Poster Poster Program Diagnostic and Interventional Radiology Physics

Deep Learning-Based Sinogram Synthesis for High-Pitch Helical CT Reconstruction

Abstract
Purpose

High-pitch helical Computed Tomography (CT) reduces radiation exposure to patients and improves robustness against motion artifacts; however, the images suffer from artifacts due to limited data sampling. Recent studies proposed deep learning approaches to address the issue by integrating iterative reconstruction algorithms or embedding Katsevich algorithms with deep neural networks. However, they require extensive computational resources due to the iterative nature of the framework. This study aims to synthesize fan-beam sinograms from helical data using a deep neural network to efficiently achieve high-pitch reconstruction.

Methods

We collected whole-body diagnostic CT images from 45 patients via AAPM Low-dose Challenge dataset. We utilized a numerical approach to model the helical scanning geometry with a helical pitch of 1.5 and employed an advanced single-slice rebinning (ASSR) algorithm to acquire sinograms per slice. The fan-beam sinograms corresponding to each slice were also acquired. The patient’s data were split into 30/5/10 for training/validation/testing. Subsequently, A residual U-Net was employed to predict the 2D sinograms from the rebinned sinograms. We compared predicted and ground-truth sinograms and FBP-reconstructed images.

Results

The mean and standard deviation of the mean absolute error (MAE) and peak signal-to-noise ratio (PSNR) between predicted sinograms and ground-truth sinograms were 4.92×10-3 ±1.5×10-3 and 40.99±2.61 dB, while those calculated from the input data were 1.14 × 10-2 ±3.5×10-2 and 32.88±2.30 dB, respectively. The mean and standard deviation of the MAE and PSNR of the images reconstructed from the predicted sinograms were 6×10-4 ±1×10-4, 32.64 ± 2.73 dB, respectively, while the results were 1.2×10-3 ± 3.0 × 10-4, 24.81 ± 2.45 dB, respectively, for the images from input sinograms.

Conclusion

We developed a deep learning framework to synthesize fan-beam sinograms from the rebinned high-pitch sinograms. The proposed approach effectively suppressed artifacts introduced in high-pitch helical CT without introducing a substantial computational burden.

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