Method and apparatus for measuring 3D geometric distortion in MRI and CT images with a 3D physical phantom

Abstract

3D printing in MRI-compatible plastic resin has been used to fabricate and implement a geometric distortion phantom for MRI and CT imaging. The sparse grid structure provides a rigid and accurate phantom with identifiable intersections that are larger than the supporting members, which produces images that are amenable to fully automated quantitative analysis using morphometric erosion, greyscale segmentation and centroiding. This approach produces a 3D vector map of geometric distortion that is useful in clinical applications where geometric accuracy is important, either in routine quality assurance or as a component of distortion correction utilities.

Claims

1. A method for measuring geometric distortions of a 3D medical imaging system, the method comprising: providing a 3D printed physical phantom comprising a plurality of control points, each having a pre-determined location; obtaining a 3D image of the 3D printed physical phantom using either magnetic resonance imaging (MRI) or computed tomography (CT); identifying the control points in the image by segmentation and morphological erosion; determining the location of the control points in the image; comparing the location of the control points in the image with the pre-determined location of the control points in the 3D printed physical phantom; and, deriving a spatial vector for each control point that quantifies the geometric discrepancy between the control points in the image and the pre-determined location of the control points in the 3D printed physical phantom.

2. The method according to claim 1, wherein the location of the control points in the image is obtained by segmenting the boundary of the 3D image of the 3D printed physical phantom using at least grey-scale threshold values, and subsequently performing morphological erosion of the boundary by removing a specified number of boundary surface elements.

3. The method according to claim 2, wherein the morphological erosion is performed until structure of the 3D printed physical phantom connecting the control points is removed, leaving isolated clusters of volume elements at known locations in the image.

4. The method according to claim 3, wherein the accuracy of the known locations is improved by obtaining a centroid of the clusters of volume elements.

5. The method according to claim 1, wherein identifying control points is performed automatically.

6. The method according to claim 1, wherein the spatial vector comprises 3D vector map.

7. A system for measuring geometric distortion errors in a magnetic resonance imaging (MRI) or computed tomography (CT) medical imaging system, comprising: a three-dimensional calibration printed physical phantom comprising a plurality of control points connected to each other by supporting structure, the three-dimensional calibration printed physical phantom suitable for imaging by the medical imaging system; and, a computer provided with machine executable instructions configured to execute an analysis program that determines the centroids of the control points in three-dimensional space from an image of the three-dimensional calibration printed physical phantom acquired using the medical imaging system, compares the centroids to the true locations of the control points, and calculates a spatial vector that relates each centroid to its true location.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) For clearer understanding, preferred embodiments will now be described in detail by way of example, with reference to the accompanying drawings, in which:

(2) FIG. 1 summarizes the three components of the system and method;

(3) FIG. 1a is flowchart showing the first component of the system and method of FIG. 1;

(4) FIG. 1b is a flowchart showing the second component of the system and method of FIG. 1;

(5) FIG. 1c is a flowchart showing the third component of the system and method of FIG. 1;

(6) FIG. 2 is a computer-aided design (CAD) depiction of an embodiment of the phantom;

(7) FIG. 3 is a photograph of the embodiment of FIG. 2, 3D printed in plastic resin;

(8) FIG. 4 is a photograph of the embodiment of FIG. 2 in a leak-tight container of paramagnetic liquid;

(9) FIG. 5 is an illustration of an exemplary axial signal intensity profile of an MRI image, along with a fitted correction function;

(10) FIG. 6 is a MRI image taken through the central fiducial bead plane of the phantom of FIG. 2;

(11) FIG. 7 is a CT image taken through the central fiducial bead plane of the phantom of FIG. 2;

(12) FIG. 8 is an isosurface of a segmented, binary volume image of the phantom of FIG. 2;

(13) FIG. 9 is an isosurface of the isolated fiducial clusters after morphological erosion of the image of FIG. 8; and,

(14) FIG. 10 is geometric distortion vector map obtained from the image of FIG. 9.

DETAILED DESCRIPTION

(15) Referring to FIG. 1, a system and method is generally shown comprising three main components. Referring to FIG. 1a, in step 1.1 a 3D phantom is created comprising a plurality of control points arranged in space in a pre-determined location. The control points are differentiable from supporting structure on the basis of size. In one embodiment, the control points have a larger size than the supporting structure used to position the control points in their pre-determined locations. This allows the control points to be readily differentiated from the support structure. The control points may comprise spheres having a larger diameter than the elongate members of the support structure, which may be cylindrical in shape. The design of the location of the control points takes into account features of the object desired to be imaged and their magnetic or density characteristics; for example, if an orthopedic implant, such as a metallic prosthesis is desired to be imaged, the phantom may be designed with a cavity conforming in shape to an exterior of the implant, thereby allowing a precise calibration to be obtained for control points on the exterior of the implant.

(16) Referring to step 1.2, once the phantom is designed, the phantom is fabricated. Although a number of techniques can be used for fabrication, it is desirable that the phantom retain sufficient structural strength and rigidity that the control points do not move from within pre-specified tolerance of their pre-determined design location. Movement of the points from their design location introduces error into the calibration and is thus undesirable. Accordingly, it has been found that forming the control points integrally with the supporting structure provides a desired degree of robustness, structural strength and rigidity to the phantom as compared with assembling the phantom from interconnecting pieces. A technique that has been found amenable to this type of fabrication is additive manufacturing, commonly referred to as 3D printing.

(17) A variety of additive manufacturing techniques are commercially available; however, since it is desired that the phantom be formed from a plastic resin, a preferred technique is photopolymeric printing using UV light as a curative. A plastic resin is chosen that is discernable from surrounding fluid (paramagnetic liquid or air, depending on whether MRI or CT imaging techniques are used), provides sufficient structural strength, and does not swell or otherwise dimensionally distort upon absorption of the surrounding fluid. Suitable examples include acrylic and/or polyacrylate resins.

(18) Referring to step 1.3, following fabrication the phantom is measured for geometric accuracy relative to the design and the precise location of the control points with reference to a particular datum (e.g. a center of the phantom) is determined. This measurement may be obtained using a co-ordinate measuring machine, a measuring microscope, a micro CT scanner, or any other suitably accurate technique.

(19) Referring to step 1.4, the phantom is then optionally immersed within a fluid tight transparent container, such as a plastic container. The container provides a protective shell for the phantom and also controls the magnetic and density properties of the space surrounding the phantom in order to provide consistent contrast between the phantom and the surrounding space.

(20) Referring to FIG. 1b, step 2.1, the phantom is imaged using a three dimensional MRI or CT scanner. The phantom is first located at an isocenter of the MRI or CT scanner so that the spatial vectors obtained through analysis of the image can be used to calibrate the MRI or CT machine relative to a known registration reference of the machine. Then, referring to step 2.2, an image is obtained using either a 3D spin-echo pulse sequence (for MRI) or a high resolution CT sequence. This is then used in step 2.3 to reconstruct a precise image of the phantom that is registered to the isocenter of the scanner.

(21) Referring to FIG. 1c, in step 3.1, the signal intensity is corrected for inhomogeneity relative to the isocenter in both the axial and trans-axial directions. This improves the uniformity of contrast between the phantom and the surrounding space across the entire image, which simplifies subsequent image processing. Referring to steps 3.2 and 3.3, the images are first interpolated to a finer isotropic resolution and then segmented based on a grey-scale threshold to obtain crisp image boundaries. Referring to step 3.4, the crisp boundaries of the segmented image are eroded on a pixel by pixel basis to obtain new image boundaries. The effect of this erosion technique on the image is similar to what would be obtained if the phantom were dipped in acid for a known amount of time and then re-imaged; in other words, all boundaries of the image are reduced in size in 3D space by the same increment. By performing successive erosion operations, the supporting structure eventually disappears from the image, leaving behind clusters of pixels representative of the control points. Thus, the control points can be readily discerned from the supporting structure on the basis of size. Referring to step 3.5, a centroid of each cluster of pixels is identified to determine a precise location for each control point represented in the image.

(22) Referring to step 3.6, by comparing the location of the control points in the image with the actual or true ground location of the control points obtained by precise measurement of the phantom, a deviation between the image and the phantom may be observed. This deviation is used to obtain a spatial vector representative of the magnitude and direction of the deviation. The spatial vector represents the adjustment required to align the control points in the image with the actual control points in the phantom. Thus, the spatial vectors for each control point may be used to calibrate the imaging system or post-process the image to improve its accuracy. The spatial vectors may be represented for ease of interpretation on a vector map (as shown in FIG. 10) that relates the location of the points on the image to the true location of the points in the phantom.

(23) The above methodology may be implemented using a calibration system or calibration kit comprising the phantom, optionally enclosed within the container, along with the measured locations of the control points and software or computer hardware configured to execute machine readable instructions for analyzing an acquired image.

(24) The instructions include the steps of identifying crisp boundaries of the image, eroding the boundaries, determining the centroid of the clusters and comparing those with the location of the control points in the phantom to arrive at a spatial vector for use in calibration and/or spatial mapping.

(25) Further features and embodiments of the foregoing will be evident to persons of skill in the art. The inventor intends to cover all features, embodiments and sub-combinations thereof disclosed herein. The claims are to be construed as broadly as possible with reference to the specification as a whole. The invention may further be understood with reference to the following Examples.

EXAMPLES

(26) [Component 1] Referring to FIG. 2, the proposed distortion phantom is comprised of 4.5 mm diameter spheres, supported by 1.5 mm diameter cylindrical struts on nominal 13 mm spacing. A prototype version of this phantom is shown in FIG. 3, consisting of a 955 matrix, was fabricated using photopolymeric 3D printing (Objet 30 Pro, Stratasys) with a transparent, rigid resin (VeroClear, Stratasys), with flexural modulus exceeding 2 GPa and water absorption of <1.5%. The 3D printer is designed to print over a 302015 cm volume, with accuracy of 0.1 mm. The fabricated plastic grid phantom was evaluated for geometric accuracy using a measuring microscope with accuracy of 0.0005 mm (STM-6, Olympus) and a micro-CT scanner (eXplore Ultra GE Healthcare). The measured control-point spacing was determined to be 13.079 mm, within 0.6% of the nominal value. Micro-CT analysis showed that control-point centroids were within 0.14 mm of their prescribed locations, on average. Referring to FIG. 4, after fabrication, the 3D plastic construct was immersed in a tissue-mimicking paramagnetic fluid to provide appropriate background signal, with T1<200 ms. A copper sulphate solution (7.8 mmol) in saline was used, following the description of Och et al. (Och 1992). The grid phantom exhibited a low volumetric packing fraction within the background liquid, displacing less than 5% of the imaging volume.

(27) [Component 2] Images were acquired at 3T (Discovery 750, GE Medical Systems) with a multi-channel knee coil, using a 3D turbo spin-echo sequence (CUBE, TR=2300 ms, TE=65 ms, flip angle=90, 0.7 mm slice thickness, 0.7 mm in-plane resolution, 62.5 kHz readout bandwidth, matrix size 320320160).

(28) [Component 3] To improve the accuracy of image segmentation, the resulting images were corrected for signal-intensity inhomogeneity in the axial and trans-axial directions, using fitted parabolic functions (FIG. 5). The resulting images of a dark grid on a bright background were interpolated to isotropic 0.43 mm resolution and segmented based on a grey-scale threshold (FIG. 6). Similar images were acquired with a micro-CT imaging system (eXplore Ultra, GE Medical Systems) (FIG. 7).

(29) [Component 3] To isolate individual fiducial locations within the grid, the segmented (binary) image (FIG. 8) was morphologically eroded to remove the struts, while retaining the beads at each intersection. In this case, erosion by a 3-pixel kernel removed 1.2 mm from every surface, thereby completely removing the struts and reducing the spheres to 2 mm diameter, as shown in FIG. 9. Each of the 175 spheres was then identified and centroided to create a 3D point cloud of observed grid locations. These measured locations were subsequently compared to the best-fit locations of a synthesized grid, based on the known grid spacing.

(30) [Component 3] FIG. 10 displays the produced 3D vector map of sub-voxel geometric deviation throughout the image volume. The derived geometric deviation map over the 11.76.56.5 cm3 volume showed average geometric deviations of 0.53 mm, ranging from 0.11 to 1.20 mm. The embodiment comprises a rigid and accurate phantom, which produces images that are amenable to fully automated quantitative analysis. This approach will be useful in any clinical application where geometric accuracy is important, either in routine quality assurance or as a component of distortion correction utilities.

(31) References: The contents of the entirety of each of which are incorporated by this reference. U.S. Pat. No. 4,644,276 issued Feb. 17, 1987. Three-Dimensional Nuclear Magnetic Resonance Phantom. U.S. Pat. No. 8,189,889 issued May 29, 2012. Systems and Methods for Characterizing Spatial Distortion in 3D Imaging Systems. U.S. Pat. No. 5,005,578 issued Apr. 9, 1991. Three-Dimensional Magnetic Resonance Image Distortion Correction Method and System. U.S. Pat. No. 5,872,829 issued Feb. 16, 1999. Method for the Detection and Correction of Image Distortions in Medical Imaging. U.S. Pat. No. 7,330,026 issued Feb. 12, 2008. Method and Apparatus for Mapping and Correcting Geometric Distortion in MRI. U.S. Pat. No. 7,535,227 issued May 19, 2009. Method and Apparatus for Correcting Distortion in MR Images Caused by Metallic Implants. US Patent Publication 2009/0169083 published Jul. 2, 2009. Method and Device for Distortion Correction in Magnetic Resonance Imaging. US Patent Publication 2012/0201438 published Aug. 9, 2012. Calibration Phantom and Method for Measuring and Correcting Geometric Distortions in Medical Images. J. F. Aubry, J. Cheung, O. Morin, L. Beaulieu, I. C. Hsu, J. Pouliot. Investigation of geometric distortions on magnetic resonance and cone beam computed tomography images used for planning and verification of high-dose rate brachytherapy cervical cancer treatment. Brachytherapy. 9, 266-273 (2010). L. N. Baldwin, K. Wachowicz, B. G. Fallone. A two-step scheme for distortion rectification of magnetic resonance images. Med Phys. 36, 3917-3926 (2009). L. N. Baldwin, K. Wachowicz, S. D. Thomas, R. Rivest, B. G. Fallone. Characterization, prediction, and correction of geometric distortion in 3T MR images. Med Phys. 34, 388-399 (2007). Z. Chen, C. M. Ma, K. Paskalev, J. Li, J. Yang, T. Richardson, L. Palacio, X. Xu, L. Chen. Investigation of MR image distortion for radiotherapy treatment planning of prostate cancer. Phys Med Biol. 51, 1393-1403 (2006). S. P. Crijns, C. J. Bakker, P. R. Seevinck, H. de Leeuw, J. J. Lagendijk, B. W. Raaymakers. Towards inherently distortion-free MR images for image-guided radiotherapy on an MRI accelerator. Phys Med Biol. 57, 1349-1358 (2012). S. J. Doran, L. Charles-Edwards, S. A. Reinsberg, M. O. Leach. A complete distortion correction for MR images: I. Gradient warp correction. Phys Med Biol. 50, 1343-1361 (2005). S. Kiryu, Y. Inoue, Y. Masutani, T. Haishi, K. Yoshikawa, M. Watanabe, K. Ohtomo. Distortion correction in whole-body imaging of live mice using a 1-Tesla compact magnetic resonance imaging system. Jpn J Radiol. 29, 353-360 (2011). D. Kittle, B. Holshouser, J. M. Slater, B. D. Guenther, N. P. Pitsianis, R. D. Pearlstein. Technical note: rapid prototyping of 3D grid arrays for image guided therapy quality assurance. Med Phys. 35, 5708-5712 (2008). S. P. Krishnan, et al. A review of rapid prototyped surgical guides for patient-specific total knee replacement. J Bone Joint Surg Br. 94(11), 1457-1461 (2012). B. H. Kristensen, F. J. Laursen, V. Logager, P. F. Geertsen, A. Krarup-Hansen. Dosimetric and geometric evaluation of an open low-field magnetic resonance simulator for radiotherapy treatment planning of brain tumours. Radiother Oncol. 87, 100-109 (2008). N. Maikusa, F. Yamashita, K. Tanaka, O. Abe, A. Kawaguchi, H. Kabasawa, S. Chiba, A. Kasahara, N. Kobayashi, T. Yuasa, N. Sato, H. Matsuda, T. Iwatsubo. Improved volumetric measurement of brain structure with a distortion correction procedure using an ADNI phantom. Med Phys. 40, 062303 (2013). T. Mizowaki, Y. Nagata, K. Okajima, M. Kokubo, Y. Negoro, N. Araki, M. Hiraoka. Reproducibility of geometric distortion in magnetic resonance imaging based on phantom studies. Radiother Oncol. 57, 237-242 (2000). J. G. Och, G. D. Clarke, W. T. Sobol, C. W. Rosen, S. K. Mun. Acceptance testing of magnetic resonance imaging systems: report of AAPM Nuclear Magnetic Resonance Task Group No. 6. Med Phys. 19, 217-229 (1992). Y. Pauchard, M. R. Smith, M. P. Mintchev. Improving geometric accuracy in the presence of susceptibility difference artifacts produced by metallic implants in magnetic resonance imaging. IEEE Trans Med Imaging. 24, 1387-1399 (2005). R. R. Price, L. Axel, T. Morgan, R. Newman, W. Perman, N. Schneiders, M. Selikson, M. Wood, S. R. Thomas. Quality assurance methods and phantoms for magnetic resonance imaging: report of AAPM nuclear magnetic resonance Task Group No. 1. Med Phys. 17, 287-295 (1990). E. Schneider and M. Nessaiver, The Osteoarthritis Initiative (OAI) magnetic resonance imaging quality assurance update. Osteoarthritis Cartilage. 21(1), 110-116 (2013). T. Stanescu, H. S. Jans, K. Wachowicz, B. G. Fallone. Investigation of a 3D system distortion correction method for MR images. J Appl Clin Med Phys. 11, 2961 (2010). T. Stanescu, K. Wachowicz, D. A. Jaffray. Characterization of tissue magnetic susceptibility-induced distortions for MRIgRT. Med Phys. 39, 7185-7193 (2012). S. F. Tanner, D. J. Finnigan, V. S. Khoo, P. Mayles, D. P. Dearnaley, M. O. Leach. Radiotherapy planning of the pelvis using distortion corrected MR images: the removal of system distortions. Phys Med Biol. 45, 2117-2132 (2000). R. Viard, N. Betrouni, M. Vermandel, S. Mordon, J. Rousseau, M. Vanhoutte. Characterization and 3D correction of geometric distortion in low-field open-magnet MRI. Conf Proc IEEE Eng Med Biol Soc. 2008, 3649-3652 (2008). D. Wang and D. M. Doddrell. A proposed scheme for comprehensive characterization of the measured geometric distortion in magnetic resonance imaging using a three-dimensional phantom. Med Phys. 31, 2212-2218 (2004a). D. Wang, D. M. Doddrell, G. Cowin. A novel phantom and method for comprehensive 3-dimensional measurement and correction of geometric distortion in magnetic resonance imaging. Magn Reson Imaging. 22, 529-542 (2004b). D. Wang, W. Strugnell, G. Cowin, D. M. Doddrell, R. Slaughter. Geometric distortion in clinical MRI systems Part II: correction using a 3D phantom. Magn Reson Imaging. 22, 1223-1232 (2004). D. Wang, W. Strugnell, G. Cowin, D. M. Doddrell, R. Slaughter. Geometric distortion in clinical MRI systems Part I: evaluation using a 3D phantom. Magn Reson Imaging. 22, 1211-1221 (2004).

(32) The novel features will become apparent to those of skill in the art upon examination of the description. It should be understood, however, that the scope of the claims should not be limited by the embodiments, but should be given the broadest interpretation consistent with the wording of the claims and the specification as a whole.