This article from Physician AI Tools describes GE HealthCare's Deep Learning Image Reconstruction for Gemstone Spectral Imaging (DLIR-GSI), branded as TrueFidelityu2122 for GSI, as a technology that enhances CT image quality using deep neural networks to reduce noise, improve spatial resolution, and boost diagnostic confidence.
2026-04
Innolitics
Photonova Spectra, Photonova Spectra Select (K253520) u2014 FDA 510(k)
This FDA 510(k) summary for Photonova Spectra references Deep Learning Image Reconstruction for Gemstone Spectral Imaging (DLIR-GSI) as a predicate device. It highlights the Photonova Spectra's next-generation deep learning reconstruction, TrueFidelity DL for PCCT, which uses a multi-layer convolutional neural network to produce low-noise images.
2026-03
FDA
Establishment Registration & Device Listing - FDA
The FDA's Establishment Registration & Device Listing includes 'Deep Learning Image Reconstruction for Gemstone Spectral Imaging' as a listed device under GE Medical Systems, LLC. This entry confirms its regulatory status as a computed tomography x-ray system.
2026-06
Medical AI Tools
Omni Legend - Medical AI Tools for Physicians
This article mentions GE HealthCare's Deep Learning Image Reconstruction for Gemstone Spectral Imaging (DLIR-GSI), branded as TrueFidelityu2122 for GSI, in the context of other GE HealthCare AI solutions. It highlights its role in enhancing CT image quality through deep neural networks.
2026-04
Scientia
Sistemas electru00f3nicos de apoyo a la toma de decisiones clu00ednicas en pacientes con cu00e1ncer de mama, pulmu00f3n, colon-recto o pru00f3stata - Scientia
This peer-reviewed article references the Deep Learning Image Reconstruction for Gemstone Spectral Imaging option as a deep learning-based CT reconstruction method. It states its intended use for producing cross-sectional images from dual-energy X-ray transmission data for various CT applications.
2021-04
Innolitics
JAK u00b7 System, X-Ray, Tomography, Computed u2014 FDA Product Code
This FDA product code listing confirms the clearance of 'Deep Learning Image Reconstruction for Gemstone Spectral Imaging' on December 10, 2020, under the product code JAK for Computed Tomography X-Ray Systems.
2020-12

Videos

Product demos, reviews, and walkthroughs for Deep Learning Image Reconstruction for Gemstone Spectral Imaging.

View all on YouTube

No FAQs available yet for this tool.

Related Tools

Surgical Reality Viewer
Surgical Reality
Oncology AI
Surgical Reality offers CE- and FDA-certified AI-powered 3D medical imaging and surgical planning software that transforms CT scans into patient-specific 3D models for thoracic surgery, enhancing precision in complex lung procedures.
Oxipit ChestLink
Oxipit (part of Sectra group)
Clinical Decision Support & Reference
Oxipit ChestLink is the world's first CE Class IIb-certified autonomous AI application for healthy chest X-ray reporting, capable of identifying normal studies with 99.9% sensitivity. It automates the reporting of normal chest X-rays, reducing radiologist workload and improving efficiency.
Cercare Medical Oncology Virtual Expert
Cercare Medical
Neurology AI
Cercare Medical's Oncology Virtual Expert is an AI-powered brain tumor segmentation module that provides fast, consistent, and semi-automated tumor analysis from standard MRI data, designed for on-premise deployment within existing clinical radiology workflows.
CARPL.ai
CARPL.ai
Developer Tools & APIs
CARPL.ai is a vendor-neutral Artificial Intelligence (AI) platform that allows radiologists to access, assess, and integrate radiology AI solutions in their clinical practice. It provides a single user interface, a single data channel, and a single procurement channel for the testing, deployment, and monitoring of AI solutions in clinical radiology workflows.
CoLumbo C-Spine
Smart Soft Healthcare AD
Clinical Decision Support & Reference
CoLumbo C-Spine is an AI-powered software that processes and analyzes cervical spine MRI exams, assisting clinicians in diagnosis and assessment.
SubtleHD-CT
Subtle Medical, Inc.
Radiology & Imaging AI
Subtle Medical's SubtleHD-CT is an AI software designed to enhance the quality and resolution of CT scans, reducing noise and improving image clarity. [7]

See all Radiology & Imaging AI tools →

More from GE HealthCare

AI Orchestrator
GE HealthCare
Practice Management & Scheduling
GE HealthCare's AI Orchestrator integrates and manages AI applications within a healthcare imaging ecosystem to streamline workflows and improve diagnostic accuracy.
Aurora
GE HealthCare
Oncology AI
GE HealthCare's Aurora is an AI-powered dual-head SPECT/CT system designed to enhance diagnostic capabilities and streamline workflows in nuclear medicine, offering excellent image quality and operational efficiency for various clinical applications.
Centricity Universal Viewer
GE Healthcare
Radiology & Imaging AI
GE Healthcare's Centricity Universal Viewer is a web-based PACS with AI-powered tools, advanced visualization, and multi-modality support to enhance diagnostic confidence and streamline clinical workflows for radiologists and specialists.
Command Center
GE HealthCare
Practice Analytics & BI
GE HealthCare's Command Center is an AI-powered platform that optimizes hospital operations and patient flow through real-time visibility and predictive analytics.

View company profile →

Suggest an Edit → | Last Verified: 2026-04-18 | First Added: 2026-04-18
AI Tool Finder
AI-powered search. Results may not be comprehensive.