MedMNIST v2 - A large-scale lightweight benchmark for 2D and 3D biomedical image classification

TLDR

MedMNIST v2 is a large‑scale MNIST‑like collection of standardized 2D and 3D biomedical images, comprising 12 2D and 6 3D datasets designed for lightweight classification across a wide range of scales and tasks. Images are pre‑processed to 28 × 28 (2D) or 28 × 28 × 28 (3D) with labels, and baseline 2D/3D neural networks and AutoML tools are benchmarked on the collection. The dataset contains 708,069 2D and 9,998 3D images and is publicly available with accompanying code at https://medmnist.com/.

Abstract

Abstract We introduce MedMNIST v2 , a large-scale MNIST-like dataset collection of standardized biomedical images, including 12 datasets for 2D and 6 datasets for 3D. All images are pre-processed into a small size of 28 × 28 (2D) or 28 × 28 × 28 (3D) with the corresponding classification labels so that no background knowledge is required for users. Covering primary data modalities in biomedical images, MedMNIST v2 is designed to perform classification on lightweight 2D and 3D images with various dataset scales (from 100 to 100,000) and diverse tasks (binary/multi-class, ordinal regression, and multi-label). The resulting dataset, consisting of 708,069 2D images and 9,998 3D images in total, could support numerous research/educational purposes in biomedical image analysis, computer vision, and machine learning. We benchmark several baseline methods on MedMNIST v2, including 2D/3D neural networks and open-source/commercial AutoML tools. The data and code are publicly available at https://medmnist.com/ .