Perhaps no technology has generated more buzz in radiology recently than Artificial Intelligence. The technology is a potential game-changer in the space, essentially giving radiologists a second set of eyes for diagnosis and image analysis.
The technology behind the MRI is still based on generating a powerful magnetic field. Although the basic technology and hardware designs have dramatically improved, most of the recent advances have been on the software and computing side, allowing faster and more accurate scans with greater detail. Overall improvements in imaging quality mean that physicians can pinpoint even tiny structures and processes in a living, breathing human being. Interestingly enough, breathing and lung tissues were sometimes difficult to scan because they are filled with air. Air has a relatively low dentistry and therefore was not always easy to image during the short between breath stages of breathing. Newer technology has worked around this to increase the accuracy and detail of scans of all forms of tissue, including the lungs.
Besides being faster, scans have also become much safer. Because an MRI does not rely on ionizing radiation in the same way that X-rays do, they are considered minimally invasive and safe overall. Even so, newer MRI technology has increased patient comfort, produced breathtakingly clear images and dramatically cut down the time required to perform an MRI. Additionally, newer MRIs are less noisy, reducing the sounds to ambient background noise rather than loud clicks and whirring noises that some patients find uncomfortable. Open MRIs also offer a more open field to reduce any anxiety caused by feeling confined. Additionally, newer MRI’s are less noisy, reducing the sounds to ambient background noise rather than loud clicks and whirring noises that some patients find uncomfortable. Open MRI’s also offer a more open field to reduce any anxiety caused by feeling confined.
There is no doubt that the market is witnessing a huge paradigm shift in the focus and design requirements for new MRI acquisitions, leading to the fast-paced development of new, fast MRI methods.
The field is still wide open. There is still a lot of potential for new algorithms and exploiting other characteristics of MR. There is a fundamental limit of signal-to-noise ratio. At some point, one has to sample long enough to get it. But every time researchers think they have had a lot of innovation in MRI and it is maturing, something new comes along. So, it is not feasible to say we are there. What’s even more exciting, is, if MR scan times can be reduced to no more than a few minutes, MR has the potential to play an even bigger role in diagnosing and treating more conditions and diseases. MR, which does not require ionizing radiation, could be a very interesting alternative for CT, especially when one combines it with artificial intelligence (AI) and adaptive algorithms.
MRI and CT scans are chock full of information that might be used to personalize treatment. Unfortunately, this information is not readily available. The latest type of AI algorithms, however, might change that, according to Albert Hsiao, assistant professor of radiology at the University of California San Diego. Hsiao observes that the explosive growth of machine learning has raised opportunities for tying (this information) directly to patient management. This is in large part because of the amazing democratization of machine learning that this new area of deep learning has enabled. Information like how much body fat one has, how fat is their liver, how much is muscle mass. those are rich pieces of information that if quantified could really help doctors guide and make more specific patient recommendations.
Physicians might use this information to create imaging phenotypes of patients. These phenotypes, in turn, might be used to personalize the treatment of patients.
Imaging phenotypes would characterize the physical aspects of a patient, seen in medical images. For the time being the information remains embedded in these images, he said. But smart algorithms might extract and quantify it. Doing so could give clinicians insights into exactly what their patients need, allowing an unprecedented degree of personalization.
Deep learning (DL) or machine learning or unsupervised learning, allows algorithms to discover rules for distinguishing certain features of images. In the early days of AI, DL allowed some algorithms to distinguish pictures of cats from those of dogs. Rather than coding the rules for distinguishing these pictures, the algorithms learned them. Today’s DL algorithms might similarly learn how to pick out and extract phenotypic information from CT and MR images. They would do so by leveraging convolutional neural networks (CNNS), which are typically used to analyze images.
Deep learning and CNNs might be applied in any medical imaging modalities, including echocardiography and X-ray angiography. But the three-dimensional nature of CT and MRI make these two modalities attractive, especially when it comes to AI efforts aimed at quantitation, for example, determining the fat fraction of a liver.
The challenge facing AI developers is to validate these algorithms across multiple patient populations. Algorithms typically are developed and tested in a single population type. A cohort of 200 to 300 patients may exemplify this. Achieving generalizability may be especially important when trying to dig information out of CT and MR images, information needed to classify patients according to imaging phenotypes, because different types of patients may express different types of imaging phenotypes.
The challenge is not, however, as difficult as might be imagined. One just basically need the data from these other demographics.
The data are required not only to refine the algorithms but to prove that they work on different types of patients. This proof is needed if the algorithms are to have wide clinical utility, and, consequently, to establish the value of unused data now buried in CT and MR images.
Advancements in AI have led to the development of image-guided, semi-autonomous robots that can guide a procedure using real-time MRI scanning. Although scientists still have to discover how to build robots that can fit within an MRI system to perform a procedure, the potential is there. The creation of MRI systems has led to significant medical breakthroughs and the continued advancement of MRI technology can be expected to continue impacting healthcare significantly in the future.