CSAW-CC (mammography)

With over 1,100 cancer cases and 10,000 healthy controls, the CSAW-CC dataset provides critical resources for researchers and developers working to build AI systems for detecting breast cancer in its early stages. It includes expert annotations of tumors, which are essential for training machine learning models, particularly Convolutional Neural Networks (CNNs), to identify malignant and benign lesions.

Key Features of the CSAW-CC Dataset:

• Image ID: A unique identifier for each mammogram image.
• Age: The patient’s age at the time of the screening.
• Screening Date: The date when the mammogram screening took place.
• Lesion Type: Classification of lesions as benign or malignant.
• Mammogram Images: High-quality images used to train AI systems for accurate cancer detection.
• Annotations: Pixel-level annotations provided by expert radiologists, including the precise location of lesions and micro-calcifications. These annotations include predicted locations for potential tumors in images captured before diagnosis.

Dataset Details:

• Cancer Cases: Over 1,100 breast cancer patients.
• Healthy Controls: 10,000+ healthy individuals.
• Purpose: To advance AI-based breast cancer detection, tumor classification, and risk prediction.

Ethical Considerations:

The CSAW-CC dataset has been ethically reviewed and approved by multiple ethical boards in Sweden:

• Ethical approval by the Ethical Review Board of Stockholm under permission number EPN 2016/2600-31.
• Additional reviews and ethical approvals were granted by the Ethical Review Authority of Sweden under permission EPM 2019-01946.

Benefits for AI & Machine Learning:

• AI Model Training: The dataset is specifically designed for training AI systems to detect early-stage breast cancer and to differentiate between benign and malignant tumors.
• Research Applications: Useful for developing AI-driven diagnostic tools, improving breast cancer prognostics, and advancing predictive models for better patient outcomes.