An Indenting Abdominal Array for 2-Fold SNR Improvement in Pancreatic MRI
Scott B. King1, Jarod Matwiy1, Calvin Bewsky1, Hung-Yu Lin1, and Masoom A. Haider2,3

1Medical Devices, National Research Council Canada, Winnipeg, MB, Canada, 2Dept of Medical Imaging, Sunnybrook Health Sciences Center, Toronto, ON, Canada, 3Department of Medical Imaging, University of Toronto, Toronto, ON, Canada


The pancreas is difficult to image because of its central location deep in the abdomen, often with overlapping artifact in parallel MRI reconstruction through the mid abdomen and drop off of signal. In this new concept for improved SNR and parallel MRI in pancreas MRI, a surface array is pushed/indented into the abdomen, bringing smaller array elements closer to the pancreas. Compared to the benchmark array, the Indenting Array demonstrated >2x SNR and 40% improved R=3 A-P parallel MRI with g-factor = 1 within the pancreas. This new “indenting” array design could have a significant impact on pancreas diagnostic MRI.


The pancreas can be a difficult organ to image because of its central location deep in the abdomen. There is frequently overlapping artifact in parallel imaging reconstruction through the mid abdomen and drop off of signal. Most array coils used for body imaging contain array elements that are 12-15 cm in diameter, evenly distributed to cover the body [1,2]. Using smaller array elements (6cm-8cm in diameter) could increase SNR near the periphery of the body, but would not increase SNR deep inside the abdomen. We report on a new concept for improved SNR and parallel MRI in pancreas MRI, whereby a surface array of smaller elements are pushed/indented into the abdomen, thereby bringing the smaller array elements (6cm-8cm in diameter) closer to the deep lying pancreas, with the added benefit of displacing the transverse colon and stomach away from the pancreas.


The anterior array design philosophy was to use two array sections, a smaller “indenting” surface array and a larger outer surface array for extended FOV coverage. The Indenting Array was shaped to contour the abdomen and sit inferior to the rib cage, containing 5-elements approximately 6-8 cm in diameter with an in-plane size of 22.1cm x 11.9cm (Fig.1b). The Outer Array contained 8-elements with dimensions ranging from 12-18 cm with an in-plane size of 48cm x 32cm (Fig. 1a). A system of “indenting” is shown in Figure 1c. Overlapping and preamplifier decoupling minimized coupling between array elements. Each of the 13 anterior elements tuned to 123.2 MHz included 1-active trap, 1-passive trap, 1-fuse, and a 1.4 Ω input impedance preamp (Hi-Q.A., Carleton Place, ON, Canada). The 13-Channel Pancreas Array was combined with a 9-Channel OEM Spine Matrix for posterior coverage. The 13-Channel Pancreas Array was safety validated and interfaced to a Siemens 3T 32-Channel TIM Trio MRI. The OEM 6-Channel Body Matrix combined with the same 9-Channel Spine Matrix was used for benchmark comparisons. Imaging experiments were carried out on healthy subjects following an approved REB protocol. The OEM “sn” sequence was used for SNR/g-factor evaluation, and 3D VIBE for clinical evaluation. SNR/g-factor analysis was performed using Musaik (Schmid and Partner Engineering AG, Zurich).


The pancreas body was initially at a depth of 95mm from the surface and OEM array, but with the indenting array system, the pancreas body was only 54mm below the smaller elements of the indenting array. Compared to the benchmark array, the 13-Ch Indenting Array demonstrated >2x SNR improvement (Fig.2 a,c) and 40% improved R=3 A-P parallel MRI with a g-factor near 1 (Fig.2 b,d) within the pancreas. This >2x SNR gain was leveraged for much better quality high resolution (704x264 matrix) VIBE imaging (Fig. 3a) and R=3 A-P parallel MRI (Fig.3c) compared to the benchmark array. Zoomed 3D VIBE images (Fig. 3a,b - bottom) obtained on the 3T Trio system at a FOV of 40x30cm, 5mm slice thickness, 18s acquisition time for 24 slices with a matrix of 704x264, have in-plane resolution of 0.57mm x 1.13mm. The image with the Indenting Array (Fig.3a) shows a sharp image through the inferior pancreatic head without noise or speckle artifact with clear definition of the lobulated architecture of the pancreatic head. The image with the manufacturers array (Fig.3b) shows noise and speckle artifact through the pancreas limiting detailed evaluation. When parallel MRI is pushed to R=3 A-P, images at 512x250 with in-plane resolution 0.78mm x 1.2mm and 12sec acquisition time with the Indenting Array (Fig.3c) are free of artifact seen in the corresponding images with the manufacturers array (Fig.3d).


This new “indenting” array design has demonstrated 2-fold SNR improvement in pancreatic MRI, and could have a significant impact on pancreas diagnostic MRI. This has the potential to improve detection of small pancreatic cancers, small distal common bile duct stones and periampullary tumors.


This research was funded by the Ontario Institute of Cancer Research. The authors thank CIMTEC for their support.


[1]Zhu et al. MRM 52:869-877 (2004). [2]Hardy et al. MRM 55:1142-1149 (2006).


Layout of (a) Outer Array, (b) Indenting Array, and (c) indenting mechanism

SNR and R=3 A-P g-factor maps for (a,b) 13-Ch Indenting Array and (c,d) 6-Ch OEM Array. Arrows indicate pancreas head and body.

Axial 5mm slice of a 24 slice 3D VIBE (TR/TE=5ms/1.67ms, FOV=40cmx30cm) for: (a,b) R=2 A-P ACQ time = 18s, matrix=704x264 and in-plane resolution 0.57mm x 1.13mm; (c,d) R=3 A-P ACQ time = 12s, matrix=512x250, in-plane resolution 0.78mm x 1.2mm; (a,c)13-Ch Indenting Array, (b,d) 6-Ch OEM Array. (Bottom) Zoomed over pancreatic head indicated with arrows.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)