Development of Ultra-Fast MRI relaxation time measurements: Application in Image Guided Therapy – FastRelaxMRI
Development of Ultra-Fast MRI relaxation time measurements: Application in Image Guided Therapy
Magnetic Resonance Imaging (MRI) is a non-invasive imaging technique that obtain images with multiple contrasts: we talk about image weighting. Each tissue possesses specific relaxation times (T1, T2, T2*), that a pathology or a treatment can modify. Quantitation is necessary in order to obtain a non-subjective scale to compare patients between them, to assess the evolution of a disease, to precisely evaluate the efficacy of a treatment, to monitor a minimally-invasive procedure in real-time.
Issues and Objectives
Several issues prevent the routine applications of MR mapping sequences.<br />The major problem is the long acquisition time. Some techniques enable to obtain values rapidly, but at the expense of accuracy. In addition, the current mapping sequences are not applicable on every organs of the human body (like lungs, tendons or hard tissues like bones), or when specific conditions are encountered (injection of contrast agents) due to the presence of short T2*. The resulting signal drop and susceptibility artifacts become frequent as the use of high magnetic fields (?3T) is getting more interest.<br />Finally, MR thermometry enables to monitor temperature, but remains blind when ablations are applied within short T2* environments, like near metallic devices or during cryo-ablations. A precise quantification of the relaxation times and thus of the temperature would bring critical informations on the control and efficiency of such therapies. In this context, however, another constraint appears : the real-time acquisition and reconstruction of the images.<br />This project focuses on the development of fast relaxation times measurement strategies that could be used anywhere on the body. <br />Consequently, the goal is double: to develop robust quantitative MR sequences that will be compatible with real-time monitoring.<br />Numerous methods have been developed to rapidly quantify relaxation times. The Variable Flip Angle (VFA) technique is the fastest to obtain T1 maps. Nevertheless, it is very sensitive to the nutation angle, preventing its application in 2D to further fasten acquisition duration. Other methods have been created to obtain T1, T2 and T2* values within the same acquisition or multi-parametric maps using the fingerprinting technique. Drawbacks are their complexity, especially in post-processings, and the long acquisition times, preventing their applications for ultra-fast mapping and consequently for real-time monitoring.<br />In the field of MR thermometry, the Proton Resonance Frequency technique fails in measuring temperature as soon as short T2* environments or when adipose tissue are present. As relaxation times have been shown to correlate with temperature values, a 3D VFA sequence was used to measure T1 of the skull during hyperthermia, but required 8min.<br />Consequently, the goal of this project is to develop innovative MR approaches in order to generate relaxation times maps, rapidly and accurately. The originality of the FastRelaxMRI project relies in the combination of multiple techniques within a single multi-parametric sequence so that it can be applied for multi-organ, multi-application human research and clinical routine. The major concern relies on the implementation of strategies that are robust enough to overcome difficult acquisition conditions, like respiration-affected organs, tissues with short T2* and the use of >1.5T high magnetic field.
In order to obtain accurate relaxation times maps very rapidly and in difficult conditions, we will base our work on Gradient Echo (GRE) sequences. Contrarily to Spin Echo sequences, there is no need of refocusing pulses that lengthen acquisition time and can be difficult to implement, especially at high magnetic fields. Non-cartesian k-space encodings, especially the ones that enable to reach an ultra-short echo time (UTE), will be implemented. This should widen the application of the GRE sequences to moving organs of the body, hard tissues, short T2* tissues and at high magnetic fields.
Our strategies will also have to be compatible with standard and innovative acceleration techniques (like Compressed Sensing). We will focus our developments on 2D acquisition strategies, due to their rapidity.
These methodologies will be implemented within two strategies. The MP2RAGE (Magnetization Prepared Two RApid Gradient Echoes) method enables to obtain 3D T1 and T2* maps of the human brain. Nevertheless, it is not adapted to either ultra-fast relaxation times measurements (it was never applied in 2D), or acquisition on short T2* tissues. In parallel, the Triple Echo Steady State (TESS) sequence can generate 3D T2 maps and potentially T1 maps. Due to its sensitivity to signal intensity, long durations are needed, making this sequence incompatible with real-time mapping in the current state. Also, it is inherently motion- sensitive.
The first goal of these methodology developments will be to obtain two 2D maps simultaneously (either T1 and T2* or T1 and T2 depending on the sequence used) in less than 5 seconds.
The second goal is to implement a motion correction module and an acceleration strategy within the previously developed sequences, in order to obtain T1, T2 and T2* maps of non-moving areas (brain, knee), motion-affected organs (liver), and motion-affected short-T2* organs (lungs) very rapidly on living organisms (humans and rabbits).
Finally, we intend to apply the new methodology to one of the new cancer treatments that get increasing interest: cryo-ablation.
After multiple methodological developments that we developed on preclinical small-animal systems (TRL 1-3), we intend to transfer our methodology to human imaging (TRL 4-5). This project should enable to obtain parametric maps of human volunteers in a similar acquisition time than weighted MR images that are currently obtained in clinics. Our ultimate goal is to apply our new methodology in clinics (TRL 6-7). Through the MRI constructor Siemens, we will include relaxometry in clinical protocols world-wide, for more accurate diagnosis, prognosis and response to treatment evaluation, and interventional MR imaging. Consequently, the grant will enable to go further towards the translation pipeline.
The new ultra-fast parametric sequences could be useful for a wide range of applications, including brain pathologies, like Multiple Sclerosis and cerebral ischemia, but also abdominal pathologies like renal ischemia or liver cirrhosis, and MR-guided therapies like cryo-ablation.
In addition, through the decrease in acquisition and reconstruction time, the new developments could lead (1) to less cost related to MR and (2) to screening many more patients. For patients, these advances should limit their discomforts, and enable more repetitive exams in order to follow responses to treatment or to maintain the surveillance of patients, like in cancer patients at-risk of developing metastases. On a fundamental point-of-view, these improvements could help in optimizing interventional MRI, and also clinical protocols. This could become the reference method to quantitatively-guide cryoablations.
The results will be published in international scientific peer-reviewed journals, like European Radiology, MRM, JMRI. Also, the results will be presented in international conferences, like ISMRM, in order to obtain feed-backs from our peers on our methodology and then improve our experiments before publication.
All codes that will be developed for MRI data acquisition and reconstruction will be available online via Gadgetron. In addition, the MR data will be available upon request.
Magnetic Resonance Imaging (MRI) enables to obtain images with high contrasts. Nevertheless, there is a need to develop quantitative methods to obtain a non-subjective scale in order to assess the evolution of a disease, to precisely evaluate the efficacy of a treatment, to monitor minimally-invasive procedures. The actual methods of relaxometry are either time-costly, not accurate, sensitive to respiration-motion or not applicable on short-T2* areas (lungs, bones, surrounding metallic implants). Consequently, this project focus on the development of the measurement of relaxation times in all these difficult conditions. This technical objective gets even more challenging as we intend to apply them for MR-guided therapies, whose specificity is real-time monitoring (both image acquisition and reconstruction).
Therefore, the goal of our project is to develop multi-parametric sequences to obtain accurate relaxometry measurements on every organ of the body, and ultra-rapidly for numerous applications. This project is crucial before a world-wide clinical use in order to reduce exam cost, patient discomfort, waiting lists and to increase the accuracy of diagnoses, prognoses and follow-ups of patients. In addition, the project is of huge interest for interventional MR-guidance, especially cryoablation therapies, in need of quantitative real-time data to ensure their safety, their efficiency and their proper use.
The project will involve the implementation of 2D non-Cartesian encodings of the k-space and Ultra-short Echo Time (UTE) into MR Gradient Echo sequences on a human research MR system at 3T. An innovative MR sequence enabling to simultaneously measure T1, T2 and T2* relaxation times will be developed.
Acceleration techniques including Simultaneous Multi-Slice and Compressed Sensing, in parallel with motion-correction methods (using a Self-Gating module and post-processing in the image domain) will be implemented. Also ultra-fast reconstruction algorithms will be developed through the Gadgetron open-source software.
These methodological developments will take place during the first two years, before any application on human volunteers. Then, high-resolution parametric maps of each body area (brain, thorax, abdomen, knees) are expected to be obtained in less than 3 min, which is faster than the current time required to obtain weighted-images for similar spatial resolutions.
The new methods will then be applied on a large-animal (rabbit) model of lung tumors. The intended spatial resolution will be 0.5mm in-plane in order to obtain multi-parametric maps of early-growing tumors.
The last years of the project will focus on the quantitative characterization of the temperature drop during a percutaneous cryoablation of the rabbit lung tumor in vivo. To do so, temperature values of the tumor and the surrounding lung region should be obtained from the parametric maps every ˜10s.
In summary, the originality of the FastRelaxMRI project relies in the combination of multiple techniques within a single multi-parametric sequence so that it can be applied for multi-organ, multi-application human research and clinical routine. Developments within each technique (encodings, motion correction, acceleration) and developments to coordinate them are the novelties of the project.
The improvements obtained in this project will be tremendous assets for the diagnosis, prognosis and surveillance of multiple pathologies, and could become the reference method to quantitatively-guide cryoablations.
Project coordination
EMELINE RIBOT (CENTRE DE RESONANCE MAGNETIQUE DES SYSTEMES BIOLOGIQUES)
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
Partnership
CRMSB CENTRE DE RESONANCE MAGNETIQUE DES SYSTEMES BIOLOGIQUES
Help of the ANR 299,952 euros
Beginning and duration of the scientific project:
February 2020
- 48 Months
