Case Study: Frost
Pharmaceutical company asks Merge OEM for sophisticated image processing support application for new MR angiography agent
Merge OEM's segmentation technique eliminates imaging artifact problem in revolutionary new MRA contrast agent
 |
||||||||||||||||||||
 |
||||||||||||||||||||
 |
||||||||||||||||||||
|
||||||||||||||||||||
A pharmaceutical company had developed a novel contrast agent for magnetic resonance imaging, but without a further image processing step the agent's clinical value would be limited. The OEM contacted Merge to custom engineer an advanced segmentation technique that would zero in on the patient's arteries only, "hiding" veins that would otherwise obscure the image and hinder the agent's market acceptance, and the company's revenue growth.
Many MRA techniques involve "bolus chasing" in which a contrast injection is quickly monitored and imaged under MRI. An angiogram builds up as the bolus of contrast moves through the arteries. The pharmaceutical firm's contrast agent in this project, however, works through an entirely different mechanism. Administered shortly before imaging begins, the agent highlights vessels (versus the contrast bolus) in the MRI image, affording many clinical advantages. A critical disadvantage is that both arteries and veins are imaged equally – unlike bolus chasing, in which image acquisition is halted before the contrast enters the venous system. Since most disease occurs in arteries, many radiologists consider the presence of veins in the image "venous contamination." This perception could greatly hinder market acceptance of the new agent, and the Pharma company looked to Merge OEM to find a post-processing fix.
The goal of this project was to segment veins in the MRI volume, separate them from the arteries and suppress the signal from the contrast-filled vessels – in a process termed Artery/Vein Separation (AVS) and Suppression. Merge's experts in segmentation techniques and application building were instrumental in advancing the project.
Technically, Merge needed to evaluate several techniques to find one that optimally segmented tubular structures, such as veins, from 3D volumes with a minimum of user interaction. The process became particularly complex as veins branched, and where veins and arteries were in close proximity. The segmentation technique could not mistake arteries for veins.
A further technical innovation was development of a form of differential MIP rendering (Maximum Intensity Projection) that was both fast and able to suppress segmented structures – in this case the venous anatomy.
In both segmentation and MIP processes, users could control the amount of any venous contamination that might occur via the user interface. Merge deployed its SBEM (Standardized Biomedical Engineering Methodology) process, running an iterative model to give fast trials of techniques and review by the client. Merge liaised with the OEM and a Merge OEM research partner at a European university.
Following successful research and development phases of the project, Merge was able to offer the OEM customer a range of flexible business models to support the introduction of the company's agent and its support in the field.