Advanced image analysis techniques, when applied to MRI data, are helping researchers gain a better understanding of neurological disorders such as multiple sclerosis (MS).

Involved in early-stage research in this area is Christopher Holland, a MD/PhD student in the department of anatomy and neurobiology at Boston University School of Medicine, who is conducting doctoral research on MS in the Center for Neurological Imaging (CNI) at Brigham and Women’s Hospital (BWH; also Boston).

Under the direction of Charles Guttmann, MD, CNI director and assistant professor in radiology at Harvard Medical School (Boston), Holland’s research is focused on the spatial relationship of white matter signal abnormalities, or lesions, as detected by MRI to anatomical architectures in the brain. The researchers’ goal is to define metrics that improve current diagnostic and prognostic methods for MS.

The pathologic hallmark of MS is the presence of demyelinating or inflammatory lesions in the brain and spinal cord. MRI, central to the diagnosis of MS, is the best method to track the appearance and progression of such lesions.

“My current research interest is to look at the spatial distribution of lesions in brain, to systematically do that in a number of subjects and then compare those distributions across subjects, or across groups, to find patterns,” Holland told Diagnostics & Imaging Week.

Holland conducts his research at the CNI’s image analysis laboratory and MRI facility, which includes a dedicated 1.5 Tesla MR scanner – a GE Sigma from GE Healthcare (Waukesha, Wisconsin) – as well as access to a large- and short-bore 3 Tesla system at the BWH Radiology Department.

The primary clinical measures of disease in MS patients found on MRI is the number of white matter lesions in the brain, and whole brain bulk, which examines the total volume of brain tissue present and allows the accounting of the loss of tissue over time, known as atrophy. Quantitative image analysis enables reproducible estimates of brain components such as white matter, white matter signal abnormalities, gray matter and cerebrospinal fluid.

“In diseases like MS and also normal aging, the brain shrinks over time, so tissue is lost,” Holland said. “In normal aging this process is slow, and in diseases like Alzheimer’s in particular, or MS, it can be dramatically accelerated.”

Based on statistical and mathematical methods, including segmentation and other image analysis techniques, the goal of Holland’s research is to “understand the pathogenesis of MS, the cause underlying what we’re seeing on the MRI scans – why some regions are more susceptible than others,” he said. “And looking forward to the outcomes, how does what we see on an MRI scan relate to clinical disability measured in the patient, both cognitive and physical ability.”

While examining lesion loads and atrophy gives clinicians “some sense of how severe that patient’s disease is, it is not a complete picture,” Holland said. “When you’re trying to quantify disease, it is going to fluctuate. Not all of the pathology observed on MRI is permanent, so it is very tough to disassociate the transient or temporary changes, such as inflammation or edema, from permanent damage.”

Holland acknowledged the challenges to correlating a lesion’s spatial distribution in the brain with a particular deficit or disability. “Think of all these different overlapping and interwoven compartments of the brain that have different functions. In MS, it’s a very murky picture,” he said.

Another challenge, according to Holland, is that measurement of disease on MRI is limited by what is called a “clinico-radiological paradox,” or difficulty correlating conventional MRI with disease severity and disability in MS patients.

“Some patients have relatively minor atrophy and low lesion volume and have huge disability. And other patients have high lesion loads, lots of atrophy and minimal, if any, recognizable disability,” he explained. “The reason for that, we believe – and the basis for part of the work I do – is that it is not necessarily how much lesion volume you have, but where those lesions are located.”

Holland is working to identify white matter foci, regions that correlate to specific function or cognitive deficits, using statistical analysis to compute probability maps.

“Part of what I’m doing as I’m looking at the distribution of lesions in space,” he said, “is taking sets of patients and creating 3-D frequency maps where we can look at how lesions are distributed in that subject group and compare those distributions to groups with different characteristics.”

This analysis was expanded by examining the relationship of these spatial distributions to the vascular architecture of the brain, and more recently to normal cerebral perfusion patterns. “I was interested in the anatomical context of brain pathology or injury, first where they are in space, and then further, the relationship to other specific anatomical structures or physiological properties,” he said.

Working with Dr. Dorota Kozinska, a professor from Poland conducting research at BWH, they developed a method to examine the relationship of individual lesions, in this case MS white matter lesions, to the vasculature of the brain.

The researchers used information from an angiogram, “which would give us the architecture of the blood vessels throughout the entire brain, and superimposed the lesions detected on dual-echo MRI scans. [Our] program would mathematically calculate for every single lesion the distance to the nearest vessel, the caliber of that vessel, and all of the geometric characteristics of the lesions.”

Their pilot study was published in 2004 in the journal NeuroImage.

“It was basically a proof-of-principle paper, because it was only applied to three patients and therefore our results were not statistically significant. But the results allowed us to speculate from our findings that larger lesions were further from the nearest blood vessel. A potential functional implication of this would be that they don’t have the same amount of blood flow,” Holland said.

“This led to the transition from looking just at structural relationships to using functional measures, in this case perfusion studies,” he added.

The team’s ongoing work seeks to elucidate the role of perfusion as a modulating factor in the pathogenesis or repair of inflammatory brain lesions, such as those associated with MS.

“It is well established that outside of the central nervous system in the rest of the body, well perfused areas heal more efficiently and completely,” Holland said.

But it is unclear if that same relationship exists in brain. “Currently we are integrating SPECT and MRI data to examine how lesions are distributed in the brain with respect to patterns of cerebral perfusion in normal and abnormal cases.

“We believe this work has the potential to yield insights into the etiology of multiple sclerosis, the discrimination of permanent vs. transient damage in brain lesions, and have implications for both patient prognosis and may lead to enhanced or novel pharmacological interventions,” he said.