Magnetic Resonance Fingerprinting of Patients with Gliomas

Advanced conventional MR Imaging of brain tumors is mainly based on contrast images and is time consuming. To overcome these limitations MR Fingerprinting (MRF) offers a fundamentally novel approach to data acquisition, post-processing, and visualization in MR using a fast, single MR sequence. MRF uses a single pseudo-randomized acquisition that causes the signals from different tissues to have a unique signal evolution or fingerprint that is simultaneously a function of the multiple tissue properties. The postprocessing involves a pattern-matching algorithm to identify the fingerprints with the closest match in a predefined dictionary of predicted signal evolutions that can then be translated into quantitative maps. From these MRF derived T1 and T2 relaxation times can be used to generate synthetic contrast images. Our hypothesis is that MRF will offer a fast single sequence for the evaluation of brain tumors which may replace several conventional MR sequences, so far used in an advanced brain tumor protocol. The simultaneously acquired T1 and T2 maps may offer information on the grade and extension of gliomas, which so far can only be gained from a conventional advanced brain tumor protocol. In the follow-up study an enhanced detection rate of recurrent tumor in high grade gliomas should be possible by MRF. Methods to proof these hypotheses will comprise the following phases: 1) Phantom testing of the MRF acquired T1 and T2 relaxation times and the conventionally acquired T1 and T2 relaxation times 2) in vivo application of the validated MRF sequence and protocol optimization on 10 volunteers. 3) in vivo application of MRF on 50 patients with low-and high grade gliomas: Validation of MR based evaluation on tumor, the tumor surrounding tissue and normal appearing brain tissue with histological specimen analyses derived from surgery. MRF-obtained T1 and T2 relaxation times in one sequence of 5 minutes will be compared to conventional advanced MRI tumor protocol which takes about 60 minutes. -1-

Advanced conventional MR Imaging of brain tumors is mainly based on contrast images and is time consuming. To overcome these limitations MR Fingerprinting (MRF) offers a fundamentally novel approach to data acquisition, post-processing, and visualization in MR using a fast, single MR sequence. MRF uses a single pseudo-randomized acquisition that causes the signals from different tissues to have a unique signal evolution or 'fingerprint' that is simultaneously a function of the multiple tissue properties. The postprocessing involves a pattern-matching algorithm to identify the fingerprints with the closest match in a predefined dictionary of predicted signal evolutions that can then be translated into quantitative maps. These MRF derived T1 and T2 relaxation times can be used to generate synthetic contrast images. In this project together with Siemens Healthineers an optimized MRF sequence for neuro applications was developed, implemented and applied in a clinical trial. An important finding in this project was the detection of motion sensitivity with the MRF sequence and the development of strategies to visualize and overcome these artefacts, which were internationally not known before. Co-registration between images acquired during MRF session and an advanced tumour protocol session with multiple sequences and quantitative assessment was performed in a collective group of 35 measured patients with histological results from biopsies and surgeries as a standard of reference Similarity between relaxation times derived from MRF and conventional techniques was evaluated. As expected, high correlations between MRF T1 and T2 and the respective conventional quantitative MR techniques could be demonstrated. Evaluation of pathological tissues of interest on the co-registered images with respect to the quantitative tissue parameters has been performed in four regions: solid part of the tumour, perifocal oedema, perilesional normal appearing white matter, and contralateral white matter of the frontal lobe with the result of a clear differentiation of these tissues by MRF derived parameters. Furthermore, it could be shown that MRF T1 and T2 values of solid tumour parts were significantly higher in IDH-mutant tumours than in IDH-wildtype suggesting the potential of using non-invasive MR imaging, particularly MRF-derived T1 and T2 relaxation values, as a biomarker for differentiating gliomas with respect to their IDH mutationan important influencing factor for individual therapy planning and prognosis. Additionally we have investigated the relationship between within-pathology MR relaxation values acquired with MRF and other metabolic imaging methods such as Positron Emission Tomography (PET) and MR spectroscopy at 7 Tesla in the same cohort of patients. The comparison MRF with PET has shown that both MRF and PET provide quantitative measurements with high predictability on grading and subtumour regions of gliomas. Nonetheless, results were not similar, reflecting the different underlying mechanisms of each method.