Design Choices in Splitting-Based Self-Supervised Sparse-View CT Reconstruction
2026-07-12 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors studied ways to improve self-supervised learning for CT scan reconstruction when only limited scan views are available. They looked at how different choices in splitting the data, preprocessing, and making predictions affect the results. Their experiments showed that the best way to split data depends on the type of noise in the measurements, and that combining multiple splitting methods helps improve reconstruction quality. They also found that some image quality metrics reveal differences between approaches that others miss, offering new insights for practical CT reconstruction without full ground truth data.
self-supervised learningsparse-view CT reconstructiondata splittingnoise structurepartitioning strategyprojection-wise splittingLPIPSHaarPSIPSNRSSIM
Authors
Nadja Gruber, Lukas Neumann, Ander Biguri, Gyeongha Hwang, Markus Haltmeier, Johannes Schwab
Abstract
Self-supervised data splitting has emerged as a promising paradigm for sparse-view CT reconstruction, enabling training from incomplete measurements without fully sampled ground truth. However, the influence of key design choices, including partitioning strategy, preprocessing, and inference, remains insufficiently understood. In this work, we introduce a unified framework that decomposes splitting-based reconstruction into these three components, enabling controlled comparison of existing methods and two incremental extensions: multi-partition splitting and an alternative inference strategy. Experiments on simulated LoDoPaB-CT data under independent and correlated noise, together with validation on the real-world 2DeteCT dataset, show that the optimal partitioning strategy strongly depends on the measurement noise structure. Lattice-based splitting performs favorably under independent noise, whereas angular masking is more robust under correlated noise and real measured data. Multi-partition splitting consistently improves over pure projection-wise splitting in several settings. Complementary perceptual and structural metrics, including LPIPS and HaarPSI, reveal differences between masking strategies that are less apparent from PSNR and SSIM alone. These results provide practical guidelines for designing self-supervised sparse-view CT reconstruction methods and highlight the limitations of common independence assumptions in realistic imaging environments.