Development of a Neural Network Algorithm for The Detection of Internal Haemorrage

Abstract

Even with the best treatment possible, the prognosis for intracranial haemorrhage (ICH), a potentially fatal medical emergency, is dismal. There should be a triage mechanism in place to quickly identify and expedite the treatment of ICH since doing so may enhance health outcomes. Prior research used a more conventional approach, which included a long series of procedures including alignment, analysis, correction, segmentation, and classification by hand. This study investigates the challenge of detecting cerebral haemorrhages and proposes a deep learning and transfer learning model to speed up the process. We built a convolutional neural network using the Transfer Learning Model to classify ICH subtypes. against guarantee the accuracy and greatness of the model's outcomes, it was compared against DenseNet121, Xception, and CNNs utilising a battery of assessment criteria. Xception outperforms rival models, and the technology generates outstanding results, as expected. In order to determine whether ICH subtypes exist, the Xception model is used.