This Research Topic features submitted papers from the MIUA 2022 conference, a UK-based international conference for the communication of image processing and analysis research and its application to medical imaging and biomedicine.
AI is currently sweeping the academic and industrial worlds. It mixes classic scientific mechanistic modelling (differential equations) with conventional machine learning and innovative deep learning techniques. However, deep learning, as is widely known, suffers from challenges like poor interpretability and lack of physical constraints; merging such approaches with numerical analysis and differential equations might lead to a new field of research via novel methods, structures, and algorithms.
Physics-informed deep learning strategies seek to address the limitations of data-driven approaches, such as (i) the large quantity of data required from data-driven models to detect and analyze medical data, and (ii) the development and collecting of clinical data that frequently does not meet the physical requirement. It is emerging that adding physical models will provide several advantages to machine and deep learning systems. When we view the problem in the context of the digital healthcare scenario, physics-informed deep learning is about applying physical principles and rules to convert computer simulations into routine applications for the clinical arena.
This Research Topic will focus on original algorithmic, methodological, and theoretical contributions to AI, scientific machine learning and physics-informed deep learning research and in particular applications in medical data analysis and broadly related to digital healthcare.
This Research Topic features submitted papers from the MIUA 2022 conference, a UK-based international conference for the communication of image processing and analysis research and its application to medical imaging and biomedicine.
AI is currently sweeping the academic and industrial worlds. It mixes classic scientific mechanistic modelling (differential equations) with conventional machine learning and innovative deep learning techniques. However, deep learning, as is widely known, suffers from challenges like poor interpretability and lack of physical constraints; merging such approaches with numerical analysis and differential equations might lead to a new field of research via novel methods, structures, and algorithms.
Physics-informed deep learning strategies seek to address the limitations of data-driven approaches, such as (i) the large quantity of data required from data-driven models to detect and analyze medical data, and (ii) the development and collecting of clinical data that frequently does not meet the physical requirement. It is emerging that adding physical models will provide several advantages to machine and deep learning systems. When we view the problem in the context of the digital healthcare scenario, physics-informed deep learning is about applying physical principles and rules to convert computer simulations into routine applications for the clinical arena.
This Research Topic will focus on original algorithmic, methodological, and theoretical contributions to AI, scientific machine learning and physics-informed deep learning research and in particular applications in medical data analysis and broadly related to digital healthcare.