Use of the decision trees to enhance the performance of deep learning models for stroke classification.

Stroke has always been one of the leading causes of death around the world. Timeless diagnostics and intervention are crucial to prevent lethal consequences. The quickest way to get some initial results is the use of non-contrast CT scans (NCCT). Traditional methods rely solely on radiologists and require a lot of time for careful analysis and diagnostics. Therefore, the use of computer-aided diagnostics (CADs) became more common. This paper reviews some of the approaches that were proposed in recent years for the 2d classification and tests them on 2 main types of stroke, namely hemorrhagic and ischemic. Additionally, it proposes and analyzes a new approach that relies on the use of Deep Learning models for feature extraction and Decision trees for the final classification. In this paper 2 private datasets obtained from CerebraAI are used, one for each type of stroke were used for the experiments. The results obtained during the experiments did not demonstrate a significant change across different metrics. More experiments on more accurately supervised datasets are required to make a final verdict about the effectiveness of such an approach.