Anthropomorphic phantoms are essential for the evaluation of image registration algorithms in multimodal imaging, quantification experiments in multinuclear MR imaging, and verification of diffusion and flow measurements. A human-like abdominal phantom incorporating a liver with lesions, a rib cage, vessels, and a lung was developed. Its tissue-mimicking characteristics were evaluated with 1H and 23Na MR and CT imaging as well as functional flow and diffusion MR imaging. The phantom exhibited morphological and functional parameters comparable to corresponding human values. It is suitable for a quantitative evaluation of a clinical workflow ranging from diagnostics to interventional procedures.
Image guided interventions are based on real-time imaging. A quantitative evaluation of a clinical workflow, ranging from diagnostics to interventional procedures, is an important step towards clinical implementation. To exploit the benefits of various imaging modalities, the image fusion of data sets with previously acquired images from computed tomography (CT) and magnetic resonance imaging (MRI) is object of recent research. Especially in MRI, this includes morphological and functional imaging providing complementary information about tumor characteristics. Anthropomorphic phantoms can mimic complex anatomy and physiology while providing an essential ground-truth for the verification of quantitative imaging. Image registration algorithms face a variety of challenges, such as different voxel sizes, geometric deformations and changing patient positions between modalities. Here, anthropomorphic phantoms serve as an additional level between geometric phantoms and in vivo studies for the validation of algorithms under practical conditions and enable the design of a feasible imaging protocol for an optimized workflow in terms of time, costs and patients comfort.
The abdominal phantom was created from several modules that were designed as close to human anatomy as possible (Fig. 1). A rib cage consisting of a spinal section with vertebral bodies and ribs surrounds a liver and lung module. It was 3D printed and made of polylactide (PLA). The lung module was also 3D printed using polystyrene (PS), splitting into the left and right lobe and acts as a geometric stabilizer of the liver. Based on a CT scan, a human liver was contoured and a mold was created. Then, silicone mixed with 20% silicone oil was poured into the mold (2.5 liter) and left to cure. Further, three oval inclusions (each 37ml, 3cm diameter) were incorporated and represented tumors with different morphological and functional characteristics. Inclusion A was made of silicone and 5% CaCO3 to induce variations in the Hounsfield units and a coloring agent (green) for visibility during biopsy. Inclusion B was only made of silicone and a coloring agent (blue). Inclusion C was made of 154 mmol NaCl solution for physiological Na+ concentration and 2% agarose acting as a stabilizer and 0.1 ml Dotarem (0.5 mmol/ml) to reduce relaxivity [1].
MR images were acquired with a 3T whole-body scanner (Magnetom Skyra, Siemens Healthineers, Germany) and a double resonant 1H/23Na transmit receive array (Rapid Biomedical GmbH, Germany). T1 and T2 weighted images were acquired (TSE sequence with TE=[8,17,34,42,51]ms, TI=[25,50,100,200,400,600,800]ms, 1.4mm isometric pixel resolution). Sodium images were acquired by using a radial density adapted 3D UTE sequence [2]. Parameters: TE/TR/FA=0.6ms/30ms/54°, 11000 projections, FoV=(350mm)3, resolution=(6mm)3, TRO=20ms, full sampling, TA=11:00min. Reconstructions of the channels were performed using adaptive combination [3]. Diffusion-weighted (EPI sequence with TR=3700ms and TE=50ms, b-value=[50,200,800]) as well as flow imaging (3D-TWIST sequence: TE/TR/FA=1.44ms/15ms/15°, temporal resolution of 2 s, total acquisition time of 100 s) was performed. For flow measurements, water served as a blood substitute and was pumped through an artificial artery. Contrast agent was injected automatically to simulate a bolus. CT images were acquired with a Somatom Force (Siemens Healthineers, Germany) with kVp=110 kV and a smoothing convolution kernel.
This research project is part of the Research Campus M²OLIE and funded by the German Federal Ministry of Education and Research (BMBF) within the Framework “Forschungscampus: public-private partnership for Innovations” under the funding code 13GW0092D.
* shared first authorship
[1] Hattori et al. Development of a MRI phantom equivalent to human tissues for 3.0-T MRI. Medical Physics 2013; 40(3):032303
[2] Nagel AM, et al. Sodium MRI using a density-adapted 3D radial acquisition technique. MRM 2009;62:1565-1573.
[3] Walsh DO, et.al. Adaptive reconstruction of phased array MR imagery. Magn Reson Med. 2000;43(5):682-690.
[4] Kalender, W. A. Computertomographie. Publicis Corporation Publ. 2006