An efficient False Negative Reduction System for the Identification of Intracranial Hemorrhage

A rapid diagnosis and efficient treatment are required for the dangerous condition known as intracranial hemorrhage (ICH). In the emergency analysis of reports of ICH, computerized tomography (CT) is frequently used as it gives better results when compared to MRI. The treating physician must be knowledgeable in neuroimaging to regulate the origin of the hemorrhage and its location and severity. However, in this paper, a new deep-learning method was proposed to improve detection accuracy by reducing false negative cases. We are mainly focused on the reduction of false negative cases as this condition is dangerous. Finally, with the proposed model we achieved an accuracy of 95%, precision of 90.9%, recall of 100%, F1-Score of 95.23%, and specificity of 90.0 on the test data. With a less complex and computationally intensive technique, our results are in line with the state of the art. Finally, our detector could help doctors make better medical decisions, make better use of available resources, and allow us to diagnose ICH more quickly and accurately.