Beyond Today's MRI: Machine Learning and AI Pioneering Tomorrow's Imaging Landscape
Yvanne Komenan *
University of Kennesaw State University, Georgia.
*Author to whom correspondence should be addressed.
Abstract
Aims: To survey the application of artificial intelligence and machine learning in magnetic resonance imaging.
Objectives: To discuss the fundamental knowledge behind the concepts of magnetic resonance imaging, artificial intelligence, and machine learning. The interconnectivity between utilizing AI models and different MRI images to achieve perfect evaluation was also examined.
Discussion: Various MRI images were discussed, including magnetic resonance angiography, anatomical MRI, diffusion MRI, and functional MRI. Supervised and unsupervised machine learning are the types of ML that have found wide applications in MRI. For supervised machine learning, the various methods under this are k-space methods, image restoration methods, cross-domain methods, direct mapping, and unrolled optimization. Nonetheless, recurrent neural networks (RNNs) and convolutional neural networks (CNNs) were the two noticeable AI models often employed during medical imaging.
Conclusions: In conclusion, artificial intelligence as a subset of machine learning has found wide medical applications to MRI. The emerging technology of AI in MRI has profound future applications in medical field.
Keywords: Machine learning, magnetic resonance imaging, cancer, artificial intelligence, diagnosis tests