Abstract:Cone-beam computed tomography (CBCT) typically requires hundreds of X-ray projections, which raises concerns about radiation exposure. While sparse-view reconstruction reduces the exposure by using fewer projections, it struggles to achieve satisfactory image quality. To address this challenge, this article introduces a novel sparse-view CBCT reconstruction method, which empowers the neural field with human tissue regularization. Our approach, termed tissue-guided neural tomography (TNT), is motivated by the distinct intensity differences between bone and soft tissue in CBCT. Intuitively, separating these components may aid the learning process of the neural field. More precisely, TNT comprises a heterogeneous quadruple network and the corresponding training strategy. The network represents the intensity field as a combination of soft and hard tissue components, along with their respective textures. We train the network with guidance from estimated tissue projections, enabling efficient learning of the desired patterns for the network heads. Extensive experiments demonstrate that the proposed method significantly improves the sparse-view CBCT reconstruction with a limited number of projections ranging from 10 to 60. Our method achieves comparable reconstruction quality with fewer projections and faster convergence compared to state-of-the-art neural rendering based methods.

What problem does this paper attempt to address?

### Problems the paper attempts to solve This paper aims to solve the image quality problem in sparse - view reconstruction of cone - beam computed tomography (CBCT). Traditional CBCT techniques usually require hundreds of X - ray projections, which leads to an increase in radiation dose and thus raises concerns about radiation exposure. Although sparse - view reconstruction methods reduce the radiation dose by decreasing the number of projections, these methods are often unsatisfactory in terms of image quality. This paper proposes a new sparse - view CBCT reconstruction method, called Tissue - guided Neural Tomography (TNT), which improves the learning process of the neural field by introducing human tissue regularization. ### Specific problems and solutions 1. **Radiation dose problem**: - **Problem**: Traditional CBCT techniques require a large number of X - ray projections, increasing the patient's radiation dose. - **Solution**: Reduce the radiation dose by decreasing the required number of projections. 2. **Image quality problem**: - **Problem**: Sparse - view reconstruction methods have difficulty obtaining high - quality images when using fewer projections. - **Solution**: Introduce the Tissue - guided Neural Tomography (TNT) method, which uses the intensity differences between soft and hard tissues to decompose the intensity field into different components, thereby improving image quality. 3. **Complexity of the learning process**: - **Problem**: A single neural network faces challenges in handling the diverse features of CBCT, especially being prone to overfitting under sparse - view conditions. - **Solution**: Simplify the learning process and improve the interpretability and controllability of the model by decomposing the intensity field into multiple components and using specialized networks to model these components. 4. **Multi - view consistency**: - **Problem**: Predicted tissue projections may be inaccurate and lack multi - view consistency. - **Solution**: Introduce a time - dependent regularization factor, so that the network is guided by tissue supervision in the early training stage, and in the later stage, the network is allowed to self - supervise to predict tissue masks that are close to the real values and multi - view consistent. ### Main contributions 1. **Novel intensity field decomposition**: - A meaningful intensity field decomposition method is proposed, which decomposes the intensity field of CBCT into shape and texture components of soft and hard tissues. This decomposition method not only simplifies the learning process but also improves the interpretability and controllability of the neural field. 2. **Utilization of tissue information**: - Use the tissue information in X - ray projections and incorporate tissue - specific guidance into the training process, which significantly accelerates the convergence speed and improves the reconstruction quality. ### Conclusion The TNT method proposed in this paper performs well in sparse - view CBCT reconstruction. It can achieve reconstruction quality comparable to or even better than existing methods with a limited number of projections (10 to 60 projections) and has a faster convergence speed. This method not only reduces the radiation dose but also improves the image quality and reconstruction efficiency.

Intensity Field Decomposition for Tissue-Guided Neural Tomography

Deep Fusion Network Based Sparse View CT Reconstructions for Clinical Diagnostic Scanners

Contrast-Medium Anisotropy-Aware Tensor Total Variation Model for Robust Cerebral Perfusion CT Reconstruction with Low-Dose Scans

Learning Deep Intensity Field for Extremely Sparse-View CBCT Reconstruction

Neural Modulation Fields for Conditional Cone Beam Neural Tomography

SNAF: Sparse-view CBCT Reconstruction with Neural Attenuation Fields

Hybrid-Domain Neural Network Processing for Sparse-View CT Reconstruction

Slice-wise Reconstruction for Low-Dose Cone-Beam CT Using a Deep Residual Convolutional Neural Network

Attention-based dual-branch deep network for sparse-view computed tomography image reconstruction

TomoGRAF: A Robust and Generalizable Reconstruction Network for Single-View Computed Tomography

Fast and accurate sparse-view CBCT reconstruction using meta-learned neural attenuation field and hash-encoding regularization

Learning 3D Gaussians for Extremely Sparse-View Cone-Beam CT Reconstruction

Tn-Net: A Spatiotemporal Plus Prior Image-based Convolutional Neural Network for 4D-CBCT Reconstructions Enhancement

HEAL: High-Frequency Enhanced and Attention-Guided Learning Network for Sparse-View CT Reconstruction

Dual-Domain Reconstruction Network Incorporating Multi-Level Wavelet Transform and Recurrent Convolution for Sparse View Computed Tomography Imaging

Tensor Convolutional Neural Network Architecture for Spectral CT Reconstruction

Geometry-Aware Attenuation Learning for Sparse-View CBCT Reconstruction

Low-dose CT image reconstruction based on a deep neural network

GPU-based iterative cone-beam CT reconstruction using tight frame regularization

Synchrotron radiation sparse-view CT artifact correction through deep learning neural networks

Learned Tensor Neural Network Texture Prior for Photon-Counting CT Reconstruction