AISSLab Breast Cancer Sub-datasets: Task-Oriented Structuring of Mammographic Images for Classification Scenarios

The AISSLab Breast Cancer Subdatasets provide a structured and task-oriented collection of mammographic images designed for AI-based classification research.
The dataset is organized into Raw and Augmented subdatasets, enabling effective training and evaluation of deep learning models.

Sub-datasets overview:

AISSLab-v1: Normal (100) vs. Abnormal (166) → 266 raw images → 1064 augmented images

AISSLab-v2: Normal (100) vs. Malignant (100) → 200 raw images → 800 augmented images

AISSLab-v3: Benign (66) vs. Malignant (100) → 166 raw images → 664 augmented images

AISSLab-v4: Benign (66), Malignant (100), Normal (100) → 266 raw images → 1064 augmented images

AISSLab-v5: BI-RADS 2 (18), BI-RADS 3 (48), BI-RADS 4 (54), BI-RADS 5 (46) → 166 raw images → 664 augmented images

This modular design allows researchers to experiment with binary and multiclass classification tasks, while the applied data augmentation helps increase sample size and reduce class imbalance.

The dataset complies with the FAIR Data Principles to ensure reusability, transparency, and interoperability.
This version transforms AISSLab into a modular and reusable benchmark for medical AI researchers working on classification tasks in mammography.

Citation

Researchers utilizing the AISSLab Subdatasets are kindly requested to cite the following works:

A. M. Al-Hejri, R. M. Al-Tam, M. Fazea, A. H. Sable, S. Lee, and M. A. Al-antari, “ETECADx: Ensemble Self-Attention Transformer Encoder for Breast Cancer Diagnosis Using Full-Field Digital X-ray Breast Images,” Diagnostics, vol. 13, no. 1, Jan. 2023, doi: 10.3390/diagnostics13010089.

R. M. Al-Tam, A. M. Al-Hejri, S. S. Alshamrani, M. A. Al-antari, and S. M. Narangale, “Multimodal breast cancer hybrid explainable computer-aided diagnosis using medical mammograms and ultrasound Images,” Biocybern Biomed Eng, vol. 44, no. 3, pp. 731–758, Jul. 2024, doi: 10.1016/j.bbe.2024.08.007.

Contact

For any inquiries regarding the dataset, please feel free to reach out via email: [e-mail]