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

The complete dataset is divided into two parts, Part 1 and Part 2, containing a total of 4,407 knee MRI study samples. This dataset is Part 2, which includes the remaining 1,207 study samples. The data resamples the original DICOM medical images into fixed-size axial slice stacks, with each sample uniformly consisting of 44 images of 336 × 336 uint8 format, ready for direct use in medical imaging model training.

**Part1：Knee‑Raptor‑Corpus Knee MRI Preprocessed Stack Dataset

Dataset Composition

Part 2 includes the following 4 core files:

• extra_vols.npy: MRI image volume data.
• extra_ids.npy: StudyInstanceUID corresponding to each image.
• extra_masks.npy: Image mask data.
• extra_meta.json: Sample metadata.