## About Dataset

The dataset consists of heart block images, along with associated masks and annotations, aimed at tasks like:

Segmentation: Identifying and delineating heart blockage areas.

Classification: Determining the presence or type of blockage.

Detection: Locating heart blockages within the images.

The dataset's hierarchical structure ensures compatibility with standard machine learning pipelines, providing well-defined splits for training, validation, and testing.

### Directory Structure

Root Directory

The root directory (heartblock/) contains three primary splits:

train/: Data used for model training.

val/: Data used for validation during model development.

test/: Data reserved for evaluating model performance.

### Subdirectories within Each Split

Each of the train/, val/, and test/ directories contains the following standardized subdirectories:

images/:

Contains raw or preprocessed images of heart block cases. These images serve as the primary input for the models and are typically medical imaging scans (e.g., X-rays, MRIs, or CT scans) showing heart blockages.

masks/:

Contains binary or multi-class segmentation masks corresponding to the images. These masks highlight the regions of interest (e.g., heart block areas) and are used in supervised learning tasks like segmentation.

annotations/ (or ann/):

Holds metadata or label files, often in formats like .json or .xml. These annotations include details such as:

Bounding box coordinates (for detection tasks).

Class labels (for classification tasks).

Pixel-wise information (for segmentation tasks).

Consistency Across Directories

Each split (train, val, test) has the same number of subdirectories in its images/, masks/, and annotations/ folders. This ensures:

One-to-One Mapping: Each image in images/ has a corresponding mask in masks/ and annotation file in annotations/.

Ease of Use: Facilitates seamless pairing of inputs (images) with their ground truths (masks and annotations) for efficient data loading.

Applications

The heartblock dataset is designed for various medical imaging tasks, including but not limited to:

Heart Block Detection: Identifying the presence and location of blockages. Segmentation: Extracting precise areas affected by heart blockages. Classification: Determining the type or severity of the blockage. Diagnosis Support: Assisting healthcare professionals in accurate and automated diagnosis.