Abstract:
"The RadSult project addresses critical challenges in radiology, such as diagnostic inaccuracies, data overload, and access disparities, by developing a machine learning-based decision support system (DSS) tailored for brain tumor detection and classification. By employing sophisticated ML algorithms, including convolutional neural networks and advanced feature selection methods, the system automates the analysis of MRI and CT imaging data, offering precise tumor detection, classification, and visual insights to assist radiologists. RadSult is designed with scalability and interoperability, ensuring seamless integration with existing healthcare IT systems while maintaining compliance with stringent data security standards like HIPAA and GDPR.
The dissertation provides a comprehensive approach to tackling variability in imaging quality, algorithmic biases, and regional disparities in healthcare resources. It includes an empirical evaluation of the DSS, leveraging iterative development methodologies and real-world testing in diverse clinical environments. The project emphasizes user-centric design, ensuring accessibility for radiologists of varying expertise levels, and incorporates extensive ethical considerations to maintain transparency and equity in healthcare applications.
Through its novel algorithmic frameworks, robust validation processes, and focus on ethical and professional standards, RadSult contributes significantly to the integration of artificial intelligence in medical diagnostics. The outcomes of this research have the potential to standardize and elevate the accuracy and efficiency of radiological assessments globally, particularly in resource-limited settings, thereby improving patient care and outcomes."
