As cameras get smaller and software becomes more powerful, imaging technology is being used for an ever broader range of medical applications, including bladder cancer detection, image-guided surgery, neuroimaging and blood cell imaging.
Bladder cancer detection
Scorpion Vision has been leading a research project that aims to provide early detection of bladder cancer (globally the fourth most common cancer in men and the seventh in women) using imaging techniques. Scorpion’s Paul Wilson became involved as a result of his wife, who worked with a clinician collaborating with a team at Cambridge University, who had pioneered the use of antibodies to stain cancer cells.
‘The technique involves dipping a cell sample in a particular type of protein,’ explains Wilson, ‘which then binds itself to any cell that is turning cancerous. It’s incredibly accurate.’
The technique involves the application of an antibody to a specific protein found in the nucleus of bladder cells recovered from urine. When this ‘bio marking’ of cells is combined with existing techniques, early studies indicate they can provide a very accurate aid to diagnosing cancer.
Once the process had been proved, the next step was to find an automated way of monitoring these cells, and verifying which cells may be turning cancerous – which is where vision technology comes in.
‘We took it on one step further, and devised a way of analysing the structure of cancerous and normal cells using algorithms,’ says Wilson. ‘We sat with the pathologists to establish exactly what it is they looked for in a cell, which included characteristics such as the relationship between the nucleus and the cytoplasm, shape and boundary of the cytoplasm, and so on. So, we customised our software package to identify all the characteristics used manually by clinical experts with those using the antibody such as colour and numbers.’
When analysed by the software Scorpion developed, this combination provides an accurate indication of cancer being present or not.
Scorpion Vision’s prototype system involves a microscope, a Sony camera, an x-y stage, and its own vision software. ‘We had to import some open CV elements [Intel vision libraries] in order to cope with the complex algorithms in this particular application. The analytic process divides the sample up into sections and then takes images. As it moves from point to point, it analyses each image, refocusing every time.
‘The testing of the system proved it to be repeatable,’ says Wilson, ‘in that thousands of samples were tested multiple times, and the system delivered the same results each time.’
Ultimately, the benefit of such a system will be to automate and therefore speed up the analysis process which, at the moment, is effectively carried out manually by pathologists on a sample by sample basis. ‘The current method is also very subjective, and can lead to errors too,’ says Wilson.
The system is being produced by pathology company Cytosystems, which intends developing it further as a medical device for the ultimate benefit of patients. Cytosystems has been invited to present the clinical results at The European multi-disciplinary meeting in Urological Cancers being held in November 2012 in Barcelona.
Basler GigE cameras are being used in an image-guided surgery system for use in various abdominal surgical procedures. Image-guided surgery is a GPS-like system whose goal is to map the position of the surgical instrument accurately onto the pre-operative CT or MRI data. Feature-rich 3D models of the organs, blood vessels, and tumours can be created from the pre-operative data, and these models provide the surgeon with the ability to see through the organ before and while they dissect, ablate, or perform some other procedure. The magic step is performing the positioning that allows the instrument to be accurately registered onto the pre-operative data and 3D models.
A non-invasive technique for imaging blood cells has been demonstrated by an Israeli research team. Credit: Andor Technology
When working with rigid anatomy such as bone, this registration procedure is straightforward. Fiducial markers that can be readily identified in a scan are attached to the patient before any pre-operative scans are taken. When the patient is brought into the operating room, these points are touched with a tracked surgical instrument and the physical and image spaces can be co-registered. In most abdominal surgery, though, the organ being operated on is not in a fixed position relative to rigid anatomy. The solution to this problem is to register using the surface of the organ in question.
A method was needed to acquire a large set of 3D points on the surface of the organ. A laser range scanner was chosen to perform this task. This scanner remains stationary while moving a laser line across the organ and simultaneously recording the position of the line using a CCD. Using triangulation, a 3D surface can be reconstructed. The primary requirements for the scanner were that it be able to capture both range and colour texture information from the same CCD and lens. They required high resolution at a faster rate. After exploring several options, the Basler Pilot GigE camera was chosen. Regulatory requirements in medical devices are quite strict, but the resulting laser range scanner passed both IEC 60601- 1 safety and IEC 60601-1-2 electromagnetic compatibility testing without modification. Julie Busby of Multipix Imaging says: ‘The Basler Pilot cameras are a great solution as they give outstanding performance and reliability that is paramount for use within medical applications.’
QImaging is working with the University of Toronto to develop a neuroimaging system for studying the effects of stroke recovery and epilepsy. The research is being led by assistant professor Ofer Levi.
‘When I first started looking into this, I realised that the imaging technology being used in optical imaging of the brain was 20 years old,’ says Levi. ‘It could barely do anything either. We began to look at what we could do better to image the brain, and quickly identified the need for a quick and stable connection between a light source and the camera.
‘The VCSEL [vertical cavity surface emitting laser] proved to be the most suitable light source, as it provided high brightness in the desired red and near IR wavelengths. We could also switch them very quickly from operating like a high-brightness LED to operating like a laser.’
For the camera, Levi turned to QImaging, as the company had a reputation in the biomedical sector of scientific imaging.
Chris Ryan, product manager at QImaging, says: ‘Ofer had seen the results that imaging could provide, and wanted us to develop an intrinsic imaging system using optical components and software, that would monitor blood flow and associated oxygenation levels.’
Over time, Levi used a number of different products from QImaging, before settling on the Rolera EM-C2 EMCCD camera, which has a number of characteristics that make it appropriate to the work he is doing, such as high frame rates and low read noise.
‘It’s a dual modality process,’ continues Ryan. ‘The team are measuring blood flow and oxygenation at the same time. They’re looking to track very small changes at very high speed. This necessitates a certain dynamic range in order to detect those changes at a magnitude of 1 in 5,000.’
The prototype system is being tested in vivo on mice and rats, where the image is taken of the cortex (the outer layer of the brain) using a VCSEL. The camera operates as the master, and the VCSEL as the slave. The system can be programmed to cycle through a predetermined sequence of light sources, where the different light sources are suited to measuring different elements, such as oxygenation or blood flow. Depending on the field of view, the speed can be up to 200fps, but for 500 x 500 pixel field, it can operate at 60fps. Levi has worked together with QImaging to develop software that allows real-time acquisition, with the only limitation being the speed of the PC. ‘We have not yet reached the limit of the camera,’ says Levi.
Levi and his team are collaborating with neuroscientists to develop a robust system that delivers as accurate a scientific measurement as possible. At the same time, Levi is looking at a CMOS-based system, which would offer a degree of portability, and this could, in time, lead to clinical studies within a couple of years.
‘When we can convince the neurosurgeons that what we are showing them is valid, and that they can use the images to pinpoint the location of a seizure, then they could be using this system to guide surgery,’ concludes Levi.
Blood cell imaging
An Israeli team has demonstrated a non-invasive technique for imaging blood cells in vivo that could eliminate the need to extract blood from many patients. Powered by the Andor Newton Electron Multiplying EMCCD camera, its high-resolution Spectrally Encoded Flow Cytometry (SEFC) probe offers primary care physicians the capability to detect directly a wide range of common medical disorders, such as anaemia and bacterial infection, and potentially life threatening conditions, including sepsis, thrombosis and sickle cell crisis.
Demonstration of the system components used by the University of Toronto research team, which features the QImaging Rolera EM-C2 EMCCD camera. Credit: QImaging and University of Toronto
As well as enabling an immediate medical response to be offered, SEFC could also allow large-scale screening for common blood disorders. Vitally, its ability to directly and continuously visualise blood cells flowing inside patients could also provide an early warning of a medical emergency, such as internal bleeding, in post-operative and critical-care conditions.
SEFC was developed by the Biomedical Optics Laboratory, headed by Dr Dvir Yelin, at the Technion-Israel Institute of Technology in Haifa. The focus is the application of advanced optics to address some of today’s clinical challenges, particularly the development of non- or minimally-invasive diagnostic tools.
According to Lior Golan, one of the researchers at the Biomedical Optics Laboratory, two major challenges needed to be solved. ‘SEFC images of fast-moving blood cells are acquired from deep under the surface of the skin through tissue that scatters the light. This means that very little light is available and necessitates the use of a spectrometer equipped with a high-speed line camera. The Andor camera provided our team with a combination of high sensitivity and the required line rate for imaging physiological blood flow. Switching between 2D image and full vertical binning mode on the camera also made the alignment of the spectrometer very easy and the ability to customise the LabView software development kit to control the camera was very convenient.’
Blood analysis is also being tackled by Biosurfit, which has developed the Spinit, a technological platform for Point of Care Testing (POCT). Point-of-care testing is medical testing at or near the site of patient care. The driving notion behind POCT is to bring the test conveniently and immediately to the patient. This increases the likelihood that the patient, doctor, and care team will receive the results quicker, which allows for immediate clinical management decisions to be made.
Using a combination of Point Grey cameras, proprietary software, and easy to use DVD-styled test cartridges, the Spinit takes one small drop of blood and gives precise results within 15 minutes. It is also capable of multi-parameter testing, and the same Spinit unit will be able to conduct different tests by using different cartridges.
A technician places a single drop of blood onto the test panel cartridge and all the blood sample preparation is done automatically. The sample flows through a series of microfluidic structures, where different operations are made including separation of blood components and mixing with other solutions. After preparation, the sample is directed to a number of detection zones on the cartridge.
Each detection zone includes a biological recognition layer (BRL) developed using antibody fragments attached on a gold surface. By using different antibody fragments and immobilisation strategies, the gold layer will act as a selective trap for a specific blood biomarker. When the patient’s blood passes through the BRL, the targeted biomarker binds to the antibody fragments. It is the concentration of trapped biomarkers that the Spinit is testing.
The Spinit uses Surface Plasmon Resonance (SPR), a well-established spectrometry technique, to measure the concentration of biomarkers. This is done by measuring the refractive index of light reflecting off the gold’s surface after exposure to the patient’s blood sample: the higher the concentration, the lower the index.
The optical module uses a Point Grey Firefly MV FireWire monochrome camera acquiring images at almost 160fps with an ROI of 50 x 800 pixels. The light source is a laser that goes through a polariser and a focusing lens. A 785nm laser utilising near IR wavelength and is always on while performing the measurement. The cartridge is spinning, much like a regular CD, and the camera is triggered when the phototransistor behind the semi-opaque mirror detects a reflection coming from the detection zone. The camera is used to acquire 3,000 images of each detection zone over a 10 minute period.
The Firefly MV has a number of advantages for the Spinit. As well as its affordability, the compact size of the camera is important as the Spinit’s small ‘desktop’ footprint measures just 215 x 236 x 306mm and weighs roughly 4kg. Additionally, Point Grey’s FlyCapture software supports Linux and allows for excellent control of acquisition parameters such as software triggering at high frame rates and variable ROI.
The Spinit will be released later in 2012 and the first cartridge will measure C-Reactive Protein (CRP). CRP is an acute phase protein actively circulating in the body during an inflammatory response. CRP is a standard test parameter used by clinicians to reduce the chances of inadequately prescribing antibiotics and improve the effectiveness of treating other infections. Future tests to be added include total blood count, a diabetes panel and a cardiac panel.