Brain hemorrhage, a type of bleeding that occurs in the brain due to the bursting of blood vessels as an after effect of trauma to the head, high blood pressure, Aneurysm, Amyloid Angiopathy or other reasons (source) causes more than 20000 people in the USA alone to lose their lives annually.
Since oxygen saturated blood does not reach some brain cells, they die after three or four minutes, and their corresponding functions are also shutdown. Besides the death of oxygen-deprived cells, other cells that are pressured from the bleeding occurring will also be damaged along with their neural functions. source
From here we can see that fast and accurate detection of such cases is vital to reduce the severity of the after effects and facilitate the treatment of patients, and consequently help save their lives.
Brain hemorrhage has different types depending on the region of the brain where the bleeding occurs and can be divided into different classes or categories. Source
Some research has been adopted for the classification and detection of brain hemorrhage but still their results do not exceed the accuracy of 90% in addition to the fact that the data required to train the models are scarce and quite expensive to preprocess.
In this research we aim to develop ML models for accurate classification and detection of brain hemorrhage while leveraging the unprocessed data available.