Dual Network Architecture for Few-view CT – Trained on ImageNet Data and Transferred for Medical Imaging | Available Intellectual Property | Rensselaer Polytechnic Institute

Dual Network Architecture for Few-view CT – Trained on ImageNet Data and Transferred for Medical Imaging

RPI ID: 2020-001-401

Innovation Summary:
This invention introduces a deep learning-based system for reconstructing CT images from a limited number of views. The model is trained to infer missing projection data and generate high-quality images from sparse inputs. It significantly reduces radiation exposure while maintaining diagnostic accuracy. The system is optimized for real-time performance and can be integrated into existing CT platforms.

Challenges / Opportunities:
Reducing radiation dose in CT imaging is a critical goal, but fewer views typically result in poor image quality. Traditional reconstruction algorithms struggle with sparse data. This invention addresses the challenge by using AI to fill in missing information, enabling safer and faster imaging. It opens opportunities for low-dose diagnostics and portable CT systems.

Key Benefits / Advantages:
✔ Low-dose imaging reduces patient exposure to radiation
✔ High-quality reconstruction maintains diagnostic fidelity with fewer views
✔ Real-time processing suitable for clinical deployment
✔ System compatibility with standard CT hardware

Applications:
• Emergency and trauma imaging
• Pediatric diagnostics
• Mobile and field-deployable CT systems
• AI-enhanced radiology

Keywords:
#FewviewCT #imagereconstruction #deeplearning #lowdoseimaging #sparsedata

Intellectual Property:
US Issued Patent 11,806,175 B2US Issued Patent 11,806,175 B2