Knee-Raptor-Corpus is a derived preprocessed dataset obtained by preprocessing the original fastMRI knee MRI data, designed to provide standardized preprocessed samples for deep learning tasks on knee MRI imaging.

The complete dataset is divided into Part 1 and Part 2, containing a total of 4,407 knee MRI studies. This dataset is Part 1, which includes the first 3,200 studies, while Part 2 contains the remaining 1,207. Each study is processed into a stack of 44 images of 336 × 336 uint8 format, with slices from sagittal, coronal, and axial sequences filled in a fixed order. The dataset does not include labels; only 58 studies come with structured labels, while the remaining studies are accompanied by free-text radiology reports.

**Part 2: Knee-Raptor-Corpus-Ext Knee MRI Slice Dataset

Dataset Composition

Part 1 includes the following 4 core files:

• all_vols.npy: MRI image stack data.
• all_masks.npy: Slice validity masks.
• all_ids.npy: Corresponding StudyInstanceUID.
• window_index.npy: Three-slice window indices.