Date of Award
8-1-2026
Degree Name
Master of Science
Department
Electrical and Computer Engineering
First Advisor
Lu, Chao
Abstract
Microscopy image denoising plays a vital role in enhancing the quality and interpretability of biological imaging data, particularly under low-light acquisition conditions. This thesis presents a two-stage supervised deep learning framework to improve denoising performance while preserving fine structural details. The proposed method first utilizes a convolutional encoder–decoder architecture to generate an initial denoised reconstruction from noisy microscopy images. A subsequent residual refinement stage is introduced to further enhance the output by learning targeted corrections, enabling improved recovery of subtle structures. This two-stage design separates coarse reconstruction from detail refinement, providing an effective approach to image restoration. The framework is evaluated on paired microscopy datasets, including live-cell and fixed-cell scenarios. Experimental results demonstrate consistent improvements across multiple quantitative metrics, including mean absolute error (MAE), peak signal-to-noise ratio (PSNR), multi-scale structural similarity (MS-SSIM), and perceptual similarity (LPIPS). The refinement stage contributes to enhanced structural fidelity and perceptual quality while maintaining computational efficiency. Overall, this work presents a practical and well-structured denoising pipeline that combines strong initial reconstruction with targeted residual enhancement, offering an effective solution for high-quality microscopy image restoration.
Access
This thesis is only available for download to the SIUC community. Current SIUC affiliates may also access this paper off campus by searching Dissertations & Theses @ Southern Illinois University Carbondale from ProQuest. Others should contact the interlibrary loan department of your local library or contact ProQuest's Dissertation Express service.