Medical imaging continues to permeate the practice of medicine, and SFU is playing an increasing role in its development and use. Here below are three of examples of SFU researchers who are making a difference in the field.

Stories by Diane Luckow

Medical-imaging research targets robotic surgery

Computing scientist Ghassan Hamarneh is leading a US$1-million medical-imaging research project to help surgeons performing image-guided robotic surgery.

The SFU associate professor’s research team is working to overlay richly detailed 3D pre-operative medical images with lesser-quality images taken quickly during surgery.

Melding the two would give surgeons a more dynamic and accurate understanding of what’s happening in the patient’s body during an operation.

The Qatar National Research Foundation is funding the research, which Hamarneh hopes will lead to a strategic alliance in medical-image computing with Qatar’s emerging research and technology sector.

His team is working closely with the Qatar Robotic Surgery Centre and other researchers to better acquire, analyze and visualize 3-D anatomical images in real-time.

That entails:

• Finding ways to process the surgical images quickly, automatically and accurately in a highly dynamic surgical scene.
• Identifying the relevant anatomical structures in the images automatically through a process called image segmentation.
• The alignment of pre-operative and surgical images, known as image registration.

The idea is to “transform the pre-operative images and models of anatomy within so they are in the correct position in accordance to the surgery,” while accounting for movements during an operation such as breathing,” explains Hamarneh.

“We also want to be able to manipulate the pre-operative images so they fit together with images taken after tissue retraction, tearing or resection.”

The team will be using advanced computer vision, 3D image and geometry processing methods, and optimization and machine-learning algorithms to overcome the hurdles.

“Not only is robotic surgery extremely useful for current minimally invasive surgery procedures (e.g. due to its higher precision), but it is also the key technology for telesurgery and future unmanned surgeries,” says Hamarneh, “so it is poised to become one of the most important forms of surgery in the coming decades.”

“We want to push the limits in medical image analysis to solve the problems of doing all this quickly, accurately, automatically and robustly.”

Applying medical imaging to skin cancer

PhD student Maryam Sadeghi is attracting international accolades for her research into a new medical-imaging computer application that can analyse and detect early features of malignant melanoma.

At last year’s World Congress of Dermatology, dermatologists selected the Iranian born computer engineer’s research from among 2,800 submissions to win an e-poster gold award for excellence in research and novelty.

Sadeghi, who came to SFU in 2008, spent the past four years meeting regularly with dermatologists and skin cancer patients to understand how malignant melanoma is diagnosed.

She learned to use a digital microscope camera to capture images of moles, or skin lesions. “It was so difficult for me to look at these images, at first,” she says. “It was upsetting.”

She also studied hundreds of images in an enormous dermoscopy “atlas” used to train dermatologists in diagnosing skin cancers.

She then began translating the information into algorithms that can examine dermoscopic images of skin lesions to segment lesions from the normal skin.

Next, she created algorithms that can recognize some critical indicators of melanoma in the images: a pigment network and streaks as positive features of cancer, and scaling as a negative feature.

So far, Sadeghi’s research has been very promising, earning her the front cover of the March 2011 issue of Computerized Medical Imaging and Graphics Journal.

“Over a set of 500 images from the dermoscopy atlas, our results show an accuracy of 94.3 per cent in classifying whether the pigment network is present or absent,” she says.

Sadeghi hopes her research will lead to a useful medical-imaging application that general practitioners can use to improve their skin cancer diagnoses and referrals.

“This would be good for countries where they don’t have easy access to dermatologists,” she says. “The program will be able to highlight clues and correlate them so that doctors can make better decisions.”

New tool enhances view of muscles

Associate professor James Wakeling is adding to the arsenal of increasingly sophisticated medical-imaging tools with a new signal-processing method for viewing muscle-activation details that have never been seen before.

Wakeling, who teaches in the biomedical physiology and kinesiology department, is fascinated with the mechanics of muscle movement in people and animals.

He’s particularly excited about a novel method he has developed using ultrasound imaging, 3D motion-capture technology and proprietary data-processing software to scan and capture 3D maps of the muscle structure—in just 90 seconds.

It’s a medical-imaging breakthrough because previous methods took 15 minutes to do the job—far too long to ask people to hold a muscle contraction.

The key to the breakthrough is the way the software processes the data, says Wakeling, who developed the program with graduate student Manku Rana.

“Now, we can get people to do muscle contractions and we can actually see how the internal structure of the muscle changes,” he says.

Wakeling’s goal is to improve the muscle models used in musculoskeletal simulation software, which predicts how people move and gauges the forces on their joints.

Current packages are missing important information about muscle contraction, such as how the muscle shape changes, how it bulges, or how the internal muscle fibres become more curved—all details that Wakeling’s technology can capture.

His research could ultimately lead to new software for predicting the outcome of orthopaedic surgeries such as tendon-transfers to treat conditions such as cerebral palsy in children.

“We’re poised to start making new observations and insights,” he says, “and to do new experiments that haven’t been possible before.”