Brain tumor MRI image for Fedarated Learning
Abstract
This dataset contains 10,417 de-identified, single-channel (greyscale) brain MRI slices. It is designed to support machine learning and computer vision research for robust, four-class brain tumor classification, particularly contributing to innovations in privacy-preserving federated learning and explainable AI (XAI) frameworks. Dataset Composition: The dataset is organized into four distinct, clinically meaningful categories based on pathology. ⦁Glioma tumor (2,547 images): MRI slices exhibiting glioma tumors. ⦁Meningioma tumor (2,712 images): MRI slices exhibiting meningioma tumors. ⦁Pituitary tumor (2,658 images): MRI slices exhibiting pituitary tumors. ⦁No tumor (2,500 images): Healthy, tumor-free brain MRI slices. Preprocessing: ⦁Slices are harmonized to 224x224 pixels and loaded as single-channel grayscale images. ⦁A unified preprocessing pipeline was applied, including resizing, optional center/zero-padding, and per-image normalization to ensure consistency. ⦁For federated learning simulations, the dataset is pre-partitioned across four clients and further divided into an 80/10/10 (train/validation/test) split conducted patient-wise to prevent data leakage. This dataset can be effectively used for: ⦁Multiclass image classification and brain tumor recognition tasks. ⦁Deep learning model development, including parameter-efficient CNNs, deeper CNNs, and hybrid architectures. ⦁Benchmarking synchronous Federated Averaging (FedAvg) and privacy-preserving multi-site training…
Read the paper
Related studies
- Active Learning during Federated Learning for Object Detection
- Fairness-aware federated learning
- Decentralized Federated Learning
- Building a Personally Identifiable Information Recognizer in a Privacy Preserved Manner Using Automated Annotation and Federated Learning
- Preservation of Patient Level Privacy: Federated Classification and Calibration Models
- Efficient explainable federated learning
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.