TBI-6000 : A Labeled Brain CT Dataset for 6 different types of Haemorrhage Classification

TBI-6000 is a curated dataset consisting of 6,000 unique and preprocessed brain CT images, specifically developed to support machine learning research in the automated classification of traumatic brain injury (TBI) and intracranial haemorrhage. The dataset is systematically organized into six balanced subdirectories, each representing a distinct diagnostic category: Epidural Haemorrhage, Subdural Haemorrhage, Subarachnoid Haemorrhage, Intraventricular Haemorrhage, Intraparenchymal Haemorrhage, and No Haemorrhage (Control Class). Each folder contains exactly 1,000 labeled CT slices, and all images are unique within and across folders, ensuring a balanced and non-redundant distribution that enhances the reliability and fairness of model training and evaluation. To maintain data integrity and facilitate seamless integration into AI pipelines, all images have been renamed using a standardized filename convention that encodes each image’s diagnostic class as a prefix (e.g., epidural_, no_hemorrhage_) followed by a unique numeric identifier. This consistent naming scheme promotes reproducibility, simplifies data parsing, and supports efficient downstream processing. TBI-6000 serves as a high-quality resource for the development and benchmarking of deep learning models targeting intracranial haemorrhage detection and classification in the context of traumatic brain injuries.