Jiahao Lin^1,2, Siyuan Liu², Rock Hadley³, Marvin Bergsneider⁴, Giyarpuram N Prashant⁴, Sophie Peeters⁴, Robert Candler^2,5, and Kyunghyun Sung^1
1Department of Radiological Sciences, David Geffen School of Medicine, University of California, Los Angeles, LOS ANGELES, CA, United States, ²Department of Electrical and Computer Engineering, University of California, Los Angeles, LOS ANGELES, CA, United States, ³Department of Radiology, University of Utah, Salt Lake City, UT, United States, ⁴Department of Neurosurgery, University of California, Los Angeles, LOS ANGELES, CA, United States, ⁵California NanoSystems Institute, Los Angeles, CA, United States

We construct an agar phantom to examine the SNR variation as the coil plane rotates from 0° to 90°, respect to B0. Our local pituitary coil has improved SNR with a factor, ranged from 3.3 to 4.2, compared to the commercial Siemens head coil as long as the surgical positioning of the coil to be within 0° and 70° with respect to the B0 field.

Valéry Ozenne^1,2,3,4, Pierre Bour^2,3,4, Marylène Delcey^2,3,4, Nicolas Cedilnik⁵, Maxime Sermesant⁵, and Bruno Quesson^2,3,4
1Centre de Résonance Magnétique des Systèmes Biologiques, UMR 5536, CNRS, Bordeaux, France, ²IHU Liryc, Electrophysiology and Heart Modeling Institute, Fondation Bordeaux Université, Bordeaux, France, ³Univ. Bordeaux, Centre de recherche Cardio-Thoracique de Bordeaux, U1045, Bordeaux, France, ⁴INSERM, Centre de recherche Cardio-Thoracique de Bordeaux, U1045, Bordeaux, France, ⁵Université Côte d’Azur, Inria, Epione, Sophia Antipolis, France

MR-guidance of electrophysiological (EP) procedures requires manual segmentation of the cardiac cavities either at the beginning of the procedure to produce the roadmap volume or after radiofrequency ablation (RFA) to assess the lesion transmurality in post ablation images. The purpose of this work is to evaluate the feasibility of automatic in-line segmentation in the context of routine preclinical EP studies.

 

Kevin Waescher¹, Simon Reiss¹, Ali Caglar Özen^1,2, Thomas Lottner¹, and Michael Bock^1
1Dept. of Radiology, Medical Physics, Medical Center University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany, ²German Consortium for Translational Cancer Research Partner Site Freiburg, German Cancer Research Center (DKFZ), Heidelberg, Germany

MR-guided catheterization with injection of Gd-based contrast agent allows for quantitative perfusion measurement of the myocardium in coronary interventions. However, Gd is contraindicated in patients with impaired renal function, and it changes the relaxation times in the target organ after injection, so that a characterization of the target tissue becomes difficult. In this study, we provide a different solution to quantify perfusion using arterial spin labelling with a catheter based coil. The results show that ASL using a local catheter based labelling coil provides an alternative technique for perfusion measurements during cardiovascular interventions without exogenous contrast agents.

Robert Moskwa¹, Azam Ahmed², and Walter Block^1
1Medical Physics, University of Wisconsin-Madison, Madison, WI, United States, ²Neurological Surgery, University of Wisconsin-Madison, Madison, WI, United States

The international Phase III MISTIE Trial, which used CT-image guidance to position a catheter at the site of intracerebral hemorrhage (ICH) failed to meet its clinical endpoint for all patients. The Trial concluded that improved guidance was needed to help a wider set of neurosurgeons. MR-guidance of ICH evacuation could meet these needs. We present an MR-guidance prototype to provide neurosurgeons with a real-time interface with similar viewpoints as commonly used in stereotactic operating room settings. The MR-guidance prototype shows potential to be a reliable tool to assist as monitoring and guiding tools and lysing drugs during clot-evacuation procedures.

Feifei Zhang¹, Yuncai Ran¹, Shujian Li¹, Jinxia Zhu², Xuemei Gao¹, Jingliang Cheng¹, and Chengcheng Zhu^3
1The first Affiliated Hospital of Zhengzhou University, Zhengzhou, China, Zhengzhou, China, ²MR Collaboration, Siemens Healthcare Ltd., Beijing, China, Beijing, China, ³Department of Radiology, University of Washington, Seattle, Seattle, WA, United States

This study investigated the feasibility of PETRA-MRA to evaluate stent angioplasty of middle-cerebral-artery stenosis before and after treatment. PETRA-MRA was shown to significantly reduce susceptibility artifacts induced by metallic devices, decrease the phase dispersion of labeled flow, and increase the signal intensity within the stent. Compared to TOF-MRA, PETRA-MRA has better consistency with DSA in image quality and stenosis measurement, especially for follow-up assessments after stent angioplasty.

Ralf Vogel^1,2, Dieter Ritter², Jonathan Weine^2,3, Jonas Faust^2,4, Elodie Breton⁵, Julien Garnon^5,6, Afshin Gangi^5,6, Andreas Maier¹, and Florian Maier^2
1Pattern Recognition Lab, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany, ²Siemens Healthcare, Erlangen, Germany, ³TU Dortmund, Dortmund, Germany, ⁴Universität Heidelberg, Heidelberg, Germany, ⁵ICube UMR7357, University of Strasbourg, CNRS, FMTS, Strasbourg, France, ⁶Imagerie Interventionnelle, Hôpitaux Universitaires de Strasbourg, Strasbourg, France

Recently, Deep Learning-based methods were used to track the position and orientation of needles in MR images in real-time. Synthetic training data can be generated in large amounts, without data privacy restrictions, and without the need of animal experiments. Therefore, we have simulated the image acquisition using virtual human phantoms containing randomly placed metallic needles in a Bloch simulator. The synthetic images were used to train a U-net to predict the position and orientation of the needle within the susceptibility artifacts of clinical images in less than $$$90\,\text{ms}$$$.

Florian Wiesinger¹, Sandeep Kaushik¹, Mathias Engström², Mika Vogel¹, Graeme McKinnon³, Maelene Lohezic¹, Vanda Czipczer⁴, Bernadett Kolozsvári⁴, Borbála Deák-Karancsi⁴, Renáta Czabány⁴, Bence Gyalai⁴, Dorottya Hajnal⁴, Zsófia Karancsi⁴, Steven F. Petit⁵, Juan A. Hernandez Tamames⁵, Marta E. Capala⁵, Gerda M. Verduijn⁵, Jean-Paul Kleijnen⁵, Hazel Mccallum⁶, Ross Maxwell⁶, Jonathan J. Wyatt⁶, Rachel Pearson⁶, Katalin Hideghéty⁷, Emőke Borzasi⁷, Zsófia Együd⁷, Renáta Kószó⁷, Viktor Paczona⁷, Zoltán Végváry⁷, Suryanarayanan Kaushik³, Xinzeng Wang³, Cristina Cozzini¹, and László Ruskó^4
1GE Healthcare, Munich, Germany, ²GE Healthcare, Stockholm, Sweden, ³GE Healthcare, Waukesha, WI, United States, ⁴GE Healthcare, Budapest, Hungary, ⁵Erasmus MC, Rotterdam, Netherlands, ⁶Newcastle University, Newcastle, United Kingdom, ⁷University of Szeged, Szeged, Hungary

MR imaging offers unique advantages for Radiation Therapy Planning (RTP) via excellent soft-tissue contrast for the delineation of the tumor target volume and surrounding organs-at-risk (OARs).  Remaining challenges include absent CT information (required for accurate dose calculation) and time-consuming manual tumor and OAR contouring.  Here we describe the application of Deep Learning for MR-only RTP in terms of synthetic CT conversion and automated OAR delineation.  Exemplary results are illustrated from an ongoing MR-only RTP study in head&neck and pelvis. 

Florian Friedrich^1,2, Juliane Hörner-Rieber³, Peter Bachert^1,2, Mark E. Ladd^1,2,4, and Benjamin R. Knowles^1
1Medical Physics in Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany, ²Department of Physics and Astronomy, Heidelberg University, Heidelberg, Germany, ³Department of Radiation Oncology, University Hospital of Heidelberg, Heidelberg, Germany, ⁴Faculty of Medicine, Heidelberg University, Heidelberg, Germany

MR-linac systems allow for real-time tumor position updates. Higher temporal resolution imaging through k-space undersampling allows for an increased number of position updates, however iterative reconstructions may negate the decrease in acquisition time and undersampling artifacts may impact tracking stability.

In this study a fast method to denoise and suppress image artifact using a U-net is presented. Undersampled Cartesian and radial cine images were acquired from a patient with a liver tumor on an MR-linac. Tumor tracking stability was assessed. Denoising was found to improve tracking stability and has potential in high temporal resolution cine imaging on MR-linac systems.

Jessica E Scholey¹, Abhejit Rajagopal², Elena Grace Vasquez³, Atchar Sudhyadhom⁴, and Peder Eric Zufall Larson^2
1Department of Radiation Oncology, University of California, San Francisco, San Francisco, CA, United States, ²Department of Radiology, University of California, San Francisco, San Francisco, CA, United States, ³Department of Physics, University of California, Berkeley, Berkeley, CA, United States, ⁴Department of Radiation Oncology, Harvard Medical School, Boston, MA, United States

We implemented a novel approach of using MRI to synthesize CT datasets acquired at MV photon energies for more accurate electron density mapping in radiotherapy treatment planning. We used a 3D deep convolutional neural network and demonstrated clinical proof-of-concept by evaluating the dosimetric impact of using synthetic datasets in a test radiotherapy treatment plan. The proposed method produced mean MAE of 72.8±17.3 HU and SSIM of 0.82 in the test dataset. The dose distributions computed on the test case produced 100% gamma passing rate (computed at 3%/3mm) indicating that synthetic MV images may be used for clinical treatment planning.   

Junzhou Chen^1,2, Pei Han^1,2, Fei Han³, Zhehao Hu^1,2, Nan Wang^1,2, Wensha Yang⁴, Anthony G Christodoulou^1,2, Debiao Li^1,2, and Zhaoyang Fan^1,2,5
1Biomedical Imaging Research Institute, Cedars-Sinai Medical Center, Los Angeles, CA, United States, ²Department of Bioengineering, University of California, Los Angeles, Los Angeles, CA, United States, ³Siemens Medical Solutions USA, Inc., Los Angeles, CA, United States, ⁴Department of Radiation Oncology, University of Southern California, Los Angeles, CA, United States, ⁵Department of Radiology, University of Southern California, Los Angeles, CA, United States

MRI can provide superior soft tissue contrasts for radiation therapy planning. However, radiation planning in the abdomen is especially difficult because respiratory motion can cause misregistration across separate scans with different contrast weightings due to inconsistent breath holds or poor patient compliance. While some studies have produced volumetric and motion-resolved images, they are all limited to a single contrast which is suboptimal for radiation planning. To address these issues, we present a free-breathing MR imaging platform that produces multi-contrast and motion-resolved volumetric images using MR-multitasking, dedicated for radiation therapy planning, and under a 9:30 mins scan time.

Evangelia Kaza¹, Jeffrey P Guenette², Christian V Guthier¹, Steven Hatch³, Alexander Marques³, Lisa Singer¹, and Jonathan D Schoenfeld^1
1Radiation Oncology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, United States, ²Division of Neuroradiology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, United States, ³Radiation Oncology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute, Boston, MA, United States

Quality characteristics of healthy volunteer head images (SNR, CNR, artifact size) were compared between two flexible coil arrangements encompassing radiotherapy immobilization masks relative to a gold standard diagnostic Head/Neck20 coil that cannot accommodate the masks. The novel arrangement of two UltraFlexLarge18 coils provided higher SNR ratios and was more spacious than the commercially recommended arrangement of two FlexLarge4 coils. Artifact size and CNR ratios were similar for the two coil setups. Clinical application of the UltraFlexLarge18 coil arrangement would be advantageous for head and neck radiotherapy MRI simulations due to higher SNR and increased patient comfort.

Austen Curcuru¹, Deshan Yang^1,2, and Michael Gach^1,2,3
1Biomedical Engineering, Washinton University in Saint Louis, Saint Louis, MO, United States, ²Radiation Oncology, Washinton University in Saint Louis, Saint Louis, MO, United States, ³Radiology, Washinton University in Saint Louis, Saint Louis, MO, United States

MRI linear accelerator hybrid systems use balanced steady state free precession sequences (bSSFP) during radiotherapy due to the sequence’s high signal to noise ratio and rapid acquisition times. However, bSSFP sequences are highly sensitive to off-resonance effects.¹ As a result, the position and velocity of the radiotherapy gantry can cause imaging artifacts that can impact tumor tracking and cause fluctuations in the main magnetic field that can result in radiotherapy/MRI isocenter misalignments. Sequence parameters were adjusted during the sequence run time to reduce these issues by integrating a navigator into the bSSFP sequence and measuring B₀ variations in real-time. 

Sihao Chen¹, Cihat Eldeniz¹, Weijie Gan¹, Ulugbek Kamilov¹, Deshan Yang¹, Michael Gach¹, and Hongyu An^1
1Washington University in St. Louis, Saint Louis, MO, United States

Magnetic Resonance Imaging Guided Linear Accelerator (MRI-LINAC) combines an MRI system with a linear accelerator radiotherapy system to treat patients. The MRI-LINAC uses MR to track organ or lesion motion and gates the radiation beam accordingly. In this study, we demonstrate the combination of a T1-weighted self-navigated respiratory motion detection method (CAPTURE) with a deep learning 4D reconstruction (Phase2Phase, P2P) method to derive the 3D deformable respiratory motion field from data acquired on a 0.35 T ViewRay MRI-LINAC system. This initial study demonstrated promising results despite the low SNR at this field strength.

Wajiha Bano^1,2, Will Holmes^1,2, Mohammad Golbabaee³, Alison Tree^1,2, Uwe Oelfke^1,2, and Andreas Wetscherek^1,2
1Joint Department of Physics, The Institute of Cancer Research, London, United Kingdom, ²The Royal Marsden NHS Foundation Trust, London, United Kingdom, ³Computer Science Department, The University of Bath, Bath, United Kingdom

Measuring T2* relaxation during the course of MR-guided radiotherapy can characterize tumour hypoxia, which is associated with treatment resistance. T2* mapping with radial trajectories allows for efficient coverage of k-space but is susceptible to errors arising from gradient delays. We propose a method that jointly estimates gradient delays and T2* using model-based reconstruction. Using the numerical phantom and the in-vivo prostate data we demonstrated that the proposed approach performs better for different noise levels for both fully sampled and undersampled datasets. This will allow better integration of T2* mapping for hypoxia imaging into an MR-linac treatment work flow.

Colleen Bailey^1,2, Rachel W Chan¹, Jay Detsky^3,4, Danny Vesprini^3,4, and Angus Z Lau^1,2
1Odette Cancer Centre, Sunnybrook Research Institute, Toronto, ON, Canada, ²Medical Biophysics, University of Toronto, Toronto, ON, Canada, ³Department of Radiation Oncology, Sunnybrook Health Sciences Centre, Toronto, ON, Canada, ⁴Department of Radiation Oncology, University of Toronto, Toronto, ON, Canada

Eight prostate cancer patients undergoing radiation treatment were scanned on an MR-Linac with conventional ADC and high-b diffusion protocols at each of five treatment fractions. With the system’s weaker gradients and longer echo times, 6/8 patients had visible lesions at the maximum b-value of 2000 s/mm². Fits to the VERDICT model demonstrated higher intracellular fraction in the lesion, inversely correlated with ADC. In three patients, ADC increased starting at the third treatment time point; a fourth patient exhibited a transient ADC increase. Future work will correlate these changes with biochemical recurrence to test their relevance as a biomarker.

Myriam Jaarsma-Coes^1,2, Teresa Ferreira¹, Marina Marinkovic², Khanh Vu², Gre Luyten², Coen Rasch³, Berit Verbist¹, and Jan-Willem Beenakker^1,2
1Radiology, Leiden University Medical Center, Leiden, Netherlands, ²Ophthalmology, Leiden University Medical Center, Leiden, Netherlands, ³Radiotherapy, Leiden University Medical Center, Leiden, Netherlands

For proton beam therapy (PBRT) of Uveal Melanoma, tantalum clips are surgically sutured on the outside of the eye to mark the tumour edge. To aid in the 3D clip localisation and  modelling of the tumor, a dedicated MRI protocol was developed and evaluated. In 51% of the clips, the MRI and per-operative measurements differed less than 1.0mm. In 28% of the clips the discrepancy between the MRI and peroperative measurement was attributed to the complex tumor geometry or anterior tumor localisation. For the majority of the patients MRI added valuable information to the peroperative measurements, improving PBRT planning.  

Pan Su^1,2, Sijia Guo^2,3, Steven Roys^2,3, Florian Maier⁴, Thomas Benkert⁴, Himanshu Bhat¹, Elias R. Melhem², Dheeraj Gandhi², Rao P. Gullapalli^2,3, and Jiachen Zhuo^2,3
1Siemens Medical Solutions USA, Inc., Malvern, PA, United States, ²Department of Diagnostic Radiology and Nuclear Medicine, University of Maryland, School of Medicine, Baltimore, MD, United States, ³Center for Metabolic Imaging and Therapeutics (CMIT), University of Maryland Medical Center, Baltimore, MD, United States, ⁴Siemens Healthcare GmbH, Erlangen, Germany

Transcranial MRI-guided focused ultrasound (tcMRgFUS) is a promising technique for treating multiple diseases. It is desirable to simplify the clinical workflow of tcMRgFUS treatment planning. Previously, feasibility of leveraging deep learning to generate synthetic CT skull from ultra-short echo time (UTE) MRI has been demonstrated for tcMRgFUS planning. In this study, 3D V-Net was used for skull estimation, by taking advantage of 3D volumetric images. Furthermore, feasibility of applying pre-trained model in new dataset was studied, demonstrating the possibility of generalization across various sequences/protocols and scanners.

Anna Crawford¹, Mark Lowe¹, Sean Nagel², Daniel Lockwood¹, Emmanuel Obusez¹, Andre Machado², and Stephen Jones^1
1Imaging Institute, Cleveland Clinic Foundation, Cleveland, OH, United States, ²Neurological Institute, Cleveland Clinic Foundation, Cleveland, OH, United States

High Intensity Focused Ultrasound (HIFU) in now entering clinical practice, for example to treat essential tremor (ET) by creating small lesions in the thalamus. Due to small size of treatment lesions, treatment success depends critically on targeting, which is classically done using measurements and landmarks. We explore an alternative method using functional imaging to guide targeting and assess efficacy, specifically using 7T task-related fMRI. We present preliminary data of the patterns of BOLD activation in a set of patients before and after HIFU.

Kain Kyle¹, Jerome Maller², Yael Barnett³, Stephen Tisch³, Benjamin Jonker³, Michael Barnett¹, Arkiev D'Souza¹, and Chenyu Wang^1
1University of Sydney, Sydney, Australia, ²GE Healthcare, Sydney, Australia, ³St Vincent's Hospital Sydney, Sydney, Australia

Investigation of thalamic segmentation tools FreeSurfer and THOMAS for use in MRgFUS planning.

Huiwen Luo^1,2, Michelle K. Sigona^1,2, Li Min Chen^2,3, Charles F. Caskey^2,3, and William A. Grissom^1,2,3
1Biomedical Engineering, Vanderbilt University, Nashville, TN, United States, ²Vanderbilt University Institute of Imaging Science, Nashville, TN, United States, ³Radiology and Radiological Sciences, Vanderbilt University Medical Center, Nashville, TN, United States

Magnetic resonance acoustic radiation force imaging (MR-ARFI) can be used to localize the focal spot in MR-guided focused ultrasound (FUS) by encoding the ultrasound-induced displacements into the phase of an MR image. A two-minute reduced-FOV 3D MR-ARFI scan with a joint image reconstruction method at 3 Tesla is proposed to image and localize the entire focus in FUS neuromodulation with a low FUS duty-cycle of 0.85%. In this work the reduced FOV MR-ARFI pulse sequence is demonstrated and the proposed k-space undersampling with joint sparse difference image reconstruction to achieve a two-minute scan time is validated.