Hyperspectral imaging system helps surgeons identify key anatomical structures
Technique could make gallbladder removal safer
Each year nearly a half-million Americans have their gallbladder removed. Now researchers at the University of Texas at Arlington are developing an imaging system that allows surgeons to distinguish between gallbladder tissue, liver tissue and crucial arteries. This could make a common minimally invasive surgery safer than ever.
In most cases, removing the gallbladder — cholecystectomy — is done laparoscopically. In fact, given a choice, few people would choose to have an open cholecystectomy because the laparoscopic procedure is much less invasive. It uses only two or three small incisions 1/4 to 1/2 in. wide compared with one 5- to 8-in. incision under the rib cage. Thus, patients recover faster and with fewer complications.
The surgery depends on a laparoscope, a tube that holds fiber optics, transmits light to the area and displays the operation on a video monitor. Although the surgery is not quite performed by remote control, the surgeon is more removed from the surgical field than would be the case in an open procedure. The field of view on the display is flattened into two dimensions, and he or she can no longer feel the organs to find and avoid arteries or other key structures.
In gallbladder surgery, one important structure that surgeons must avoid cutting is the duct that carries bile from the liver to the stomach. The bile duct is often difficult to see in the surgical field and is connected to the gallbladder and the pancreas. Given that some patients have an irregular anatomy, the duct can be overlooked or misidentified easily. Damaging it in surgery can lead to problems such as infections from bile leaking into the abdominal cavity or a blocked bile duct that causes jaundice from bile backing up into the liver, complications that are painful and potentially life-threatening.
Fortunately, such injuries are rare, averaging perhaps 0.5 percent of all cases. But the public’s tolerance for preventable complications is waning. National organizations, such as the Institute for Healthcare Improvement, insurers, and Medicare and Medicaid, are focusing attention on reducing preventable errors. Medicare, the single biggest health care buyer in the US, has announced that it will no longer pay for treating preventable errors.
Researchers have developed a hyperspectral imaging system that allows surgeons to distinguish between the liver and gallbladder during laparoscopic gallbladder removal. Conventional laparoscopic surgery relies on visible light (left), which highlights the tissue surface and makes the liver and gallbladder difficult to tell apart. Near-infrared light penetrates deeper into the tissues, enabling researchers to create a spectral image that relies on lipids and water to highlight the gallbladder and cystic duct relative to the nearby liver (right). Courtesy of Karel J. Zuzak.
“This has created a pressing need to develop new technologies enabling surgeons to visualize bile ducts during surgery and eliminate bile duct injury complications,” explained Karel J. Zuzak, lead author of a paper on the new imaging system, which was published in the June 15 issue of Analytical Chemistry. Zuzak’s system, which has been tested in preclinical, proof-of-principle studies, used the spectral characteristics of oxygenated and deoxygenated hemoglobin, water and lipids to distinguish tissue. The hyperspectral system relies on a liquid-crystal tunable filter to rapidly wavelength-sweep an image of the field of view that is passed to a focal plane array, from which it is deconvolved and presented as an enhanced image.
The system uses a conventional surgical laparoscope connected to a visible light source with a visible to near-infrared liquid lightguide from Oriel. In the viewing optical path, Zuzak and his colleagues mounted a Cambridge Research liquid-crystal tunable filter and a lens to filter and focus light passing to a Princeton Instruments focal plane array detector. Via custom-built software, the image can be switched to any one of five principal component images, each of which is created by gray-scale encoding each pixel according to the relative contribution of the principal component; for example, water, oxygenated or deoxygenated hemoglobin, or lipids.
Each image provides a different level of contrast between the tissues in adjacent structures. For example, spectra showing a concentration of lipids and water are more likely to be seen in gallbladder tissue than in liver tissue. In addition, the system can plot the averaged spectrum from within a region of interest against the spectra of surrounding tissue.
The group has tested the system in a swine model. According to Zuzak, it takes about 41 s to acquire the spectroscopic images and an additional 8 s to analyze them. “Imaging the anteriorly placed bile ducts obscured by connective tissue ... within 49 s is a vast improvement over current techniques,” he said. “For example, IOC [intraoperative cholangiography, an x-ray technique for visualizing the gallbladder] requires finding the cystic duct, dissecting it out of the hepatoduodenal ligament, inserting a catheter, injecting contrast agents and imaging with x-rays.”
Although video-rate imaging would be very useful, Zuzak explained that surgeons do not require it. “They only need a snapshot at the start of an operation to orient the hidden structures relative to those easily visualized to help them safely proceed.” He added that the group has begun collaborating with the National Institute of Standards and Technology to integrate the system into a clinical device that would allow a surgeon to see both the video-rate surgical field and a hyperspectral image.
He said that this could be done either with two displays or by superimposing hyperspectral data over the white-light image to highlight tissue of interest.
According to Zuzak, the group also is exploring using a spectrally tunable source as an alternative to the rapidly tunable liquid-crystal filter to sort out the spectra. “Texas Instruments’ digital light processing chip provides full digital control over an illumination source,” he said. The chip-controlled source can produce any arbitrary spectra or monochromatic light using micromirrors to select the predetermined wavelengths of light. A wavelength can be tuned in less than 50 μs, compared with about 150 ms for the liquid-crystal filter. That allows the source to scan through a spectrum fast enough that the human eye sees white light, while a camera can be set to collect spectra from relevant wavelengths.
“This spectral light engine can be seamlessly integrated into the surgical environment without major modifications to the existing instrumentation, providing spectroscopic images at video rates,” he said.
The present system is custom-built. Excluding the surgical laparoscope, it costs between $50,000 and $100,000, depending on the individual components, Zuzak explained. “Once commercialized, these devices would become much less expensive.”
Zuzak’s research interest is in translating medical technology from the research bench to the clinical bedside to bring better health to everyone. Toward that end, he initiated collaboration with Dr. Edward Livingston, chairman of gastrointestinal and endocrine surgery at the University of Texas Southwestern School of Medicine. Livingston has developed a keen interest in the project and has become essential to its long-term success by providing the clinical resources and opportunities for utilizing the hyperspectral imager. The key element for developing technology that will provide better health care is the consortium between biomedical engineers and clinicians, because if the technology stays in the lab or works only in an animal, it does no good.
The group has begun testing the usefulness of the technique in human gallbladder surgery. At the same time, according to Zuzak, a team of specialists from the University of Texas at Arlington, the University of Texas Southwestern Medical Center and the NIST Optical Technology Division is pushing ahead with work on a video-rate hyperspectral imaging system.
- Electromagnetic radiation detectable by the eye, ranging in wavelength from about 400 to 750 nm. In photonic applications light can be considered to cover the nonvisible portion of the spectrum which includes the ultraviolet and the infrared.
MORE FROM PHOTONICS MEDIA