Password Magazine - Issue 31
|
February 2008
|
|||||||
|
Fans of television medical dramas will recognize the scene: dedicated doctors examining hundreds of computed tomography (CT) images, racing to find a diagnosis and save lives. Ironically in the real world, medical scans now deliver so much data that doctors barely have enough time to use it to best effect. But that’s about to change thanks to new automated image analysis and 3D organ modeling.
When Wilhelm Röntgen made the first shadowy X-ray images of his wife’s hand in 1895, he could hardly fathom where those initial steps would lead. Today, medical imaging shows doctors the structures and organs inside the body in remarkable detail and clarity, using an array of non-invasive technologies of which the X-ray was the first. The X-ray is still at the heart of numerous imaging techniques including radiography, angiography, fluoroscopy and CT. Over time, this group has expanded to include others such as magnetic resonance imaging (MRI), positron emission tomography (PET), single photon emission computed tomography (SPECT) and ultrasound. As the number of modalities (the term used for types of imaging technology) has grown, so too has the number of scans conducted. In the USA alone, the number of CT scans grew from 30 million in 1999 to 60 million in 2006. The data bottleneck With the growth in computing power over the last 30 years, systems capable of producing 3D and even 4D (adding time to show organ movement) representations of our insides are now common. The new imaging capabilities lead to more personalized organ models, better diagnosis and easier surgical interventions. Yet these advances have also led to a dilemma – clinicians now have access to so much data, they don’t have time to deal with it all. For instance, a single cardiac CT scan may generate a sequence of ten 3D images, each made up of 300 2D ‘slices’. But analyzing all this data can take a highly trained specialist an hour or more, time which is better spent on face-to-face patient care. Rapid analysis According to Guido Pardo-Roques, Senior Director of Global CT Research at Philips Healthcare, there’s a twopart solution. “The first step is to decrease the time it takes to produce images in order to provide more reliable clinical information,” he says. “The second is to automate image analysis enabling faster and more accurate diagnosis and therefore treatment.” The latter is where Philips has recently taken a big step forward, as Jürgen Weese, Principal Scientist at Philips Research, explains: “Speeding up image analysis is vital for the healthcare industry, especially in cardiac CT. To simplify the work of healthcare professionals, we’ve developed automated analysis techniques that provide an accurate model of a person’s heart in just ten seconds.” Besides its speed, the technology’s ability to provide personalized models is truly innovative. Previously, doctors planning a procedure or making a diagnosis had to rely on generic 3D models and 2D images. But by using advanced boundary-detection algorithms (see More), Philips technology overlays the patient’s CT scan data onto a standard reference model creating a patient-specific 3D representation that greatly aids diagnosis.  A helping hand The technology is also intended to help doctors in the planning and execution of image-guided procedures such as minimally invasive heart repair. A prime example is cardiac radio-frequency ablation, a common intervention to correct irregular heartbeat involving the insertion of a catheter into the heart chambers. Until now, cardiologists relied on X-ray fluoroscopy to guide their instruments. However, while X-ray fluoroscopy shows bone and instruments clearly, it reveals little of the structures of the heart. Philips’ new automated analysis and 3D organmodeling techniques allow a realistic model of the patient’s heart to be superimposed on the fluoroscopy images. As the model accurately describes the anatomy and includes important clinical information such as cardiac ‘landmarks’ like heart valves, cardiologists can position instruments more easily. Weese likens the effect to switching the lights on in a dark room. The technology has already been integrated into Philips’ EP Navigator workstation and feedback from clinical tests supports the idea that it may lead to a significant breakthrough. Reza Razavi, Professor of Pediatric Cardiovascular Science at King’s College, London (UK), explains: “The software revolutionizes the ease with which we can analyze the very large datasets that are now being produced by our CT and MRI scanners. Not only is this a great value in diagnostics, but it also allows us to quickly integrate our 3D images into our Xray interventional program, which is of particular help in ablation of arrhythmias.” Not just for the heart Philips is not the only company pursuing ways to speed image analysis. But the ability of its technology to create dynamic and accurate models of the heart – showing the four chambers, the myocardium and the major attached vessels – puts it at the forefront of innovation. Looking ahead, the team behind the technology is working on extending it to all modalities, adding even more detail and creating reference models for common anatomical variants (for instance, patients with five pulmonary veins rather than the normal four).  Personalized diagnosis The team is also looking for ways to apply the techniques to the entire body. They’ve already developed an experimental system that extracts a 3D model of the lungs, the ribcage and the spine from CT scans. Long term, the hope is to help radiologists identify lung diseases and even model a patient’s entire breathing process. This would greatly improve accuracy and patient care in radiotherapy image-guided treatments of lung tumors. As these image-analysis and modeling techniques continue to evolve, it appears that they not only provide a way past the data bottleneck, but also open the door to new levels of personalized diagnosis and treatment. |
|