Acquired Brain Injury (ABI), is an alarming medical condition caused by ruptured or a lacking blood supply to the skull, that determines the neuronal activity of a human. Ischemic strokes account for more than 70% of all reported strokes cases. In recent times the mortality rate due to this condition has increased manifold. This study aims to efficiently improve the performance of detection of ABIs through CerebroNet, a novel stroke detection framework integrating the U-Net architecture which assists in understanding spatial hierarchies. This includes an encoder-decoder structure with skip connections for fine-tuning the dataset. The final layer uses an activation function for classification, identifying stroke regions in the image. By employing brain imaging datasets from kaggle sources, the framework enhances segmentation and classification tasks through convolutional image recognition pipelines. Cerebro-Net achieved and demonstrated an accuracy of above 97% compared to traditional machine learning models like Random Forest, SVMs and Decision Trees. This study ultimately aims to solve the issues of limited labeled data of cerebral activity and the ability of certain learning models to handle variability and to prevent overfitting.