Consistency could be found in our lives all around the world. Routine allows us to thrive: we drive to the same place almost every weekday, to do the same kind of work and to interact with the same people. If our consistent routine is disrupted, we sometimes find ourselves "going on autopilot", or beginning to follow the routine that we are breaking. So, why wouldn't consistency be as important in areas like radiology?
In radiology, consistency is incredibly important. We want the most consistent magnetic field, reducing inhomogeneity in our MR scanners to produce the best possible images. We want to have consistent imaging procedures so that the staff knows exactly what they are supposed to be doing. Finally, we may want consistency in the images produced of the same areas of the body. This is especially true in longitudinal studies of patients. New softwares from MR scanner manufacturers could further enable consistency in imaging.
How does consistency help radiologists?
In many cases, consistency could help radiologists read their scans faster. If they have to find anatomical landmarks, they should be able to look in the same area across the scans of different patients. This could allow them to find abnormalities in certain patients quicker, enabling faster, more personalized results. However, it is not just in reading radiographs that consistency comes into play.
Consistency is also important in the scanning of images. The radiographer has to be able to scan the patient as consistently as possible, especially for the imaging in longitudinal studies. Longitudinal studies produce images of one area of the body over a period of time. These studies may allow physicians to spot differences between scans, potentially showing disease progress. This is also important in medical research, because it could show the changes caused by a disease and provide insight into the effectiveness of treatments.
How can radiologists improve consistency?
There are a couple of different software programs that may help radiologists and radiographers enhance the consistency in their department, as well as across multiple departments. Automatic slice prescription powered by AI may be one of the most innovative efforts to improve consistency, and other assisting software programs could help as well.
Automatic slice prescription is a newer idea that has been pushed forward through the use of AI.1 The software program is trained using deep learning to easily recognize landmarks in the body. This allows the program to decide which angles and areas need to be imaged. The imaging technique used in MRI is called slicing. It takes an image of an angle of a portion of the body, called a slice. The slices acquired depend heavily on the radiographer and the scanner. Automatic slice prescription software could eliminate some of the inconsistencies due to these factors, because the computer does the selection itself.
Another way that consistency can be increased in MR imaging is through the use of quantification software programs. These programs calculate the levels of different things observed through the scan, much like radiologists traditionally had to do while interpreting the data. In the past, the numbers could be influenced by the radiologists' beliefs about what was normal and what was abnormal. Now, radiologists can take their own measurements and compare them to the computer's. This means that they can double check their measurements to make sure they are consistent. One doctor in particular, Dr. Matthew Bramlet from Children's Hospital of Illinois and the University of Illinois College of Medicine in Peoria, Illinois, has increased confidence in his measurements with the help of one of these programs.2 He also feels that the software program is easier to use and provides more reproducible, faster values.
Both automated slice prescription and measurements produce by the computer allow for more consistency across the MR spectrum. This could enable further studies of disease and consistent longitudinal studies. In turn, researchers studying diseases could use the acquired scans to study the disease itself. Consistent longitudinal studies may allow doctors to more accurately assess disease. In the future, this could provide the same kind of consistency in radiology as we see in everyday life.
For more information on Dr. Bramlet's story, please read SIGNA Pulse "An efficient and reproducible toolset for cardiac MR image analysis."
1. "GE Healthcare's FDA approved MR neuro deep-learning software, AIRx, increases consistency and productivity." dotmed.com. 14 March 2019. Web. 1 May 2019. <https://www.dotmed.com/news/story/46576?s=newsreg>.
2. Matthew T. Bramlet. "An efficient and reproducible toolset for cardiac MR image analysis." SIGNA Pulse. Spring 2018. Web. 1 May 2019. <http://www.gesignapulse.com/signapulse/spring_2018/MobilePagedReplica.action?pm=2&folio=20#pg20>.