NeuroQuant: Reducing MRI Analysis Time from 100 Hours to Less than 10 Minutes

Breakthrough: Unmasking Early-Stage Alzheimer's Disease
Reducing MRI Analysis Time from 100 Hours to Less than 10 Minutes

--------------------------------------------------------------------------------

“Debbie, have you seen my keys?”

We all ask questions like that. In an active life, it’s easy to lose track of things when we’re distracted, in a hurry, or in multi-tasking overload.

But mild forgetfulness, when it appears to be progressive, can have a different meaning for people in their 60s, 70s, or 80s. It may stem from a variety of causes, including stress; but it can also be an early indicator of Alzheimer's disease (AD) or less common neurological problems. This sums up the challenge facing neurologists when they see a patient with what is known as mild cognitive impairment (MCI). How can they diagnose AD with a high degree of certainty in the early stages – when it may still be possible to arrest it?

Professor James Brewer of the University of California at San Diego (UCSD) is doing groundbreaking work with a Mac-based diagnostic system known as NeuroQuant that automates analysis of magnetic resonance (MRI) images of the brain. He is the first physician to use this technology on a clinical basis to examine patients’ brains for the tell-tale effects of Alzheimer's disease.

“Diagnosing Alzheimer’s may be helped by taking volume measurements of regions of the brain,” says Brewer. “The old way of doing this is unbelievably painstaking. The MRI is 256 slices with 256x256 resolution, so every voxel is about a millimeter in each direction. One used to have to flip through the MRI slice by slice and trace the hippocampi and other brain regions on each slice using a program called Amira. That could take even an expert neuroanatomist 100 hours.

”Now I have a beautiful setup. The iMac sits on my desk and runs NeuroQuant, which processes an MRI and delivers a quantitative analysis of the entire brain structure in less than 10 minutes. This is a simple, powerful arrangement for doing both research and clinical practice.”

Brewer’s work will be of intense interest to clinical neurologists – and to anyone whose loved one is facing memory loss.

The Tragedy of Patient HM: Life without Memory
The mystery behind memory loss in AD began to unravel in Boston’s Hartford Hospital in 1957. A man known today as Patient HM came to the hospital suffering from severe epilepsy. He was 27 and unemployable, incapacitated by up to 11 seizures a day. Electroencephalography revealed that the source of the seizures was Patient HM’s hippocampi – a pair of seahorse-shaped organs not far from the inner ear. (“Put a fingertip in your ear,” says Brewer, “and it will be about an inch and a half from a hippocampus.”)

The hospital offered Patient HM an experimental procedure with no guarantee of success: removal of part of his medial temporal lobes, including his hippocampi. Patient HM agreed, and the procedure worked. From that point on his seizures were reduced to about two a year. But there was an extraordinary side effect: from that day forward he formed no new memories of any event. He could acquire knowledge and skills, and his pre-operation memories were intact, but nothing that happened afterwards remained in his memory, including the death of his mother and the first moon landing. He described his condition as constantly waking up from a dream and wondering where he was. Word of his plight inspired production of the acclaimed 2000 movie Memento, which Brewer says portrayed the symptoms accurately. The hero was forced to tattoo himself to retain vital information.

Once it became clear that the hippocampi are the gateways of episodic memory -- the memory of events – scientists were able to focus on them as the probable site of AD. Studies revealed that while the brain of a healthy person shrinks by nearly one percent per year, the hippocampi of an Alzheimer's patient shrinks at five times that rate.

“We’ve known since the early 90s that the hippocampus starts shrinking very early in the development of Alzheimer's disease,” says Brewer. “There are variants of the disease, but they’re very atypical.”

Today, with an aging US population and an increasing number of dementia patients, there’s a pressing need to apply this knowledge to clinical diagnosis. But neurologists have been severely technology-limited. What they need is a practical, quantified way to compare their patients’ hippocampi with the hippocampi of healthy people of the same age. MRI scans can deliver high-resolution DICOM-format image slices of a patient’s brain. But to manually quantify 256 hippocampi MRI image slices is, as Brewer points out, a mind-numbing, weeks-long job – an impossibility for routine diagnosis.

NeuroQuant: Automated, Mac-Based MRI Analysis
When NeuroQuant receives a set of DICOM images, it checks image quality and automatically segments (assembles) the image slices into a single 3D volume image. It then electronically isolates the brain from the skull and its membranes, segregates grey matter from white (connective nerve tissue), identifies and quantifies all regions of the brain with great accuracy, and delivers a statistical report including colorized images – all in about 8.5 minutes of processing on Brewer’s iMac.

NeuroQuant, which is FDA-approved, was launched by CorTechs Labs, Inc. in September 2007 at the Society for Neuroscience meeting.

“We primarily used Macs for development,” says CorTechs CEO Michael Smith. “We used OsiriX, an open-source DICOM viewer, to review images. When it came time to decide what platform to release it on, we had been demoing it on Macs, and running it internally on Macs, and we were just getting rave reviews, so it seemed like a no-brainer to roll it out on the Mac. It runs on any Mac, and will perform an analysis in well under five minutes on an 8-core Mac Pro. By marrying it to OsiriX, we give our users a really rich visual experience.”

The most high-profile use of NeuroQuant is for Alzheimer's disease, but it has great potential for the diagnosis of other diseases such as epilepsy, where it may replace the highly invasive examination now in wide use. It is already at work in clinical trials research in the pharma industry.

“We’re excited about NeuroQuant,” says Smith, “because we think it will not only make a big difference in the clinical assessment of patients with neurodegenerative disorders, but it will also end up being a very important product for our development as a company. Much of that we have to credit to the form factor, and the ease of use and nice experience, that we get by rolling this out on the Apple platform. And the leverage we get by having a tool like OsiriX on that platform has strengths that clearly complement this tool. We’re delighted to be able to offer our customers the ability to use NeuroQuant and OsiriX in a tightly integrated fashion.”

A Powerful Tool for Clinical Assessment of PatientsBrewer, who is both professor and clinician as well as a Mac enthusiast, divides his time between teaching residents at the Neurosciences and Radiology Department in the UCSD Medical Center and heading up the Human Memory Laboratory, a research facility in a nearby professional center. The lab, according to Brewer, is very Mac-ish. “We try to do everything we can on Macs.”

Patients referred to Brewer with mild cognitive impairment receive a seven-minute, full-volume MRI brain scan at the UCSD Radiology Department. DICOM image files from the exam move to the hospital’s PACS system. Brewer pulls them onto his iMac using OsiriX. NeuroQuant also resides on Brewer’s iMac. He sends the DICOM files directly to NeuroQuant for processing.

The results – colorized images and statistical reports – are now available on the NeuroQuant web-based browser. They tell Brewer the volume of the patient’s hippocampi, and how they compare to those of healthy patients of the same age. After he adds his diagnostic comments, NeuroQuant outputs the report as a PDF for billing and for informing the referring doctor. It also outputs a DICOM-compliant image series that he sends back to the hospital’s PACS server.

The Bottom Line: Better Care for AD Patients“What we’re trying to do here is bring a brand new concept to practice using NeuroQuant and the Mac platform,” says Brewer. “When doctors attribute memory impairment to AD, they’re usually right – but many times they’re not, and they’re prescribing drugs inappropriately. That’s not a major concern right now, because the drugs we have for AD don’t cause a lot of side effects, other than, say, some mild stomach upset. But the stronger drugs now being developed will have much more serious side effects. You're not going to want to throw those willy-nilly at a patient. We’re saying, let’s get some more objective evidence.

“That’s what we’re trying to provide with NeuroQuant. Radiologists can look at an MRI and see atrophy that looks like AD. But by the time they can see that, the disease is pretty severe, and we can already tell clinically that the person has AD. We’re trying to push the diagnosis back in time to where we can make an intervention such as enrolling them in a clinical medication trial and sparing as many nerve cells as possible.

“You're not usually going to bring those cells back from death. If a person’s already demented when you intervene, they're likely to stay demented. But if we can use NeuroQuant to catch AD at the mild cognitive impairment stage, where the worst problem they have is losing keys or forgetting names, we can keep it from getting worse. That’s the stage where we’d like to diagnose this stuff.”

Although Brewer has not actively sought outside referrals for his clinical work, he’s getting AD patient referrals every week for the exam. And he is also breaking ground in research: he recently completed a study of 269 patients with mild cognitive impairment with the help of just two post-doctoral fellows – an impossibility without NeuroQuant and his iMac.


At A Glance Professor Jim Brewer of UCSD operates a neurology practice at the Human Memory Laboratory in La Jolla, California. Like other neurologists, he sought a practical way to measure shrinkage of the hippocampus, the memory gateway to the brain, to gain objective evidence of early Alzheimer's disease as the cause of mild memory loss. Existing methods required manually tracing the organ on 256 image slices from an MRI brain scan – a process that can take 100 hours.

Brewer’s practice became a beta site for NeuroQuant, a Mac-based image analysis system that processes MRI DICOM brain-scan images, quantifies the volumes of all regions of the brain, and provides statistical data showing hippocampus size for healthy patients of the same age. Brewer runs NeuroQuant on an iMac, along with OsiriX, an open-source DICOM viewer and PACS workstation. Pulling a patient’s DICOM files from the hospital PACS, Brewer can quantify hippocampus size in about 8.5 minutes.