A Dataset for Breast Cancer Histopathological Image Classification.

Area of Science:

• Medical Image Analysis
• Computational Pathology
• Machine Learning in Healthcare

Background:

• Standardized datasets are crucial for validating medical image analysis methods.
• Current breast cancer image datasets suffer from heterogeneity in acquisition and evaluation, hindering method comparison.
• Automated classification of histopathology images holds promise for computer-aided diagnosis.

Purpose of the Study:

• To introduce a large, publicly available dataset of breast cancer histopathology images.
• To establish a standardized evaluation protocol for automated image classification tasks.
• To facilitate collaboration between medical and machine learning researchers for clinical applications.

Main Methods:

• Compilation of 7909 breast cancer histopathology images from 82 patients.
• Inclusion of both benign and malignant image samples.
• Preliminary evaluation using state-of-the-art image classification systems.

Main Results:

• The dataset is publicly accessible at http://web.inf.ufpr.br/vri/breast-cancer-database.
• Preliminary classification accuracy ranged from 80% to 85%, indicating potential for improvement.
• The dataset supports a two-class automated classification task (benign vs. malignant).

Conclusions:

• The newly released dataset and protocol will enable reproducible research in breast cancer image analysis.
• Advancements in automated classification can lead to valuable computer-aided diagnosis tools for clinicians.
• This initiative aims to foster interdisciplinary research to improve breast cancer diagnosis and treatment.