Expanding the Boundaries
By Dan Harvey
Radiology Today
Vol. 18 No. 6 P. 18

Volumetric imaging software offers new options for measuring disease-related brain changes.

Volumetric MRI, which collects signals from an entire tissue sample to form a single image, represents an advance in technique and technology. Until recent years, volumetric software served as an invaluable tool for researchers seeking to expand the borders of imaging territory; now, due to the all-important FDA clearance, the software has moved into the clinical setting. The software's unique capabilities, advantages, and value in improving diagnosis of degenerative brain disease, especially Alzheimer's, have fostered this shift.

Applied to neuroradiology, volumetric imaging measures the size of the different structures within the brain. From there, it quickly detects the subtlest changes in brain volume. Reduced volume of certain brain structures is a significant indicator of a problem. For instance, atrophy of certain anterior brain structures is found in patients with frontotemporal dementia. Reduced hippocampus and amygdala volumes are identified in patients with schizophrenia.

Furthermore, the software indicates when a patient is not suffering early Alzheimer's disease; cognitive problems can result from other disorders, and the software can override an inappropriate diagnosis. Steve Fung, MD, a neuroradiologist at the Houston Methodist Hospital in Texas, makes an additional and important point: Early Alzheimer's disease can present with age-appropriate brain volumes. In such cases, FDG PET and beta-amyloid PET scans may be more sensitive.

Automatic Improvement
Previously, using volumetric data for degenerative brain disease detection was time-consuming. New software has changed that.

"Volumetric imaging dates back to the 1990s, but it was manually oriented," says Barton Branstetter, MD, a neuroradiologist and professor of radiology, otolaryngology, and biomedical informatics in the neuroradiology division within the department of radiology at the University of Pittsburgh.

"Manual segmentation required someone sitting down with each image and measure," adds Joseph Mettenburg, MD, PhD, also a neuroradiologist and an assistant professor in the department of radiology at the University of Pittsburgh.

"So it was labor intensive," Branstetter says. "Then we went through a period where it was semiautomatic." That is, the radiologist provided some input and the computer finished the job.

"Now, a radiologist doesn't need to be involved," Branstetter continues. "The acquired data can be sent directly to the computer software. Values are sent without further intervention."

Thus, both the efficiency and practicality of the process have been enhanced. Another benefit is better precision with better reproducibility and reliability.
"It's far better now, with the improvements in the computer's ability to automate the process as well as improvements in MR capability and the elimination of reader subjectivity," Branstetter says.

Software Providers
Currently, there are two companies offering FDA-cleared volumetric imaging software. One is San Diego-based CorTechs Labs. Its flagship product, NeuroQuant, enables the acquisition of volumetric brain image data and automatically conducts accurate and consistent measurements of cortical and subcortical volumes to help identify evidence of neurodegeneration.

Within minutes, the solution's fully automated MRI postprocessing capabilities enable physicians to accurately measure brain structure volumes and assess the presence of neurodegeneration with greater confidence. Also, NeuroQuant automatically segments and measures volumes of brain structures and compares them with norms, which reduces subjectivity in the diagnostic process. Collaborating with CorTechs Labs, Philips is providing NeuroQuant in its IntelliSpace Portal 9.0.

The other FDA-approved provider of volumetric imaging software is Brainreader, a medical device company located in Horsens, Denmark, which markets Neuroreader. As with NeuroQuant, Neuroreader quickly detects changes in a patient's brain volume resulting from neurodegenerative diseases. Users upload MR images to Neuroreader, which compares the images against an FDA-approved database of healthy brain scans. In a matter of minutes, the software can measure 45 different brain structures and report back on changes as small as hundredths of a millimeter, according to Brainreader.

Results of a study published in a 2016 issue of the Journal of Alzheimer's Disease showed that Neuroreader was as accurate as traditional methods for detecting the subtlest changes in brain volume, in a fraction of the time.

Current and Potential Applications
Fung has found volumetric imaging software to be a boon to diagnosis: "The software provides z-scores of segmented brain volumes relative to normal age and sex-matched control population, and [it allows me to] rank individual right and left hippocampal, frontal, parietal, occipital, and temporal lobar z-scores to determine if disproportionate volume loss is present. Disproportionate hippocampal and temporoparietal volume loss would suggest Alzheimer's disease."

Conversely, disproportionate frontotemporal volume loss would suggest frontotemporal dementia. "I would question primary progressive aphasia, if there is asymmetric left hemispheric volume loss," Fung says. "Usually, the volumetric findings support and make me more confident with my visual interpretation. If I didn't 'see it' the first time, it does force me to look again."

While volumetric imaging is proving to be a significant advance in the diagnosis of Alzheimer's disease, industry professionals are already considering potential applications. Some of these include improved diagnosis of other neurological disorders such as depression, fragile X syndrome, Rett syndrome, and Tourette syndrome. The new software may eventually improve the assessment of brain trauma following serious injury, as well, but Branstetter says that's farther down the road.

"There is not a well-defined present clinical application," Branstetter says. "Right now, that's an area for more research. There may be some future applications, as brain trauma—from concussion, for instance—lends itself to this type of automated analysis."

Meanwhile, its impact on Alzheimer's diagnosis has been helpful. "It's a reasonable biomarker to test potential treatments," says Cyrus A. Raji, MD, PhD, a neuroradiology clinical fellow at the University of California, San Francisco department of radiology and biomedical imaging. "We can test a new drug using brain volume and other types of brain imaging, such as PET amyloid imaging for one example. There's a lot of interest in using biomarkers to determine if a patient is being helped. So, one day, volumetric imaging may help speed up development of a cure. The potential is there."

Fostering Collaboration
For something as astonishing and world changing as an Alzheimer's cure, collaboration across several specialties would certainly be necessary. New volumetric imaging software is already fostering collaboration in other areas of research, diagnosis, and patient treatment.

"Look beyond the software to the entire process," Fung says. "Many times, findings are subtle, a slight asymmetry, for example. Quantitative information backs up data. The asymmetry would become evident when everything is put together. Collaboration means that you're not reading blindly. When you combine the radiologist's report with the neurosurgeon's clinical information, subjective information now becomes something more reliable: a numerical value. But what happens when a patient has a lesion? A more focused analysis is required, to better find out what is going on."

The FDA first cleared NeuroQuant in 2007 and Neuroreader in 2015. Since those clearances, Raji has witnessed a convergence of cognitive neurology and geriatric psychology. He even has contributed to that convergence with a new concept that he published a report about: cognitive health.

"It refers to the reality that many types of cognitive problems exist, such as trouble concentrating, focusing, or performing mental tasks appropriately. Alzheimer's could be the cause, but the problem may be due to vascular disease or traumatic brain injury or depression, among other causes." The advantage of identifying a cause, he says, is that it enhances patients' care and management.

Raji provides a specific example of collaboration: At UCLA, he collaborated with geriatric psychiatrist David Merrill, MD, PhD, who operates his own cognitive health clinic. "He wants to show that when people suffer certain mental problems, it does not always mean Alzheimer's," Raji explains. "We look at his patients, analyzing brain volumes using Neuroreader. When we started, we thought we'd be detecting people with Alzheimer's, but we have found also that the technology is good at detecting potential cognitive declines that aren't Alzheimer's."

The collaborators had a patient present to the clinic with multiple memory problems who was absolutely convinced that he was suffering from early Alzheimer's. With the Neuroreader, they discovered that all of the patient's brain volumes were normal, except for one area—the cerebellum, which is not affected by Alzheimer's. After an in-depth analysis, they were able to determine that the causal factor was alcohol abuse.

"When we told him the news, he really felt empowered," Raji recalls. "He went from thinking he had an incurable disease to understanding that the cause was the alcohol. He subsequently quit drinking. So, having that quantitative data in that case proved extremely helpful for the patient's care and subsequent management."

Mettenburg has also witnessed increased collaboration in the new software solution environment; he now works directly with many colleagues in different departments. "We've received a lot of feedback about volumetric analysis outside of the neurology department. Even within neuroradiology, several subspecialists are particularly interested."

Interest has also grown at memory impairment clinics that look at all neurogenerative diseases—not just Alzheimer's—at movement disorder and multiple sclerosis specialists, and among epileptic neurologists and surgeons. "They're all very interested in volumetric analysis of certain brain regions," Mettenburg says.

Looking Ahead
Fung says users are currently pushing the boundaries of what the software can or should be doing, but whatever they come up with will have to undergo extreme vetting (validation process, peer review, etc). "I consider the software imaging to be just like medicine, in that we are pushing the boundaries of medicine," he says. "But you have to be careful, based on the software and how its algorithm works. Research continues but requires validation. Also, we need to best determine both the potential benefits and the potential limitations."

For Branstetter, it is hard to forecast exactly where it is all headed. "We've really just opened the door on using this in a routine clinical manner," he says.

"We will have a better grasp on how to use it with the individual patient, with more and more clinical context," Mettenburg adds. "We may see new growth areas such as head injuries. I see it moving into many more discipline areas for many different disease states, besides the ones already mentioned."

Goals are multifold: to establish a diagnosis and prognostic value, Branstetter says. "Prognostic value may be of particular importance—to be able to tell someone that the memory deficits experienced are definitely the early stages of something more prolonged. Also, a longitudinal assessment of how rapidly a patient is progressing will help us establish clinical thresholds such as with intervention, medication, or therapies as appropriate. So that is the future of how this will be applied."

Mettenburg adds a final, important point: "These evaluations will need to incorporate other modalities for evaluation and diagnosis. So there could be components integrated with the volumetric analysis."

— Dan Harvey is a freelance writer based in Wilmington, Delaware.