Neural Discrete Representation Learning for Sparse-View CBCT Reconstruction: From Algorithm Design to Prospective Multicenter Clinical Evaluation
🤖 AI Summary
To address excessive radiation dose in sparse-view cone-beam CT (CBCT) for intraoperative needle guidance, this paper proposes DeepPriorCBCT—a novel three-stage deep learning framework that uniquely integrates neural discrete representation learning with clinical usability optimization. It achieves diagnostic-quality image reconstruction at merely 1/6 the standard dose. The method comprises three modules: sparse-view reconstruction, multi-scale feature enhancement, and structure–semantics co-optimization. Trained on a large-scale retrospective dataset of 8,675 cases from 12 clinical centers, it was rigorously validated via a prospective multi-reader crossover study. Five blinded radiologists rated its image quality as statistically indistinguishable from full-dose CBCT (p > 0.05) while meeting real-time intraoperative navigation requirements. To our knowledge, this is the first low-dose CBCT reconstruction study validated through both large-scale multicenter retrospective training and prospective clinical evaluation—establishing a clinically translatable paradigm for safe, precise interventional imaging.
Application Category
📝 Abstract
Cone beam computed tomography (CBCT)-guided puncture has become an established approach for diagnosing and treating early- to mid-stage thoracic tumours, yet the associated radiation exposure substantially elevates the risk of secondary malignancies. Although multiple low-dose CBCT strategies have been introduced, none have undergone validation using large-scale multicenter retrospective datasets, and prospective clinical evaluation remains lacking. Here, we propose DeepPriorCBCT - a three-stage deep learning framework that achieves diagnostic-grade reconstruction using only one-sixth of the conventional radiation dose. 4102 patients with 8675 CBCT scans from 12 centers were included to develop and validate DeepPriorCBCT. Additionally, a prospective cross-over trial (Registry number: NCT07035977) which recruited 138 patients scheduled for percutaneous thoracic puncture was conducted to assess the model's clinical applicability. Assessment by 11 physicians confirmed that reconstructed images were indistinguishable from original scans. Moreover, diagnostic performance and overall image quality were comparable to those generated by standard reconstruction algorithms. In the prospective trial, five radiologists reported no significant differences in image quality or lesion assessment between DeepPriorCBCT and the clinical standard (all P>0.05). Likewise, 25 interventionalists expressed no preference between model-based and full-sampling images for surgical guidance (Kappa<0.2). Radiation exposure with DeepPriorCBCT was reduced to approximately one-sixth of that with the conventional approach, and collectively, the findings confirm that it enables high-quality CBCT reconstruction under sparse sampling conditions while markedly decreasing intraoperative radiation risk.
Problem
Research questions and friction points this paper is trying to address.
Develops a low-dose CBCT reconstruction method for thoracic tumor procedures
Validates the model using large-scale multicenter retrospective and prospective trials
Reduces radiation exposure to one-sixth while maintaining diagnostic image quality
Innovation
Methods, ideas, or system contributions that make the work stand out.
Deep learning framework reduces radiation dose to one-sixth
Validated with large multicenter dataset and prospective clinical trial
Achieves diagnostic-grade image quality comparable to standard scans
🔎 Similar Papers
No similar papers found.
