Abstract:
Artificial intelligence (AI) has significantly enhanced medical diagnostics, particularly through medical imaging. However, these images often suffer from noise due to various factors like reduced radiation exposure. Efficient noise reduction is crucial for accurate diagnosis. This master thesis addresses the problem of denoising medical images using Convolutional Autoencoders (CAEs). CAEs leverage the power of Convolutional Neural Networks (CNNs) to effectively distinguish between noise and essential diagnostic information. By training on large datasets, CAEs learn intricate noise patterns specific to different imaging modalities, preserving critical anatomical details. The proposed method aims to improve image clarity, ensuring reliable diagnostics and better healthcare outcomes. This study demonstrates the potential of CAEs in enhancing the quality of medical imaging, thereby supporting more precise medical evaluations.
