ITDS | SPMS

STROKE DETECTION AND CLASSIFICATION USING CONVOLUTIONAL NEURAL NETWORK

Project Description
Supervisor

In Ghana and many developing and less developed countries in the world stroke is considered to be one of the deadliest diseases we face. It affects a large number of people every day. Which leads to deaths or may lead to permanent symptoms that may affect the daily life of the affected person and people around them affecting them both physically or financially. The traditional method of discovering stroke in disease is not only time consuming is also prone to human error. It requires professionals like radiologists to scan the brain the for using medical imaging modalities and doctors who spend several hours to look through the image to detect the presence of stroke. But in cases where there are areas that do not have access to these resources the stroke would either not be found out or the discovery would take longer than usual making it riskier in situations where there are no available resources. Our project is developed to tackles some of the issues stated in the previous section. We developed a web application that is powered by two deep learning models one that is used to detect the instance of stroke and the other to classify the detected instance between the two main types, which is accessible to not only doctors but any individual who has access to a soft copy of their brain CT scan. The webapp is easy to use faster than the average doctor in detecting stroke and available various places. The user just has to upload the image of their scan to the webapp and it will pass through the detection model before further classifying the image when it detects a stroke. This will help users detect stroke faster, much more accurately and earlier than most trained professionals. We employed a few powerful techniques to ensure that our models’ performance is not only very high but also very effective in real-world situations, we used augmentation to tackle class imbalance and overfitting issues, the use of hyperparameter tuning for finding the best possible combination for our model to perform well and transfer learning to train faster and a deeper model. In all our models performed extremely well compared to other trained models available online and during the research for the project we discovered how important machine learning and AI is in the modern-day medicine and we are glad to contribute to its advancement.

Machine Learning: Machine Learning (ML) research in Computer Science and Information Technology focuses on the development of algorithms and models that enable computers to learn from data and improve their performance over time without being explicitly programmed. It is a subset of Artificial Intelligence that uses statistical techniques to give machines the ability to learn patterns, make decisions, and predict outcomes based on data. Supervised learning, a key area of ML research, involves training models on labeled data, where the input-output relationships are predefined. This method is widely used for tasks such as classification (e.g., spam detection) and regression (e.g., predicting house prices). Unsupervised learning, on the other hand, involves finding hidden patterns in data without predefined labels, with clustering and association being typical applications in areas such as customer segmentation and anomaly detection. Reinforcement learning is another area of ML that focuses on teaching agents to make decisions by interacting with their environment and receiving feedback in the form of rewards or penalties. It is often applied in robotics, game playing, and autonomous systems, where continuous learning and adaptation are required.

To design a web-based application that utilizes deep learning models, that are able to recognize stroke and classify them into their various classes, utilizing convolutional neural networks (CNNs) and leveraging ResNet50, Albumentations, and Random Search to improve accuracy and generalization on data from medical images (e.g. CT images).

2024/2025

Feb 3, 2026

YAKANYA PRINCE UEB3207421, SAEED MUSDEEN UEB3229421, JESSICA ANANE DANQUAH UEB3219821

BSc Information Technology