Engineering in Medicine
Behind every great medical advance, there’s a great engineer.
By Elizabeth Kelsey
Photographs by John Sherman
Anyone who goes to the doctor benefits from the work of engineers. Every medical device represents a collaboration between doctors eager for better ways to treat patients and engineers eager to push technological boundaries. Dartmouth engineers have focused on medical technologies since the 1960s, when Professor John Strohbehn started a biomedical engineering program at Thayer. Collaborating with clinicians at Dartmouth Medical School, Strohbehn directed his inventive skills to a wide range of medical applications, including mathematical models for X-ray tomography, an interactive image processor for clinical use, hyperthermia techniques for destroying cancer cells with heat, and a frameless stereotactic operating microscope for neurosurgeons. His work inspired several graduate students who today are lead researchers at Thayer, including Professor Stuart Trembly Th’82, who developed a microwave thermokeratoplasty technique to correct nearsightedness, and Keith Paulsen Th’86, who heads the engineering side of Dartmouth’s comprehensive medical imaging programs.
Thayer School is home to a wide range of medical projects. Some, such as the joint implant work of professors John Collier ’72 Th’77, Douglas Van Citters ’99 Th’03, ’06, and Dr. Michael Mayor, have led to major improvements in medical devices and techniques. Others have resulted in new companies, such as GlycoFi, the protein engineering start-up that professors Tillman Gerngross and Charles Hutchinson sold to Merck in 2006 for $400 million, or Gerngross’ latest bioengineering venture, Adimab.
In the following, Dartmouth engineers explain some of the many projects they are working on with colleagues at Dartmouth Medical School and Dartmouth-Hitchcock Medical Center to give doctors new tools for better care.
The Researcher: Professor Ursula Gibson ’76
WHAT IT IS
Nanomedicine is the intersection between nanomaterials and medical research. We’re using knowledge of self-assembly of small-scale materials to try to make therapeutic substances that will interact with diseased tissue.
WHY NANOMEDICINE CAN WORK FOR CANCER
Cancer tissues have a vasculature that is not well developed. They essentially have little pores in them that are about 300 nanometers in diameter — a scale at which people have started to gain control over what they can fabricate. The leaky vasculature allows us to preferentially deposit nanoparticles inside a tumor.
If we can structure materials on a very fine scale, we can give them a plethora of functionalities for targeting cancer. We could incorporate a drug into a polymer structure, add some magnetic nanoparticles that can absorb energy from an externally applied field, and attach an antibody to the outside of the polymer to help it find the cancer or bind preferentially to the cancer for a multiplicative effect.
There’s a lot of work now to figure out how to get therapeutic drugs and materials into tumors without having them intercepted by the body’s natural defenses, how to optimize entrance into the tumor, and how to activate the drugs and materials once they get there. There are certain conditions that are present in a tumor — the pH is different, the oxygen concentration is different — and people would like to leverage all of those into a system that gives a higher release factor in cancerous regions and a lower release anywhere else.
The Researcher: Professor Ian Baker
WHAT IT IS
You can heat up nanoparticles by applying an alternating magnetic field. If you introduce the particles into a tumor — either by injecting them or by antibody-tagging them — you can use the heat to kill cancer cells.
Many groups have been working on using iron-oxide particles for magnetic hyperthermia. We’re making iron nanoparticles coated with iron oxide. We wanted to use iron because it is more magnetic — and therefore has greater heating abilities — than iron-oxide. But nano-sized iron particles exposed to air quickly form an oxide and produce so much heat that they catch fire. I had the idea of using iron particles for their heating effect and coating them with iron oxide to make them safe and make them visible in magnetic resonance imaging. But I really hadn’t much clue how to do it. Fortunately, post-doc Qi Zeng figured it out. More recently another post-doc, Guandong Zhang, managed to tweak the processing so that now our particles range from 8 to 20 nanometers, heat well, have a biocompatible coating, and have good MRI contrast.
The ultimate goal is to tag these nanoparticles with tumor-specific antibodies, and put them into the bloodstream to find their way to a tumor and attach to it. Then, using a technique like MRI to ensure the particles are at the tumor, we’ll magnetically heat the tumor to destroy it without damaging surrounding tissue.
LOW-POWERED ELECTRONICS FOR MEDICAL DEVICES
The Researcher: Professor Kofi Odame
WHY THEY’RE NEEDED
Low-power electronics are crucial for medical devices that need to be portable or implantable.
A cochlear implant works by capturing sound, processing it, and — bypassing the inner ear — stimulating the patient’s auditory nerve. Currently only a small part of the device is actually implanted. An external part contains a microphone, microprocessor, and batteries, which need to be replaced or recharged every 24 hours or so.
I want to make an implant that will allow users to discriminate between sounds they’re interested in hearing — say a conversation — and background noise. I’m currently working on an algorithm and designing an electronic circuit that will imitate a healthy cochlea.
But a fully internal cochlear implant, with all of its parts surgically placed inside the patient’s head, will never become a reality unless the implant is made to consume so little power that batteries would only need to be changed once every several years, if at all.
EPILEPSY MONITORING DEVICE
Many people with epilepsy need to be monitored long-term. Monitoring in hospitals is expensive and disruptive. Portable EEG monitors exist, but they’re bulky and awkward. My idea is to create a device that is small enough to fit in a hat. The device would be doing wireless transmission of electroencephalogram (EEG) data, which consumes a lot of power. I’m designing circuits that selectively transmit only the important portions of EEG data in a low-power fashion.
The Researcher: Professor John Collier ’72 Th’77
MAKING BETTER IMPLANTS
We look at the problems orthopedists and patients have with hip and knee implants and figure out how to solve them.
It used to be that loosening of implants was a big problem. Loosening is no longer a problem [thanks largely to a porous coating Collier developed in the 1980s]. It used to be that a success rate of 70–80 percent at five years was good. Now it’s 97–99 percent at ten years. We’re looking to push that to 97–99 percent at 20 years or more.
But making better materials for implants is tough. People use implants in a very active environment over many years. We do in-vitro simulations, but when a new material is put in the body, we commonly get surprised. The body environment is complicated — with proteins, enzymes, free radicals, cells, low pH, and a whole lot of other things going on — and implants are submersed in it for a long time. We don’t know how to model all those conditions.
IMPLANTS AND OUTCOMES
Orthopedic surgeon Michael Mayor thought that if we’re putting implants into patients, we ought to examine the devices when they come out. He started our implant retrieval lab. Since the 1970s he has evaluated nearly 10,000 retrieved implants for corrosion and other damage. We compare components that have been in patients to predictions from our tests. There’s a constant evolution of tests and evaluation of components.
Cross-linked polyethylene components show promise. Polyethylene is a long-chain monomer; you can improve its wear resistance by connecting the chains to one another.
GET AN EXPERIENCED SURGEON
Even with a sophisticated prosthetic device, surgeons can get into difficulty if they haven’t done many implant procedures. The device might not be exactly the right size or it might not be tensioned or positioned exactly. Any one of those things is sufficient to cause the device to fail. It’s very difficult to make devices robust enough to survive if they’re put in the wrong position.
MULTI-MODAL BRAIN IMAGING
The Researcher: Professor Solomon Diamond ’97 Th’98
IMAGING ALZHEIMER’S DISEASE
My graduate students and I want to understand how neurodegenerative diseases like Alzheimer’s work, so we’re developing brain imaging tools and technologies to dynamically measure neurovascular coupling — the relationship between evoked neural activity and subsequent changes in cerebral blood oxygen, volume, and flow.
We’re building a special brain imaging lab at Thayer that will allow us to run three different noninvasive neuroimaging techniques simultaneously. One technique is electroencephelography (EEG), which measures electrical potentials on the scalp that arise from ionic currents in the brain’s neurons; the second is magnetoencephelography (MEG), which measures the magnetic field around the head that also comes from neural currents; the third is near-infrared spectroscopy (NIRS), which uses near-infrared light to rapidly measure brain blood volume and oxygen dynamics.
FASTER THAN FMRI
Our instruments will allow us to explore neurovascular relationships with more precise timing and better specificity to neural currents than can be done with an MRI. With functional MRI, you get an image stack of the whole brain about every two seconds, and you can see how the stack evolves over time due to the cerebral blood dynamics. You can tell what part of the brain was active when the subject saw a certain stimulus or performed a particular task. But you wouldn’t know the moment-to-moment waveform of blood dynamics, and you wouldn’t have any concurrent measure of the neural currents. We will be able to measure cerebral blood and neural dynamics in our lab.
Once all the instruments are functioning, we’ll start studying the neurovascular coupling relationships in healthy normal controls. With that data as a foundation, we can then begin the pilot clinical work, measuring people who are at risk for developing Alzheimer’s or who are in the early stages of it. The background work could take five years, but I envision early clinical work at the pilot level starting in two to four years.
The Researcher: Professor Karl Griswold
I think of proteins as the nanoscale machines that enable life. Knowing how well proteins work in their natural context, engineers and scientists have envisioned ways in which we can use these molecules to enable practical applications. However, when you yank proteins out of their natural context, they frequently lose some or all of their desirable qualities. This is where protein engineering comes into play. We take natural protein sequences as a starting point and redesign them to meet our own performance criteria.
CYSTIC FIBROSIS ANTIBIOTICS
Seeing proteins as next-generation antibiotics, I wanted to develop enzymes to treat bacterial infections. That led me to work on cystic fibrosis.
Most CF patients die from complications associated with Pseudomonas aeruginosa bacterial infection of the lungs. In many infections, bacteria adhere to a surface — such as the lining of the lung, a catheter, or an orthopedic implant — and cover themselves with biofilm, a blanket of proteins, nucleic acids, carbohydrates, and other components. By growing inside this complex matrix, bacteria gain protection from the human immune system and antibacterial drugs. This is what makes Pseudomonas in the lungs of CF patients so insidious. Once contracted, the bacteria generally cannot be eradicated.
We’re looking for enzymes that will break through Pseudomonas’ biofilm armor. We’re identifying natural enzymes that have infection-fighting potential, outlining their limitations, and then developing improved versions in the lab.
BREAST CANCER IMAGING
ALTERNATIVES TO MAMMOGRAPHY
Mammography isn’t perfect. That’s why Thayer School engineers and Dartmouth Medical School (DMS) researchers have been working together for a decade on four alternative imaging technologies to detect and help treat breast cancer.
Keith Paulsen Th’84, Thayer School’s Robert A. Pritzker Professor of Biomedical Engineering, and Dartmouth Medical School radiology professor Dr. Steven Poplak are the principal investigators. Together they lead a team of some 40 engineers, radiologists, pathologists, computer programmers, data analysts, and other collaborators who are refining the technologies and conducting clinical trials at Dartmouth-Hitchcock Medical Center.
The four technologies — microwave imaging spectroscopy (MIS), electrical impedance spectroscopy (EIS), near-infrared imaging (NIR), and magnetic resonance elasticity (MRE) — differ from mammography in a key way. Mammography detects tissue that looks like a tumor. The other technologies detect tissue that acts like a tumor.
MIS and EIS measure the ability of different regions of the breast to hold or conduct electricity. Part of what defines a tissue as cancerous is how its cells and blood vessels are organized. Normal tissue is quite orderly. Cancer, however, is “just a jumble,” says Thayer professor Alex Hartov Th’88, EIS project leader. “A lot of membranes, a lot of vascularity — all these things are associated with different electrical properties.”
NIR uses a “unique spectral window,” says Thayer professor Brian Pogue, who leads the NIR project. Penetrating deep into tissue, NIR reveals information about hemoglobin and oxygen saturation levels, which can indicate a tumor.
MRE uses an MRI scanner and specialized coils to vibrate breast tissue and measure whether it is elastic or stiff. “Almost all cancer is stiff,” says MRE project leader John Weaver, a professor of radiology at DMS and adjunct professor at Thayer.
All four technologies differ from mammography in another way as well: computational complexity. Since mammography utilizes X-rays, which penetrate the body in more or less a straight path, constructing an image from X-ray data is a relatively easy linear problem. But the electromagnetic waves used in MIS, EIS, and NIR and the mechanical waves generated during MRE travel through the breast in complex patterns. Generating images from this data requires complex differential equations.
The mathematical and engineering problems involved in getting these modalities to work are “huge,” says Thayer professor Paul Meaney Th’95, head of the MIS project.
Even so, the Thayer-DMS group has developed both the software and the hardware — free-standing machines for MIS, EIS, and NIR, plus specialized equipment that’s used inside an MRI scanner for MRE and NIR. And while researchers elsewhere are working on their own versions of the technologies, no program is as comprehensive and well developed as the Dartmouth collaboration.
“We’re looking at all four technologies in a common setting,” explains Paulsen. “We can look at them together and comparatively and synergistically.”
Size is part of the success, says Paulsen. “It’s sort of the big company/small company paradigm,” he explains. A big company can’t adapt itself quickly to something new, he says, “because there’s huge infrastructure and huge investment in teams. We’re a much more nimble, small enterprise.”
BEYOND BREAST CANCER
Thayer researchers explain new applications for their imaging technologies.
Professor Keith Paulsen Th’84
We’re developing a fluorescence signature to guide surgeons in removing brain tumors. We’ve found that high-grade tumors fluoresce, so the fluorescence can be used as a kind of surgical road map. We hope to be able to make low-grade tumors fluoresce, too, so they can be treated before they get worse.
MOLECULAR IMAGING OF GLIOMA BRAIN TUMORS
Professor Brian Pogue
Glioma brain tumors don’t always show up in an MRI scan. The most problematic parts of a tumor cannot be seen structurally because they’re a microscopic invasion. With better molecular tracers and optical imaging, though, we’ve been able to detect them. This is very exciting, because we can now demonstrate that we can see something that standard clinical imaging doesn’t capture.
Our glioma research has been done on mouse models of human cancer. We use an injectable agent that isn’t FDA-approved yet, but would be a good candidate for future clinical use. We need to do phase-one trials with new drugs and imaging systems to make sure they don’t harm the patient. The process takes years and depends on funding, competing technologies, and other factors. If all the right factors come together, we can consider doing imaging studies on humans.
IMAGING FOR METASTATIC CANCER
Professor John Weaver
We’re using imaging to monitor neoadjuvant therapies to treat metastatic cancers. Rather than immediately removing the primary cancer, the oncologist leaves it in place and attacks it with a mix of chemotherapies. If the chemotherapies shrink the primary tumor, you can assume that they also attack unseen metastases. Obviously, you want to find out really quickly if the chemo cocktail is working. Instead of waiting a month, we want to use imaging to find out in a week.
MICROWAVE IMAGING OF ULTRASOUND HEATING
Professor Paul Meaney Th’95
If you’re using heat to destroy a tumor, you want to monitor the temperature during the treatment. It has been challenging to find a noninvasive method. We use a focused utrasound system to heat the tumors. We’ve integrated a microwave imaging system into it, allowing us to continuously image the patient while the heating occurs.
EIS FOR PROSTATE CANCER
Professor Ryan Halter Th’06
A big problem with screening for prostate cancer is that the prostate-specific antigen (PSA) test is not specific to just cancer. Benign conditions can also cause PSA levels to rise. Ultrasound-guided biopsy is used to definitively diagnose prostate cancer in men with elevated PSA, but it misses some malignant lesions and doesn’t accurately characterize the extent of the disease. We’re incorporating electrical impedance spectroscopy (EIS) sensors into standard biopsy needles and ultrasound probes to improve this detection and disease characterization. We’re also developing EIS-enhanced probes that surgeons can use during surgery to ensure clear margins.
— Elizabeth Kelsey is a contributing editor at Dartmouth Engineer
For more photos, visit our Engineering in Medicine set on Flickr.