Dark Mode
Human Sperm Shape Classification Data
Patient Health Records & Digital Health
Tags and Keywords
Trusted By
"No reviews yet"
Free
About
The Human Sperm Head Morphology dataset (HuSHeM) provides classified images of human sperm heads crucial for automated morphology analysis and research in fertility and andrology. The data was meticulously collected from semen samples of fifteen patients at the Isfahan Fertility and Infertility Center. Sperm heads were extracted from initial high-resolution images, fixed and stained using the Diff-Quick method, and subsequently classified by three independent medical specialists. Only samples where there was collective consensus on the shape classification were retained, ensuring high data quality for machine learning and diagnostic support systems.
Columns
This product is structured as an image dataset organized into four distinct directories, representing the classification of the sperm heads. Each directory contains individual RGB images (131x131 pixels) corresponding to a specific shape class:
- 01_Normal: Contains images of sperm heads exhibiting normal morphology (54 files).
- 02_Tapered: Contains images of tapered sperm heads (53 files).
- 03_Pyriform: Contains images of pyriform or pear-shaped sperm heads (57 files).
- 04_Amorphous: Contains images of amorphous or irregularly shaped sperm heads (52 files).
Distribution
The dataset consists of four directories holding a total of 216 individual sperm head images. Each image file is in RGB format with a fixed resolution of 131x131 pixels. The initial images from which these heads were cropped had a resolution of 576x720 pixels, captured using an Olympus CX21 microscope with a Sony color camera. The distribution across classes is relatively balanced: 54 Normal, 53 Tapered, 57 Pyriform, and 52 Amorphous heads. The expected update frequency for this dataset is never.
Usage
This dataset is ideal for:
- Training and testing deep learning models for automated sperm morphology classification.
- Research focused on computer-assisted diagnosis in fertility clinics.
- Developing image processing techniques specific to biological cell analysis.
- Academic projects in pattern recognition and medical imaging.
Coverage
The data was sourced from semen samples provided by 15 patients at the Isfahan Fertility and Infertility Center. The coverage is geographically specific to this center. The classification methodology included an initial consideration of five classes (Normal, Pyriform, Tapered, Amorphous, and Others); however, only the four most established classes are included in the final product.
License
CC BY-SA 4.0
Who Can Use It
- Image Processing Engineers: For developing new algorithms for morphological feature extraction.
- University Researchers: For studies in andrology and reproductive health conditions.
- Healthcare Technology Developers: For building AI tools to assist fertility specialists.
- Data Science Students: For learning classification techniques on high-quality biological image data.
Dataset Name Suggestions
- HuSHeM Sperm Head Morphology Image Collection
- Human Sperm Shape Classification Data
- Andrology Morphology Analysis Dataset
Attributes
Original Data Source: Human Sperm Shape Classification Data
QUALITYLoading...