Dark Mode
Breast Tumor Segmentation and Classification Dataset
Patient Health Records & Digital Health
Tags and Keywords
Trusted By
"No reviews yet"
Free
About
A valuable collection of ultrasound images focusing on breast tumors, specifically labelled as either benign or malignant. This product is designed to facilitate robust research and development efforts in medical imaging, particularly concerning the automated segmentation and classification of breast malignancies. The raw images were sourced from established clinical examples and subsequently annotated by an experienced radiologist, ensuring accurate classifications for segmentation and classification model training.
Columns
The supporting metadata file (e.g., CSV) provides structural details about the image collection, including:
- levels: Indicates distinct levels within the data structure (9 total values).
- Name: Unique identifiers for entries (9 unique values).
- Type: Specifies the nature of the entry, predominantly indicating directories.
- Description: Provides context, with the most common entry being "Root directory."
- File Count: Details the quantity of files associated with a given entry, ranging from 2 up to 811 total files.
Distribution
The collection consists of 811 high-quality breast ultrasound images. All images are standardised to a resolution of 256 × 256 pixels. The image types are distributed into two primary classes: 358 images representing benign breast tumors and 453 images representing malignant breast tumors. The data product has a high usability rating of 10.00 and is not expected to receive future updates.
Usage
This collection is ideally suited for several key applications in medical artificial intelligence:
- Breast Tumor Classification: Training and evaluation of machine learning and deep learning models to distinguish between benign and malignant tumors using ultrasound scans.
- Breast Tumor Segmentation: Developing algorithms aimed at accurately delineating tumor regions within the ultrasound images.
- Comparative Studies: Assessing the performance metrics of different computational techniques applied to breast ultrasound analysis.
- Educational Purposes: Providing labelled data for learning and experimentation in the field of medical image analysis for students and researchers.
Coverage
The scope is focused strictly on medical imagery pertaining to breast tumors, categorised by malignancy. The data represents clinical ultrasound examples derived from the source material. No specific geographical location or time range is noted, but the context is global clinical research imagery.
License
Creative Commons Attribution 4.0 International License (CC BY 4.0).
Who Can Use It
Intended users include:
- Artificial Intelligence Researchers: For developing and refining deep learning models for clinical diagnostics.
- Medical Image Analysts: For developing automated tools for segmentation.
- Academics and Students: For educational projects and academic experimentation in computer vision and medical technology.
Dataset Name Suggestions
- BUS_UC Breast Ultrasound Images
- Annotated Benign and Malignant Breast Ultrasound Data
- Breast Tumor Segmentation and Classification Dataset
Attributes
Original Data Source: Breast Tumor Segmentation and Classification Dataset
QUALITY