creator Duan, Bin 31633460 2024 Spring 2024 Dissertation Illinois Institute of Technology CS / Computer Science advisor Yan, Yan Computer science Biomedical Image Processing Biomedical Image Toolkit Image Compression Image Registration Neuron Segmentation Neuron Tracing en Biomedical imaging is fundamental to advancing medical research and clinical diagnostics, offering critical insights into complex biological structures and processes. However, the inherent complexity and variability of biomedical images demand the development of specialized tools and algorithms for accurate and efficient analysis. In this thesis, we present a suite of solutions aimed at addressing two central challenges in biomedical imaging: efficient data storage and advanced image analysis.First, we develop high-ratio compression techniques that drastically reduce the storage requirements of biomedical images while preserving their analytical fidelity. These methods enable us to archive and manage vast datasets more efficiently, without compromising the critical details necessary for research and diagnostics. By minimizing compression artifacts, we ensure the integrity of the images, allowing for fast data transmission and seamless long-term storage without loss of quality.For image analysis, we introduce advanced algorithms that significantly enhance the precision and performance of key imaging tasks. Our segmentation methods, leveraging multi-scale non-local correlations, allow us to accurately delineate complex tissues and cellular structures in challenging biological images. In the area of neuronal tracing, we create algorithms that improve the accuracy of mapping intricate connections in densely labeled multi-spectral datasets, providing deeper insights into biological networks. Furthermore, we propose a robust image registration algorithm that corrects alignment errors in multi-modal and longitudinal datasets, ensuring precise and reliable integration for downstream analyses.By combining these innovations, we offer a practical toolkit that streamlines both the storage and analysis of biomedical images. Our work has the potential to significantly enhance research and diagnostic processes, providing tools that improve efficiency, accuracy, and scalability in biomedical imaging. born digital application/pdf In Copyright http://rightsstatements.org/page/InC/1.0/ Restricted Access