Sparse-View CT

Sparse-view CT reconstructs an image from projections taken at only a small number of angles, typically tens rather than the hundreds or thousands of views in a standard scan. Fewer views mean lower dose and faster acquisition, but filtered back-projection then produces strong streak artifacts because the angular sampling is too coarse.

It is used where dose, scan time or hardware limits the number of views: low-dose screening, interventional and radiotherapy cone-beam imaging, 4D-CBCT (where each breathing phase sees only a fraction of the projections), and industrial or security inspection.

The main approaches are regularised iterative reconstruction (TV and other sparsity priors), networks that remove streaks from an FBP image, sinogram interpolation, dual-domain and unrolled networks that enforce data consistency, and, more recently, per-scan representations (implicit neural fields, Gaussian splatting) and diffusion priors. Benchmarks usually simulate sparse sinograms from LoDoPaB-CT or the Mayo data; measured datasets such as 2DeteCT and the walnut collections test whether a method holds up on real data.

Browse all 157 Sparse-View CT papers in the searchable directory

Start here

If you are new to this topic, these are good first reads, chosen by the maintainers.

• Image reconstruction in circular cone-beam computed tomography by constrained, total-variation minimization (PMB, 2008)
  The TV-constrained baseline that most sparse-view papers still compare against.
• A Sparse-View CT Reconstruction Method Based on Combination of DenseNet and Deconvolution (TMI, 2018 · code)
  An early, simple CNN that post-processes the FBP image, with code; a useful first deep-learning baseline.
• DRONE: Dual-Domain Residual-Based Optimization Network for Sparse-View CT Reconstruction (TMI, 2021 · code)
  A dual-domain network that refines both the sinogram and the image while enforcing consistency, with code.
• NAF: Neural Attenuation Fields for Sparse-View CBCT Reconstruction (MICCAI, 2022 · code)
  A self-supervised neural attenuation field for sparse-view CBCT that fits each scan without training data, with code.
• R2-Gaussian: Rectifying Radiative Gaussian Splatting for Tomographic Reconstruction (NeurIPS, 2024 · code)
  The first Gaussian-splatting framework for sparse-view tomographic reconstruction; fast, with code.

Papers

The 30 most recent of 157 papers, newest first.

Browse all 157 Sparse-View CT papers in the searchable directory

Datasets, benchmarks and toolkits

Entries from the directory whose tasks or tags match this topic.

• LoDoPaB-CT (official page) · Dataset — Projection + image. 42,895 simulated human thorax slice pairs (LIDC/IDRI) at a single noise level (N0 = 4096) — other dose levels have to be re-simulated.
• 2DeteCT (official page) · Dataset — Projection + image. 5,000 real experimental 2D slices of mixed objects, 3 dose levels, multiple source-detector geometries. Excellent for reproducible real-data CT benchmarking.
• AAPM DL-Sparse-View CT data (official page) · Dataset — Projection + image. Data of the 2021 AAPM DL-sparse-view CT challenge: 4,000 simulated 2D breast-CT phantoms, each with an ideal noiseless 128-view fan-beam sinogram over 360° and its FBP image, plus 100 test cases.
• SparseBeads (official page) · Dataset — Projection + image. 48 measured scans of bead phantoms whose image sparsity varies systematically, released specifically as a benchmark for sparsity-regularised reconstruction.
• FleX-ray Walnut CBCT Collection (CWI) (official page) · Dataset — Projection + image. 42 walnuts scanned on three different orbits each, so that artefact-free reference volumes can be computed from the combined views and used as supervised-learning ground truth.
• Carved cheese (FIPS) (official page) · Dataset — Projection only. A single 2D slice of a carved cheese from the Finnish Inverse Problems Society, published at three resolutions together with the measurement matrices that encode the X-ray transform as an explicit linear operator.
• AAPM DL-Sparse-View CT Challenge (official page) · Benchmark — Recover a 2D breast-CT phantom from ideal noiseless 128-view fan-beam data: 4,000 training pairs, 100 test cases, ranked by RMSE on a public leaderboard. A common reference point for deep-learning sparse-view CT.
• DIRECTT Reconstruction GUI · Toolkit — A Python GUI built on the ASTRA toolbox for reconstructing X-ray CT data, with the DIRECTT algorithm included.

Guides

Guides that cover this topic.

• Sparse-View CT Reconstruction: Methods, Datasets and Where to Start
• How to Choose a CT Reconstruction Dataset

Other topics

• Surveys & Reviews (24)
• Classical & Iterative Reconstruction (71)
• Limited-Angle Tomography (46)
• Low-Dose CT Denoising (69)
• Metal Artifact Reduction (23)
• Spectral & Dual-Energy CT (19)
• Cone-Beam CT (77)
• Dynamic & 4D CT (24)
• Scatter Correction (11)
• Geometry Calibration & Motion Compensation (28)
• Deep Unrolling & Model-Based Learning (68)
• Self-Supervised & Untrained Methods (35)
• Diffusion & Generative Priors (68)
• Implicit Neural Representations (25)
• Gaussian Splatting (26)