Does reduced FOV Diffusion Weighted Imaging inherently yield lower ADC ?
Suchandrima Banerjee1, David Aramburu-Nunez2, Ramesh Paudyal2, Thomas Chenevert3, Michael Boss4, and Amita Shukla-Dave2,5

1Global MR Applications & Workflow, GE Healthcare, Menlo Park, CA, United States, 2Department of Medical Physics, Memorial Sloan-Kettering Cancer Center, New York, NY, United States, 3Department of Radiology, University of Michigan Health System, Ann Arbor, MI, United States, 4Applied Physics Division, National Institute of Standards and Technology, Boulder, CO, United States, 5Department of Radiology, Memorial Sloan-Kettering Cancer Center, New York, NY, United States


The benefits of reduced field-of-view (rFOV) imaging with the single-shot echoplanar diffusion sequence such as lower distortion and better discrimination of tumor from benign tissue have been demonstrated in several anatomies. In most of these published works, lower ADC was reported using rFOV compared to the standard full FOV (fFOV) method, irrespective of the technique by which rFOV was achieved. In this work we conducted controlled experiments in 3 phantoms to avoid some of the confounding factors present in vivo and investigated if there is a systemic underestimation of ADC in rFOV DWI compared to fFOV DWI.

Target Audience:

MR physicists and clinicians interested in diffusion and quantitative imaging


The benefit of reduced field-of-view (rFOV) imaging with the single-shot echoplanar diffusion sequence with regard to lower distortion and better discrimination of tumor from benign tissue has been demonstrated in several anatomies [1-5]. In most of these publications, lower Apparent Diffusion Coefficient (ADC) was also reported using rFOV compared to the standard full FOV (fFOV) method, irrespective of the technique by which rFOV was achieved [6-8]. In cases where rFOV and fFOV were acquired at the same resolution, distortion was lower in the former, and could have contributed to measurement differences especially at tissue boundaries. In cases where rFOV was acquired at a higher spatial resolution, less partial voluming in heterogeneous tissue environment could have contributed to lower/ “truer representation” of ADC in rFOV images. On the other hand, lower SNR due to a smaller excitation volume could also have caused ADC underestimation [9]. A thorough quantitative study is needed to establish one to one correspondence in diffusion measurements between the two approaches that would going forward allow, for example, inclusion of both rFOV and fFOV data in longitudinal studies. As a first step we conducted controlled rFOV vs. fFOV experiments at 2 field strengths and three phantoms to investigate if rFOV scans inherently yield lower ADC values.


A Sphere Phantom (GE, Waukesha, WI) doped with CuSO4 at room temperature and the new National Institute of Standards and Technology (NIST) and RSNA-QIBA ice-water diffusion phantom were used for 3T experiments. The ice-water phantom is constructed of varying concentrations of polyvinylpyrrolidone (PVP) in aqueous solution to generate physiologically relevant ADC values at 0 °C; the vials are arranged in different positions (c=central; o=outer; i=inner) to sample any spatial dependence of ADC [10]. Scans were acquired on a GE MR 3.0T system (Discovery MR750, Waukesha, WI) with a 12 channel receive array using the standard fFOV EPI DWI and FOCUS, where a 2D spatially selective excitation is used to limit the phase FOV (pfov) extent [3]. FOCUS scans were acquired with pfov factors of 1 and 0.5. A FOCUS pfov = 0.5 scan was additionally acquired with 2 averages, to have “equivalent” SNR as fFOV scans. All other scan parameters were kept identical between acquisitions (FOV=24cm, Slice thickness=5mm/5mmgap, in-plane: 1.875x1.875 mm2, TR/TE=8000/63 ms, b-value =0 and 1000 s/mm2). Scans were also acquired in a 4 cylinder phantom (GE, Waukesha,WI) constructed of cylinders having relaxation properties similar to cerebrospinal fluid (CSF) (agarose gel doped with .08 mM CuS04), grey matter (GM) (.9 mM NiCl2) , white matter (WM) (2 mM NiCl2) and skull fat (vegetable oil) on a GE 1.5T system (Discovery MR450W, Waukesha, WI) using a 12 channel receive array. These acquisitions also consisted of standard fFOV EPI and FOCUS sequences (FOV=18cm, Slice thickness=4mm/1mm gap, in-plane resolution=1.4x1.4 mm2, TR/TE=4000/66 ms, b value= 0 and 1000 s/mm2) with 1 average for all except the last FOCUS scan, similar to 3T.


Measurements were in close agreement between all the acquisitions in the GE sphere (Figure 1) and the ice-water phantom (Figure 2) at 3T. In the ice-water phantom, distortion was expectedly much lower in the FOCUS pfov=0.5 scans due to shorter readout, compared to fFOV as seen in Figure 2. Regions of interest for ADC analysis were placed away from vial boundaries as much as possible, to minimize effects of distortion on the measurements. In the 4 cylinder phantom, there was small underestimation in ADC measurements obtained from FOCUS pfov=0.5 images compared to fFOV, which was virtually eliminated in the “SNR matched” FOCUS pfov=0.5 scan (Figure 3).


Unlike in 3T experiments, FOCUS pfov=0.5 images were in the low SNR regime at 1.5T, such that noise bias created an underestimation in ADC values that was removed in the “SNR matched” FOCUS pfov=0.5 scan. This could also explain why the biggest measurement difference was in WM, which had the lowest SNR of the CSF, GM and WM cylinders, due to its shortest T2 relaxation time.


Our initial observations indicate that rFOV diffusion methods do not inherently yield lower ADC values. But users wishing to combine and compare metrics obtained from rFOV and fFOV acquisitions should give careful consideration to choice of scan parameters such as number of averages and optimal b value. For a more complete investigation, experiments would have to be conducted on phantoms with diffusion properties, other factors such as higher spatial resolution would have to be explored, and a wide range of scalar and tensor diffusion metrics would have to be analyzed.


No acknowledgement found.


1.WheelerKingshott C et al, Neuroimage 2002;16:93-102. 2. Wilm B et al, MRM 2007;57:625-30 3.Saritas E et al, MRM 2008; 60:468-7 4 Zaharchuk G et al, AJNR 2011;32:813-20 5.Singer L et al, Acad Radiol.6. Lu Y et al, J Comput Assist Tomogr. 2015 May-Jun;39(3):334-9 7.Park J et al, ASNR 2015 8. Poplawski MM et al, RSNA 2014#NRS419 9. Dietrich O et al, MRM 2001, 45:448-453 10. Boss M, Chenevert T, Attariwala R, Shukla-Dave A, Jackson E, Amaro E, accepted to RSNA 2015


Figure 1: ADC measurements in the sphere phantom

Figure 2: Images and ADC measurements from the ice-water phantom

Figure 3: Differences in ADC measurements between FOCUS and standard fFOV EPI DWI in the 4 cylinder phantom

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