LoDoPaB-CT Glimpse: Processed CT Image Slices for Reconstruction Research

Processed image subsets from the LoDoPaB-CT dataset, used by the GLIMPSE research paper. The dataset contains training and test slices stored as .npy float32 arrays, along with out-of-distribution brain images in .jpg format. It was uploaded by AmirEhsan1995 to Hugging Face and was last updated on June 7, 2026.

Use Cases

• Training image reconstruction models based on the provided .npy training slices.
• Evaluating model generalization using the separate test set of CT slices.
• Testing model robustness on out-of-distribution data based on the provided .jpg brain images.

Strengths

• Data is split into dedicated train and test sets, facilitating standard ML workflows.
• Images are provided in a processed format (.npy arrays) ready for computational use.
• Includes a specific 'ood' (out-of-distribution) split for robustness testing, as mentioned in the description.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

Derived from the LoDoPaB-CT dataset, processed for the GLIMPSE research pipeline.

Collection Method

Subsets were processed and rendered by the GLIMPSE data pipeline using ODL / scikit-image.

Freshness

Last updated 2026-06-07 09:47:07; freshness should be verified.