Date on Master's Thesis/Doctoral Dissertation
12-2024
Document Type
Master's Thesis
Degree Name
M.S.
Department
Bioengineering
Committee Chair
El-Baz, Ayman
Committee Member
Giridharan, Guruprasad
Committee Member
Elmaghraby, Adel
Author's Keywords
Kidney-transplant; renal rejection; artificial intelligence; biomarkers
Abstract
They kidney is a vital organ for which humans are fortunate to find a spare through transplantation to sustain critical body functions, offering a hope to those struggling with renal failure. Kidney transplant procedure is the optimal treatment for people who suffer from end-stage renal failure. However, there are posed challenges due to the risk of immune rejection and the limited availability of donors. Early detection of renal rejection can provide timely intervention and accurate diagnosis that are critical to improve the transplant outcomes. This study explores the innovative approaches for addressing the current challenges through biomarkers identification, imaging techniques, and provide accurate classification of renal rejection based on the advanced deep learning models. The first chapter provides an overview of renal rejection in kidney-transplant patients, highlighting current treatments, challenges, and the research goals behind this study. It emphasizes the importance of improving renal rejection diagnosis to support better patient outcomes. In the second chapter, it reviews biomarkers for kidney transplant rejection, including histopathological, clinical, and genetic markers used over the past ten years. The focus is on identifying key markers that predict renal rejection and differentiate between its subtypes. This chapter aims to explore essential markers for diagnosing each rejection subtype. The third chapter introduces a novel approach for 3D renal segmentation in kidney-transplant patients using multi-modal imaging (Bold-MRI, T2-Weighted Imaging, and DW-MRI). By applying a fusion of shape features and Generative Adversarial Networks (GANs), the framework accurately segments kidney structures. Testing was conducted on 57 Bold-MRI, 60 T2-Weighted Imaging, and 86 DW-MRI patients, providing precise binary masks generation that help diagnose patients at risk of rejection. The fourth chapter proposes an AI-driven approach for diagnosing renal rejection by integrating genomic and clinical markers to differentiate between antibody-mediated rejection (ABMR) and T-cell-mediated rejection (TCMR). Utilizing a transformer-based classifier, the system combines feature reduction on genomic data with clinical data. The proposed computer-aided diagnostic (CAD) system achieved an overall accuracy of 94%, offering a safer alternative to biopsies while maintaining high diagnostic precision. Finally, the last chapter concludes the study by summarizing the methodologies and findings. It highlights the clinical significance of the proposed frameworks and provides future directions for improving renal rejection diagnosis. Additionally, it emphasizes the potential of combining advanced imaging, biomarker discovery, and AI techniques to enhance patient outcomes and streamline non-invasive diagnostic methods.
Recommended Citation
Sharaby, Israa, "AI-driven approach for diagnosis of renal transplant rejection based on biomarkers identification and integration." (2024). Electronic Theses and Dissertations. Paper 4486.
Retrieved from https://ir.library.louisville.edu/etd/4486
