and perfusion maps from volunteers and patients with hepatic tumors. Materials ... and maps from a patient with a cavernous haemangioma. The. T2*W image ...
Breath Hold 3D Perfusion and Permeability Mapping in the Abdomen Using a Novel Ultrafast First-Pass Leakage-Profile Model X.P. Zhu, K.L. Li, J.M. Hawnaur, *J.C. Waterton, Y. Watson, P. Taylor, A. Jackson Division of Imaging Science and Biomedical Engineering, Stopford Medical School, University of Manchester, Manchester, *AstraZeneca, Alderley Park, Macclesfield, Cheshire, UK Introduction An ultra-fast 3D method for simultaneous mapping of endothelial permeability and blood volume has been developed. This method uses the leakage profile of contrast agent (CA) to target tissues during the first pass of CA bolus. Here we describe the mathematical model and implementation of the new technique. We also present 3D abdominal permeability and perfusion maps from volunteers and patients with hepatic tumors. Materials and Methods MR imaging We applied this technique to volunteers and patients with hepatic tumors. MR images were obtained on a 1.5 T Philips ACS NT scanner with maximum gradient strength of 23 mT/m and maximum slew rate of 105mT/m/ms. MR imaging was performed using Tl weighted 3D gradient echo images (TlW-GRE, TE/TR = 1.1/4.3 ms). An array of 3D TlW-GRE images were firstly acquired using different flip angles to allow calculation of intrinsic 3D longitudinal relaxation rate maps (RlO). The TlW-GRE sequence with the highest flip angle was repeated to produce a dynamic data set immediately after a intravenous bolus injection of 0.1 mmol/kg of gadodiamide with breath holding for a period of 30 seconds. Modeling lesion leakage urofile Assuming the gadodiamide concentration, [Gd], in plasma Cp >> [Gd] in the leakage space (CZ), during the first pass of CA. The passage of gadodiamide Erom plasma into the leakage space vl can be simplified as dC -.....L= dt
k 2-c Y,
If we consider the increase of [Gd] in the leakage space as the sole contributor to the MRI signal of tumor in a relaxivity dependent dynamic contrast study, the [Gd] concentration of tumor, C, = v[C, and = k b 1 C ,(t)dt’
(3)
Eq. 3 describes the0relation between the permeability kfi, CA concentration measured from target tissues Clt), and the sum of accumulated C,(f) in the vascular compartment. We have called this relationship the leakage profile of the tissues. Manning nermeabilitv and blood volume We initially compute 3D RlO maps and 4D concentration (C(t)) maps (x, y. z, t) [2]. Mean C,(t) curves (arterial input function) were obtained for each patient from voxels in the coeliac axis. The C,(t) data were fitted using a gamma function concatenated with a steady-state concentration. The cumulative sum of C,(t) in Eq. 2 was then calculated to provide the leakage profile in tissues. Calculation of the permeability kfi for each voxel was accomplished in two stages: 1) kfi in tumor voxels with negligible first pass peaks was calculated by fitting the leakage profile to the corresponding tissue; 2) For kb in tumor voxels which have significant first pass peaks, the fit was corrected to remove the effect of residual CA. Finally, 3D leakage-free rBV maps, denoted as rBV,,,ted (rBV, in brief), were calculated, using the Eq.4. 3D high spatial resolution rBV, and kb maps with a matrix size of 2.56~256~25 were generated using an in-
Proc. Intl. Sot. Mag. Reson. Med. 8 (2000)
Figure 1 shows a sagittal slice from 3D maps of rBV, (la) and kb (lb) map from a healthy volunteer. Cardiac chambers and large vessels with high blood volume were clearly depicted on the rBV, map. The kb map demonstrates tissues hyperpermeable to CA, such as liver parenchyma. (kb = 0.35 + 0.12 min.‘). Figure 2 shows sagittal sections from 3D images and maps from a patient with a cavernous haemangioma. The T2*W image (2a) shows a hyperintense lobular lesion with distinct border (arrow). The rBV, map (2b) and the kfi map (2~) reveals the hypoperfused tumor surrounded by a vascular rim. The tumor itself is also less permeable to CA than liver parenchyma during the first pass of CA bolus. Again, normal blood vessels, such as pulmonary arteries and vessels in the liver , are pronounced on the rBV, but not on the kfi maps. Renal cortex has higher kh and rBV,, than the liver.
1-1
Fig. 1 a) rBV, map; b) kn map. Mean kh values ( f SD) of liver = 0.35 + 0.12 mid.
’
where kb denotes permeability surface area product during the 1st pass of contrast bolus, VI is the fraction of lesion tissue which the leakage space occupies. Integrating Eq. 2 gives
C,(t)
house IDL-C program.
Fig. 2 a) TZ*W shows
Cc)
a cavem~u~ haemangioma(mow); b) rBV, map; c) km map. Mean kD values(f SD) of kidney = 0.64+0.23 mid.
Discussions and Conclusions This new method of blood volume and permeability mapping based on contrast agent leakage profile was first applied to a group of patients with intra-cranial tumors [l]. Results showed that the new method provided more robust measures of endothelial permeability of brain tumors than the conventional methods based on multi-compartment models. Noteworthy was the speed advantage of this new method, which made it possible to obtain both blood volume and permeability maps of thoracic and abdominal organs when patients were breath holding. 3D abdominal blood volume and permeability maps have been obtained in less than one minute. In liver and kidney, unlike in brain, intra-vascular agents, such as gadodiamide used in this study, firstly distributes within the intra-vascular compartment and then rapidly leaks throughout the interstitial extra-vascular space. The new technique permits separation these two concurrent effects on a pixel by pixel basis. The resolved blood-volume/permeability maps have demonstrated diagnostic quality allowing radiologists to a visualize lesions and to provide quantitative information on structural and functional changes of capillary beds in liver and kidneys. This new technique is fast and robust with a potential to be used in more extensive clinical trials, for assessing tumor more grade, therapeutic effectiveness and/or providing appropriate assessment of turnour vasculature. References 1) Li et al. submitted to ISMRM 2000; 2) Kassner A, Annesley D, Zl& XP et al, Abnormalities of the contrast m-circulation phase in cerebral tumors demonstrated using dynamic suiceptibility contrast-enhanced MR imaging: a possible marker of vascular tortuosity, (in press), JMRI.
