04/22/2025
By Danielle Fretwell

The Francis College of Engineering, Department of Electrical and Computer Engineering, invites you to attend a Doctoral Dissertation Proposal defense by Bahareh Morovati on: "Photon-Counting Computed Tomography Image Reconstruction and Data Correction through Deep Learning Technique."

Candidate Name: Bahareh Morovati
Degree: Doctoral
Defense Date: Friday, May 2, 2025
Time: 10-11 a.m.
Location: Ball Hall, Room 302

Committee:
Advisor: Hengyong Yu, Professor, Department of Electrical and Computer Engineering, UMass Lowell

Committee Members*
1. Yan Luo, Professor, Department of Electrical and Computer Engineering, UMass Lowell
2. Seung Woo Son, Associate Professor, Department of Electrical and Computer Engineering, UMass Lowell

Abstract:
Medical imaging plays a crucial role in disease diagnosis, treatment planning, and monitoring. Various imaging modalities, such as X-ray, Magnetic Resonance Imaging (MRI), Ultrasound, and Computed Tomography (CT), provide critical insights into anatomical and functional abnormalities. Among these, CT is widely used for its ability to produce high-resolution cross-sectional images. The recent advancement of Photon-Counting CT (PCCT) extends conventional CT by utilizing photon-counting detectors (PCDs) to capture energy-resolved information, introducing an additional energy channel dimension. This capability enables improved material differentiation, enhanced contrast resolution, and reduced radiation dose.
Despite these advantages, PCCT faces significant challenges, including increased noise, charge splitting, pulse pileup, and detector
imperfections, which degrade image quality and affect quantitative accuracy. To address these issues, we propose a deep learning-based approach for PCCT reconstruction and data correction. Our method incorporates a dual-domain correction strategy, leveraging both sinogram and image domains to mitigate noise and detector-related artifacts. Additionally, we introduce a prior term into the reconstruction process to preserve structural details and account for spectral correlations across energy bins, ensuring more accurate results and preserving more details. Furthermore, we employ dynamic clustering and enhanced channel weighting for spectral-image similarity-based tensor decomposition with enhanced-sparsity reconstruction for sparse-view PCCT. This approach exploits the underlying spectral structure in the energy dimension, reducing noise, improving image quality while maintaining anatomical consistency and giving great results in sparse-view scenarios.
By integrating deep learning techniques, we aim to enhance PCCT imaging through high-quality reconstruction and artifact correction. The proposed methods are expected to improve diagnostic accuracy, enable better material decomposition, and reduce the dependency on high-dose acquisitions, thereby advancing the clinical applicability of PCCT in medical imaging.