Making MRI even faster: FLASH 2
Within the newly founded Biomedical NMR company, Frahm and his team continued with their MRI research, but struggled for many years to achieve any further image acceleration. Then, in 2010, they succeeded in developing a new high-speed MRI method, “FLASH 2”, which reduces the acquisition time for a single image down to as little as 20 milliseconds or one-fiftieth of a second. This technology makes it possible to record live videos of moving organs and other physiological processes at a rate of 50 frames per second with high image quality. Major diagnostic applications of FLASH 2 include real-time MRI studies of the beating heart, the flow of blood or cerebrospinal fluid, joint movements and articulation and swallowing processes, as well as minimally invasive surgery.
The original FLASH technology significantly
reduced MRI acquisition times and enabled continuous imaging, though still at a
temporal resolution of about one second. It was therefore not yet fast enough for
true real-time imaging of moving objects. Instead, it relied on a
synchronisation of the MRI data acquisition with the electrocardiogram and some
retrospective sorting of data from different heartbeats to generate a single
synthetic “cardiac cycle”.
Frahm and his team overcame this problem by developing a new MRI technology which enabled the desired real-time imaging. FLASH 2 extends the FLASH principle by the use of extreme (e.g. 20-fold) data undersampling and a novel type of image reconstruction. Undersampling is the acquisition of a few spatially encoded data for each image and directly achieves a corresponding acceleration or reduction in measuring time. Serial image reconstructions are then defined as iterative solutions to a nonlinear inverse problem where the ill-conditioned numerical problem for the actual frame is effectively constrained by temporal regularisation to a preceding frame. In other words, FLASH 2 takes advantage of the similarity of successive frames in an MRI film at high temporal resolution.
Technically, FLASH 2 also exploits a low flip angle gradient-echo technology for rapid continuous imaging, but uses a different spatial encoding scheme, wherein the data for a single MRI frame is sampled using radial “spokes” rather than rectilinear (parallel) lines.
Figure 4: Radial sampling co-ordinate system with 15 spokes
Secondly, in order to further reduce
imaging time, an undersampling approach is used in which for a single frame
only a small and equally distributed subset of spokes is sampled. For the
following frames, the undersampled radial spokes are turned by a given angle
such that a temporally interleaved spoke arrangement is achieved.
Figure 5: Undersampled temporally interleaved spoke arrangement
According to FLASH 2, each frame is reconstructed from its undersampled radial MRI data using a regularised nonlinear inversion algorithm. Despite a high computational demand, the algorithm allows for real-time image reconstruction when using a high degree of parallelisation and a dedicated computer with graphical processing units. This computer may be connected to an existing MRI system in order to bypass the conventional image reconstruction method.
Taken together, the faster speed of FLASH 2 is the result of the use of extremely undersampled radial acquisition schemes in conjunction with advanced numerical mathematics to solve the reconstruction problem. When implemented on multiple graphical processing units, FLASH 2 allows for true real-time MRI: acquisition, reconstruction and display of serial frames in real time.
Due to the high commercial potential of FLASH 2, the Max Planck Society filed two PCT patent applications and entered the national phases before the European Patent Office and the United States, Japan and Chinese Patent Offices. The advantage of using the PCT route is that applicants can defer patent prosecution and translation costs until the commercial value of the invention becomes clear. The FLASH 2 applications were all granted and can therefore be licensed to interested parties.
What do you think Max Planck Society’s business model will be this time?
Commercialising FLASH 2
In many respects, the possibility to
obtain real-time MRI recordings rather than still images came as a shock to the
radiology community. Over the previous 25 years, while MRI diagnostics had been
refined, they had not really changed, and a whole generation of radiologists
and clinicians had either accepted that it was not possible to use MRI to study
joint movements or swallowing processes, or had learned to use means such as
synchronisation to the electrocardiogram to carry out cardiac or flow MRI
examinations. In other words, once real-time MRI had been invented, its
clinical translation into routine medical applications asked for a change of
paradigm. To achieve this goal requires vision, courage and extensive clinical
trials. The benefits to be expected are manifold: patients will no longer need
to hold their breath as they do for many of the current examinations or they
will undergo a more comfortable MRI examination instead of a stressful non-MRI procedure
(e.g. reflux diagnostics). Clinicians may develop new or simplified diagnostic procedures
(e.g. providing access to patients with swallowing disorders or cardiac
arrhythmias). Economic advantages are also expected thanks to shorter
examination times (e.g. when using comprehensive real-time cardiac MRI protocols).
The strategy for commercialising FLASH 2 therefore focuses on making the product known to radiologists and clinicians, creating the medical need for it and finding interested partners among medical MRI manufacturers. To this end, a two-stage commercialisation procedure is used, comprising a first, intermediate commercialisation stage with mostly university-based radiological and clinical partners, followed by a final commercialisation stage with MRI manufacturers.
In the intermediate commercialisation stage, renowned medical research facilities may acquire low-cost licences for FLASH 2 for patient-oriented research in various fields of diagnostic imaging. It is expected that this research will result in peer-review contributions at conferences and in scientific journals. These publications are expected to broaden and change the current practice and in doing so to demonstrate the enhanced diagnostic potential of FLASH 2, which in turn will define clinical need, radiological demand and commercial interest. Furthermore, these research activities are expected to provide valuable feedback to Frahm and his team which will enable them to further optimise FLASH 2. At present, FLASH 2 can be implemented on Siemens MRI scanners with an additional graphical processing computer under the support of Frahm and his team.
In the second commercialisation phase,
FLASH 2 will be licenced to MRI manufacturers using a number of potential
licencing models. The most commercially promising model is the royalty-based
licence scheme, under which manufacturers pay a fixed amount of money per
product sold. For FLASH, this licencing scheme proved to be very lucrative. Another
licencing model used by Max-Planck-Innovation GmbH (MPI) (formerly Garching
Innovation) is based on milestone payments (following an initial down-payment),
whereby payments are triggered at significant stages of success in the project
for which the licence is granted (e.g. market introduction, sale of the first
100 products, and so on).
Generally speaking, licence agreements with MPI stipulate whether the Max Planck Society or the licensee is responsible for litigation. A clear definition of who is to bear the costs in any litigation proceedings is essential, particularly in view of the very high cost of litigation in the US. If the Society declines to start litigation, as the patent owner it (and MPI) is party to the proceedings but does not have to bear any of the costs.
The FLASH case study clearly shows just how important - and how easy - it is to file patent applications for inventions prior to their first public disclosure. Research facilities should encourage scientists to notify their intellectual property departments as soon as possible after an invention has been made, and before they publish anything about it. The case study also demonstrates that patent owners need to be prepared to go to court to defend their intellectual property rights and that this may involve a significant outlay in terms of both time and money.