Article of the Week: November 17

Medical imaging and nano-engineering advances with artificial intelligence

Author

Marina Pulini-Franks MD
November 17, 2023

Medical imaging and nano-engineering advances with artificial intelligence

Medical imaging is a broad field of research and artificial intelligence used to explore such images is termed as AI-Imaging. AI-imaging is further divided into sub-branches including the computational, theoretical and practical experiments in wet and dry labs. The current research focuses on the background of medical imaging, recent advances in the field of medical imaging for oncology, challenges and possible solutions. During this research, some computational and programing tools are outlined. The process of image segmentation is important as it can help to explore the medical images in more detail. During this research, the steps involved in image segmentation are outlined and the numerical experiments are performed on a set of breast cancer medical images. It is concluded during this research that the achievements in this domain are always credited by the smart programing & computational tools and computer vision. The current research also outlines the step-wise protocols of deep learning, designed for different types of medical imaging such as X-rays, CT-scan and MRI are documented to provide a comprehensive understanding, that can help in bridging the two domains of medicine and computer vision, in a reliable and fruitful manner.

journals.sagepub.com/doi/abs/10.1177/23977914231161443

Main Point

This scholarly article discusses the evolution and significance of artificial intelligence (AI) and deep learning in the field of medical imaging. It traces the historical development of AI techniques, from early "if-then rules" to the advent of convolutional neural networks (CNNs) and radiomics. The article highlights the use of AI for data preprocessing, software development, and time series forecasting in medical imaging, along with the challenges and advancements in this domain.

5 Salient Points

  • Historical Evolution of AI in Medical Imaging: The article traces the development of AI in medical imaging, from early "if-then rules" to the recognition of deep learning in 2012 with the success of convolutional neural networks (CNNs) in ImageNet Classification. It emphasizes the pioneering works of Alan Turing's "Turing Test" and the early application of machine learning in medical imaging during the 1980s.

  • Advancements in Deep Learning and Applications: The article discusses the profound impact of deep learning algorithms, such as U-net and Holistically Nested Network (HNN), in medical imaging. These algorithms have been extensively used for organ and tumor segmentation, improving diagnostic accuracy. It highlights the efficiency of AI-driven software tools in enhancing the interpretation of X-rays, CT scans, and MRIs.

  • Importance of Data Preprocessing: Data preprocessing is crucial in medical imaging to clean and enhance data before applying AI and deep learning algorithms. Image preprocessing techniques, such as noise removal and feature enhancement, significantly contribute to the accuracy of medical image analysis.

  • Radiomics and Feature Extraction: The article introduces the concept of radiomics, where diverse features are extracted from segmented medical images, including texture, shape, density, and pixel intensity. These features are utilized in computational models to decode tumor phenotypes and predict treatment outcomes.

  • Challenges and Future Outlook: The article addresses challenges in AI adoption, including data collection, standardization, and selecting appropriate models for specific medical imaging tasks. Despite challenges, the article acknowledges the promising future of AI in medical imaging, with ongoing advancements in deep learning and the potential to revolutionize medical diagnostics.

🎯Research Appendix