A brain hemorrhage is an eruption of the brain's arteries brought on by either excessive blood pressure or blood coagulation, which may result in fatalities or serious injuries. It is the kind of medical emergency that requires a clinician to quickly identify the site of internal bleeding before beginning therapy. Convolutional Neural Network (CNN) and CNN + LSTM hybrid models for deep learning are suggested in this study for the categorization of brain hemorrhages. The 200 head CT scan images dataset is utilized to increase the deep learning models' precision and processing capability. Because big datasets are rarely immediately available in essential situations, the main goal of this work is to apply the abstraction capability of deep learning on a smaller amount of pictures A unique architecture called Brain Haemorrhage Classification based on Neural Network was made utilizing the CNN model together with picture augmentation and dataset misbalancing approaches. The performance of the recommended technique is assessed using accuracy, precision, sensitivity, specificity, and F1-score. The experimental results are further evaluated utilizing comparative analyses of the balanced and unbalanced dataset using CNN and CNN + LSTM. Unbalancing the dataset yields encouraging results, outperforming CNN in accuracy. The results show the effectiveness of the proposed method for quick implementation in real-world circumstances and precise prediction to preserve the patient's life in the interim.
