Poster Poster Program Diagnostic and Interventional Radiology Physics

Beyond Hardware Limits: A Super-Resolution Convolutional Neural Network for Photon-Counting-Detector CT In Temporal Bone Imaging

Abstract
Purpose

To develop a super-resolution convolutional neural network to generate high-resolution temporal bone images beyond the resolution limitation of a commercial photon-counting-detector (PCD) CT.

Methods

With IRB approval, 10 patients’ temporal bones were scanned on a commercial PCD-CT (NAEOTOM Alpha, Siemens). Scans were performed in the ultra-high-resolution mode (120x0.2 mm collimation). Quantum iterative reconstruction (QIR3) was performed at 0.2 mm slice thickness, Hr84 (commercial images used in routine practice, network input) and Hr96. Hr96 images were denoised using a custom AI algorithm and served as the training label. A total of 10,000 patches (128x128 pixels) were extracted from input and label images for training. A U-Net architecture was used with 4 down-sampling and up-sampling, ELU activation function, and combination of loss functions including: MSE, SSIM, and Gradient loss with weights of 0.3, 0.5, and 0.2, respectively. The trained model was applied to Hr84 images to further enhance resolution. Qualitative and quantitative image quality assessment was performed by visual inspection and line-profile edge sharpness for both phantom and patient images.

Results

The network output showed substantial enhancement in resolution compared to the commercial (clinical) PCD-CT images, improving the visualization and delineation of fine osseous and membranous structures, including delicate trabecular and cellular patterns that were previously poorly resolved. Similar improvement was observed on middle ear prostheses and cochlear implant electrodes. Quantitative measurements showed the 10-90% rise distance (of line profiles) of 0.234 mm and 0.156 mm for clinical PCD-CT images and network output, indicating a 33.3% improvement.

Conclusion

The proposed super-resolution network improved the visualization and delineation of microstructures in the temporal bone compared to clinical PCD-CT images. The combination of PCD-CT and AI-based super-resolution enhances spatial resolution beyond standard reconstructions, enabling improved depiction of clinically relevant, microscopic temporal bone anatomy.

People

Related

Similar sessions

Poster Poster Program
Jul 19 · 07:00
B-Trac – Breast Tissue Rotation and Compression Apparatus for Calibration

Mammography (compressed 2D) and MRI (uncompressed 3D) capture breast tissue under different conditions, complicating tumor localization across modalities. To bridge this gap, we developed a customizable physical platform to simul...

Dayadna Hernandez Perez
Diagnostic and Interventional Radiology Physics 0 people interested
Poster Poster Program
Jul 19 · 07:00
Comprehensive Medical Physics Assessment of Digital Mammography Equipment: A Three-Year Multi-Site Evaluation of Technical Performance and Radiation Safety at 24 Saudi Arabian Healthcare Institutions (2022–2024)

To conduct a comprehensive multi-center audit evaluating the technical performance, image quality, and radiation safety of digital mammography systems across 24 unique healthcare facilities in Saudi Arabia. This study aims to est...

Sami Alshaikh, PhD
Diagnostic and Interventional Radiology Physics 0 people interested
Poster Poster Program
Jul 19 · 07:00
Starting Small: Implementing a CT Protocol Optimization Program

This talk describes our organization’s CT optimization program, and how we implemented it to make efficient use of limited physicist time.

Robert J. Cropp, PhD
Diagnostic and Interventional Radiology Physics 0 people interested