Biomedical Image Reconstruction: A Survey

Samuel Cahyawijaya
2023-01-04
Abstract:Biomedical image reconstruction research has been developed for more than five decades, giving rise to various techniques such as central and filtered back projection. With the rise of deep learning technology, biomedical image reconstruction field has undergone a massive paradigm shift from analytical and iterative methods to deep learning methods To drive scientific discussion on advanced deep learning techniques for biomedical image reconstruction, a workshop focusing on deep biomedical image reconstruction, MLMIR, is introduced and is being held yearly since 2018. This survey paper is aimed to provide basic knowledge in biomedical image reconstruction and the current research trend in biomedical image reconstruction based on the publications in MLMIR. This survey paper is intended for machine learning researchers to grasp a general understanding of the biomedical image reconstruction field and the current research trend in deep biomedical image reconstruction.
Image and Video Processing,Computer Vision and Pattern Recognition
What problem does this paper attempt to address?
The paper primarily explores the development history, current trends, and future research directions in the field of biomedical image reconstruction. Specifically, the paper aims to: 1. **Provide Basic Knowledge**: Offer readers fundamental knowledge of biomedical image reconstruction, covering various classical techniques from traditional analytical methods (such as the central slice theorem, filtered back projection) to iterative methods (such as algebraic reconstruction techniques, maximum likelihood expectation maximization, etc.). 2. **Overview of Technological Evolution**: Describe the paradigm shift in the field of biomedical image reconstruction in recent years due to advancements in deep learning technology. The paper highlights the advantages of deep learning-based methods in improving image quality and processing speed and introduces several commonly used deep learning architectures (such as U-Net, Generative Adversarial Networks GAN, etc.). 3. **Analyze Research Trends**: Reveal current research hotspots and potential future directions by statistically analyzing articles published over the years at the Machine Learning for Medical Image Reconstruction (MLMIR) workshop. For example, most work focuses on Magnetic Resonance Imaging (MRI) and tomographic imaging, while self-supervised pre-training methods and Transformer architectures are gradually gaining attention. 4. **Evaluate Methods and Datasets**: Discuss the datasets and evaluation metrics used to assess the effectiveness of biomedical image reconstruction, such as Structural Similarity Index (SSIM), Peak Signal-to-Noise Ratio (PSNR), and Root Mean Square Error (RMSE). 5. **Future Work Outlook**: Predict that in the coming years, self-supervised pre-training, U-Net variants, and Transformer models may become mainstream research directions, and point out some challenges that still need to be addressed in practical applications, such as high computational resource consumption.