Technology Category
Intelligent Robots: Multimodal Perception & Sensor FusionComputer Vision: Multi-modal VisionConstraint Satisfaction and Optimization: Satisfiability Modulo Theories
Enhancing Synthetic CT from CBCT via Multimodal Fusion: A Study on the Impact of CBCT Quality and Alignment
🤖 AI Summary
Intraoperative cone-beam CT (CBCT) suffers from severe artifacts and poor visual quality due to low radiation dose and rapid acquisition, limiting its clinical accuracy. To address this, we propose a registration-aware robust multimodal fusion framework that synthesizes high-fidelity synthetic CT (sCT) by jointly leveraging preoperative CT and intraoperative CBCT. Our method integrates deep learning–based image-to-image translation with registration-aware feature fusion and incorporates controlled-variable analysis on synthetically generated data to disentangle confounding factors. This work is the first to quantitatively characterize the coupled influence of CBCT image quality and CT–CBCT registration accuracy on sCT performance. Evaluated on real clinical data, the proposed method reduces average Hausdorff distance by 32% and improves SSIM by 0.15. Under conditions of low-quality CBCT with accurate registration, PSNR increases by up to 4.8 dB, demonstrating strong clinical reproducibility.
Technology Category
Application Category
📝 Abstract
Cone-Beam Computed Tomography (CBCT) is widely used for real-time intraoperative imaging due to its low radiation dose and high acquisition speed. However, despite its high resolution, CBCT suffers from significant artifacts and thereby lower visual quality, compared to conventional Computed Tomography (CT). A recent approach to mitigate these artifacts is synthetic CT (sCT) generation, translating CBCT volumes into the CT domain. In this work, we enhance sCT generation through multimodal learning, integrating intraoperative CBCT with preoperative CT. Beyond validation on two real-world datasets, we use a versatile synthetic dataset, to analyze how CBCT-CT alignment and CBCT quality affect sCT quality. The results demonstrate that multimodal sCT consistently outperform unimodal baselines, with the most significant gains observed in well-aligned, low-quality CBCT-CT cases. Finally, we demonstrate that these findings are highly reproducible in real-world clinical datasets.
Problem
Research questions and friction points this paper is trying to address.
Improving synthetic CT quality from CBCT using multimodal fusion
Analyzing impact of CBCT-CT alignment on sCT generation
Addressing CBCT artifacts via preoperative CT integration
Innovation
Methods, ideas, or system contributions that make the work stand out.
Multimodal fusion enhances sCT generation
Integrates CBCT with preoperative CT
Improves alignment and quality impact
🔎 Similar Papers
No similar papers found.
