Lung Cancer CT scan image for Federated Learning
Abstract
This dataset contains 3,200 high-quality slice-level lung CT images categorized into three distinct diagnostic classes. It is designed to support machine learning, deep learning, and federated learning research for the automatic detection and classification of lung cancer, contributing to innovations in privacy-preserving medical AI, multi-institutional collaboration, and explainable diagnostic systems. Dataset Composition: The dataset is organized into three distinct classes based on standard radiological features and pathological definitions: ⦁ Malignant (1,310 images): Confirmed carcinomas, such as Adenocarcinoma, typically characterized by irregular margins, nodular densities, and surrounding parenchymal distortion. ⦁ Benign (905 images): Non-cancerous nodules, such as Hamartomas or Granulomas, characterized by circumscribed structures, smooth margins, or calcification. ⦁ Normal (985 images): Scans exhibiting clear lung parenchyma with no evidence of nodules or pathological focal points. Data Preprocessing and Structuring: ⦁ Images are formatted as single-channel (grayscale) CT slices. ⦁ The dataset underwent a standardized preprocessing pipeline where images were resized to 224×224 pixels to ensure uniformity across samples. ⦁ Scans were enhanced via Contrast Limited Adaptive Histogram Equalization (CLAHE) to handle differences in brightness and contrast, and normalized using shared per-channel mean and variance to ensure consistent pixel intensity (range of 0 to 1).…
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.