1426

Visualization and quantification of epiphyseal cartilage vasculature using quantitative susceptibility maps of pediatric knee specimens
Kai D. Ludwig1,2, John Strupp1,2, Casey P. Johnson1,2, Stefan Zbyn1,2, Mikko J. Nissi1,2,3, Ferenc Tóth4, Kevin Shea5, Cathy S. Carlson6, and Jutta M. Ellermann1,2

1Center for Magnetic Resonance Research (CMRR), University of Minnesota, Minneapolis, MN, United States, 2Radiology, University of Minnesota, Minneapolis, MN, United States, 3Applied Physics, University of Eastern Finland, Kuopio, Finland, 4Veterinary Population Medicine, University of Minnesota, St. Paul, MN, United States, 5Orthopaedic Surgery, Stanford University, Stanford, CA, United States, 6Veterinary Clinical Sciences, University of Minnesota, St. Paul, MN, United States

Synopsis

Visualization of the vasculature within the epiphyseal cartilage of the distal femur of the skeletally immature knee joint is possible with susceptibility-weighted MRI. A post-processing pipeline is described to segment and quantify the vascular density within two distinct vascular beds of the distal femoral epiphyseal cartilage using quantitative susceptibility maps. The described post-processing may allow identification of vascular abnormalities at early stages of development and also may improve assessment of therapeutic interventions.

Purpose

In children, growth and development of the knee joint involves gradual replacement of the highly vascular sub-articular epiphyseal cartilage with bone through a process called enchondral ossification. Interruption of the vascular supply of the epiphyseal cartilage is known to cause failure of the enchondral ossification that may lead to juvenile osteochondritis dissecans (JOCD) or other developmental diseases. Imaging the vasculature of the epiphyseal cartilage is important to better understand the pathogenesis of JOCD and to evaluate potential treatments. Susceptibility-weighted imaging (SWI) with quantitative susceptibility mapping (QSM) post-processing is a recently-developed technique that allows visualization of vessels within the epiphyseal cartilage utilizing tissue-inherent contrast [1-3]. However, a need remains for methods to quantify these vessels to identify abnormalities. In the distal femur, where the medial condyle is the primary predilection site for JOCD, there is a need to better characterize the temporal changes to three distinct epiphyseal cartilage vascular beds as the femur develops: two peripheral beds and one central bed. The purpose of this work was to develop an image-processing pipeline from QSM images to better visualize the central and peripheral vascular beds within the epiphyseal cartilage of the distal femur and to determine the vascular density within these beds.

Methods

Specimens. A juvenile human cadaveric distal femur, obtained from a 3-month old female donor (Allosource), was immersed in perfluoropolyether for clean and susceptibility-matched background during imaging. Imaging. MRI was performed at 9.4 T (Agilent Technologies) with a quadrature volume RF coil (Millipede, Varian NMR Systems). Quantitative susceptibility mapping (QSM) datasets were acquired using a spoiled 3-D gradient-recalled echo sequence (TR/TE = 40/14 ms, matrix = 3843, resolution = 100 µm isotropic, bandwidth = 16 kHz, flip angle = 15°, and scan time = ~98 min). Analysis. MRI data was post-processed in Matlab (Mathworks) as previously described [1,2] to obtain QSM images. A Hessian-based Frangi filter [4,5] converted QSM images to maps of vessel likelihood, which were then binarized. The binary vessel masks were skeletonized using parallel medial axis thinning [6,7], and a 3-D network graph was calculated [6]. The two peripheral and single central vascular beds were manually segmented from the network graph and the vessel area density (VAD) and vessel skeleton density (VSD) were calculated [8]. VAD is the ratio of the imaging voxels with vasculature to total imaging volume in the binary vessel map while VSD is the ratio of the imaging voxels with skeletonized vasculature to the total imaging volume in the skeletonized vessel map.

Results

Vascular supply to the epiphyseal cartilage of the distal femur can be visualized using QSM post-processing (Figure 1). The vascular architecture becomes evident after skeletonization of the vascular data: two peripheral and one central vascular beds of the knee are clearly outlined and demonstrate the presence of a watershed zone that lacks anastomoses between the peripheral and central vascular beds (Figure 2). Further post-processing of the QSM images, with the step-wise pipeline shown in Figure 3, allows mapping of the skeletonized vascular network within the distal femur along with 3-D segmentation and representation of the vascular beds (Figure 4). Quantitative vascular density values can then be computed for each vascular bed (Table 1). The results demonstrate that the peripheral vascular beds had greater vessel area density (VAD = 9.7×103 a.u.) and vessel skeleton density (VSD = 1.5×103 a.u.) than the central vascular bed (VAD = 12.7×103 a.u.; VSD = 2.0×103 a.u.).

Discussion

We have demonstrated a post-processing pipeline to visually and quantitatively analyze QSM images of epiphyseal cartilage vessels in the distal femur of a skeletally immature knee joint. Using freely available tools, our pipeline enables high-quality delineation of the vascular beds and semi-automatic segmentation of the 3-D volume encompassing the vascular networks within these beds. This is a powerful tool to demonstrate the existence of distinct vascular networks without anastomoses, quantify their densities, and follow their development and relationship over time during skeletal maturation. This tool will be useful in the study of JOCD, as it is hypothesized that reduced vascularity within the epiphyseal cartilage predisposes specific anatomical regions (i.e., predilection sites) to development of the disease. The study of other diseases that involve the vasculature of the epiphyseal cartilage, such as Legg-Calve-Perthes disease [9,10], may also benefit from this method.

Acknowledgements

This study was supported by the NIH (R01AR070020, K01AR070894, K01OD021293, and P41EB015894), the Academy of Finland (#285909 and #293970) and the W. M. Keck Foundation.

References

1. Nissi MJ, Tóth F, Zhang J, Schmitter S, Benson M, Carlson CS, Ellermann JM. Susceptibility Weighted Imaging of Cartilage Canals in Porcine Epiphyseal Growth Cartilage Ex Vivo and In Vivo. Magn Reson Med. 2014;71:2197-2205

2. Nissi MJ, Tóth F, Wang L, Carlson CS, Ellermann JM. Improved Visualization of Cartilage Canals Using Quantitative Susceptibility Mapping. PLos ONE. 2015;10(7):e0132167

3. Tóth F, Nissi MJ, Ellermann JM, Wang L, Shea KG, Polousky J, Carlson CS. Novel Application of Magnetic Resonance Imaging Demonstrates Characteristic Differences in Vasculature at Predilection Sites of Osteochondritis Dissecans. Am J Sports Med. 2015;43(10):2522-7

4. Kroon D-J. Hessian based Frangi Vesselness Filter. 2009. Matlab implementation:

https://www.mathworks.com/matlabcentral/fileexchange/24409-hessian-based-frangi-vesselness-filter

5. Frangi A, Niessen W, Vincken K, Viergever M. Multiscale vessel enhancement filtering. Medical Image Computing and Computer-Assisted Interventation-MICCAI. 1998;98:130–137

6. Kerschnitzki M, Kollmannsberger P, Burghammer M, Duda GN, Weinkamer R, Wagemaier W, Fratzi P. Architecture of the osteocyte network correlates with bone material quality. Journal of Bone and Mineral Research. 2013;28(8):1837-1845. Matlab implementations:

https://www.mathworks.com/matlabcentral/fileexchange/43400-skeleton3d

https://www.mathworks.com/matlabcentral/fileexchange/43527-skel2graph-3d

7. Lee TC, Kashyap RL, Chu CN. Building skeleton models via 3-D medial surface/axis thinning algorithms. Computer Vision, Graphics, and Image Processing. 1994;56(6):462-478

8. Chu Z, Lin J, Gao C, Xin C, Zhang Q, Chen C-L, Roisman L, Gregori G, Rosenfeld PJ, Wang RK Quantitative assessment of the retinal microvasculature using optical coherence tomography angiography. J Biomed Opt. 2016;21(6):066008

9. Kim HK, Bian H, Aya-ay J, Garces A, Morgan EF, Gilbert SR. Hypoxia and HIF-1alpha expression in the epiphyseal cartilage following ischemic injury to the immature femoral head. Bone. 2009;45(2):280-8

10. Johnson CP, Wang L, Toth F, Aruwajoye O, Carlson CS, Kim HK, Ellermann JM. Quantitative Susceptibility Mapping Detects Neovascularization of the Epiphyseal Cartilage After Ischemic Injury in a Piglet Model of Legg-Calvé-Perthes. J Magn Reson Imaging (in press)

Figures

Figure 1. A representative slice of the input 3D GRE MRI data acquired on the skeletally immature distal femoral condyle of the cadaveric knee joint at 3 months of age. (A) Magnitude MR anatomical image, (B) quantitative susceptibility map (QSM), and (C) overlay of the QSM (red) and magnitude image (gray scale) highlights the location of vasculature within the sub-articular epiphyseal cartilage of the developing distal femur.

Figure 2. Maximum intensity projections (MIPs) of the 3D susceptibility map at two different projection angles (α) demonstrating the watershed zone (arrows) of the vasculature network separating the two peripheral and single, central vascular beds within the epiphyseal cartilage.

Figure 3. The step-wise image processing pipeline used to convert the susceptibility maps into the 3-D vessel network graph for calculation of vessel density in specific anatomical vascular beds of the epiphyseal cartilage shown as either a maximum intensity projection (MIP; top row) or as a single, central slice (bottom row).

Figure 4. A 3-D-volume rendering of the segmented central vascular bed (blue volume) and two peripheral vascular beds (red [lateral knee] and green [medial knee] volumes) with a projection of the entire segmented knee joint below the 3-D-rendering. The watershed architecture devoid of any anastomoses is evident after completion of the skeletonization.

Table 1. Vascular density values for the central and peripheral vascular beds within the epiphyseal cartilage of the skeletally immature distal femoral condyle (3 months of age).

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
1426