Ars.els-cdn.com



Physical Correction Model for Automatic Correction of Intensity Non-Uniformity in Magnetic Resonance ImagingStefan Leger1, Steffen L?ck1, Volker Hietschold2, Robert Haase3, Hans Joachim B?hme4 and Nasreddin Abolmaali1, 51 OncoRay - National Center for Radiation Research in Oncology, Faculty of Medicine and University Hospital Carl Gustav Carus, Technische Universit?t Dresden, Helmholtz-Zentrum Dresden – Rossendorf, Dresden, Germany2Institute and Policlinic for Radiology, Faculty of Medicine and University Hospital Carl Gustav Carus, Technische Universit?t Dresden, Dresden, Germany3Scientific Computing Facility, Max Planck Institute for Cell Biology and Genetics, Dresden, Germany4Department of Artificial Intelligence, Faculty of Computer Science/Mathematics, University of Applied Science Dresden, Dresden, Germany5Clinic for Radiology, Teaching Hospital Dresden - Friedrichstadt, Technische Universit?t Dresden, Dresden, GermanySupplement A: MRI phantom experimentSuppl. Figure 1a illustrates the 3D-view of the used MRI water phantom. Due to the fixed geometry of the head coil the phantom lies close to the lower coil segments, while there exists an intermediate space between the upper segments and the phantom. This leads to a higher image signal around the lower coil segments (orange arrows, suppl. figure 1a) and a lower image signal close to the other coil segments. Furthermore, it is perceivable that the image signal is slowly decreasing in the direction towards the coil centre. We quantified this assumption by plotting the intensity values along the blue and the red arrows, which are shown in Suppl. Figure 1a-b. The signal decrease from the lower coil segments to the image centre can be described by an exponential function with sufficient accuracy, which is motivated by the damping law, whereas in longitudinal direction we assume a Gaussian like intensity profile.Supplementary Figure 1: In (a, b) the intensity profile is plotted in x, y (blue arrow) and longitudinal direction (red arrow). The data (dots) are fitted by (a) an exponential and (b) a Gaussian function.Supplement B: PCM optimiser configurationFor the parameters of the swarm intelligence optimisation algorithm we used a colony size of 100 and a trail limit of 75. The number of iterations was set to 500. To reduce the computation time we defined a convergence threshold of 20 which stops the optimisation process if no new global optimum could be found until the convergence threshold is reached. In addition to that, we down sampled each volume by a factor of 4. Furthermore, the image background was identified by the Otsu threshold method ADDIN CSL_CITATION { "citationItems" : [ { "id" : "ITEM-1", "itemData" : { "author" : [ { "dropping-particle" : "", "family" : "Otsu", "given" : "N", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" } ], "container-title" : "Automatica", "id" : "ITEM-1", "issued" : { "date-parts" : [ [ "1975" ] ] }, "title" : "Otsu1975", "type" : "article-journal" }, "uris" : [ "" ] } ], "mendeley" : { "formattedCitation" : "[1]", "plainTextFormattedCitation" : "[1]", "previouslyFormattedCitation" : "[1]" }, "properties" : { "noteIndex" : 0 }, "schema" : "" }[1] and was not considered during the correction process. For the penalty parameter σr = 0.1 was chosen.Supplement C: N4 configurationAccording to Tustison et al. ADDIN CSL_CITATION { "citationItems" : [ { "id" : "ITEM-1", "itemData" : { "DOI" : "10.1109/TMI.2010.2046908", "ISBN" : "1558-254X (Electronic)\\r0278-0062 (Linking)", "ISSN" : "02780062", "PMID" : "20378467", "abstract" : "A variant of the popular nonparametric nonuniform intensity normalization (N3) algorithm is proposed for bias field correction. Given the superb performance of N3 and its public availability, it has been the subject of several evaluation studies. These studies have demonstrated the importance of certain parameters associated with the B-spline least-squares fitting. We propose the substitution of a recently developed fast and robust B-spline approximation routine and a modified hierarchical optimization scheme for improved bias field correction over the original N3 algorithm. Similar to the N3 algorithm, we also make the source code, testing, and technical documentation of our contribution, which we denote as ??N4ITK,?? available to the public through the Insight Toolkit of the National Institutes of Health. Performance assessment is demonstrated using simulated data from the publicly available Brainweb database, hyperpolarized 3He lung image data, and 9.4T postmortem hippocampus data.", "author" : [ { "dropping-particle" : "", "family" : "Tustison", "given" : "Nicholas J", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Avants", "given" : "Brian B", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Cook", "given" : "Philip A", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Zheng", "given" : "Yuanjie", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Egan", "given" : "Alexander", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Yushkevich", "given" : "Paul A", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Gee", "given" : "James C", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" } ], "container-title" : "IEEE Transactions on Medical Imaging", "id" : "ITEM-1", "issue" : "6", "issued" : { "date-parts" : [ [ "2010", "6" ] ] }, "page" : "1310-1320", "title" : "N4ITK: Improved N3 bias correction", "type" : "article-journal", "volume" : "29" }, "uris" : [ "" ] } ], "mendeley" : { "formattedCitation" : "[2]", "plainTextFormattedCitation" : "[2]", "previouslyFormattedCitation" : "[2]" }, "properties" : { "noteIndex" : 0 }, "schema" : "" }[2] we used two different spline distance configurations for the N4 correction algorithm, 50 mm (N450) and 100 mm (N4100), respectively. The other N4 parameters were set to: full width at half maximum = 0.15, convergence threshold = 0.0001, maximum number of iterations = (100,100,100), bins = 200, volume re-sampling factor = 4. We used the N4 implementation in the Advanced Normalization Tools (ANTs) ADDIN CSL_CITATION { "citationItems" : [ { "id" : "ITEM-1", "itemData" : { "author" : [ { "dropping-particle" : "", "family" : "Brian B. Avants", "given" : "Nick Tustison and Gang Song", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" } ], "id" : "ITEM-1", "issued" : { "date-parts" : [ [ "0" ] ] }, "title" : "Advanced Normalization Tools (ANTS)", "type" : "article-journal" }, "uris" : [ "" ] } ], "mendeley" : { "formattedCitation" : "[3]", "plainTextFormattedCitation" : "[3]", "previouslyFormattedCitation" : "[3]" }, "properties" : { "noteIndex" : 0 }, "schema" : "" }[3] framework. Other configurations of the N4 algorithm were tested without improving the results.Supplement D: DatasetsWe performed four evaluation studies to demonstrate the applicability of the PCM algorithm. Dataset I consists of 9 different simulated MRI brain volumes (T1, T2, PD) created by the BrainWeb-MRI simulator ADDIN CSL_CITATION { "citationItems" : [ { "id" : "ITEM-1", "itemData" : { "DOI" : "10.1.1.51.3917", "ISSN" : "1053--8119", "abstract" : "Introduction The increased importance of automated computer techniques for anatomical brain mapping from MR images and quantitative brain image analysis methods leads to an increased need for validation and evaluation of the effect of image acquisition parameters on performance of these procedures. Validation of analysis techniques of in-vivo acquired images is complicated due to the lack of reference data (\"ground truth\"). Also, optimal selection of the MR imaging parameters is difficult due to the large parameter space. BrainWeb makes available to the neuroimaging community, on-line on WWW, a set of realistic simulated brain MR image volumes (Simulated Brain Database, SBD) that allows the above issues to be examined in a controlled, systematic way. Features The SBD was generated by varying specific imaging parameters in an MRI simulator, which starts from a digital phantom, and performs a realistic, first-principles modeling of the imaging process based on", "author" : [ { "dropping-particle" : "", "family" : "Cocosco", "given" : "Chris A.", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Kollokian", "given" : "Vasken", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Kwan", "given" : "Remi K.-S.", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Pike", "given" : "G. Bruce", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Evans", "given" : "Alan C.", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" } ], "container-title" : "3-rd International Conference on Functional Mapping of the Human Brain", "id" : "ITEM-1", "issue" : "3", "issued" : { "date-parts" : [ [ "1996" ] ] }, "page" : "1996", "title" : "BrainWeb : Online Interface to a 3D MRI Simulated Brain Database", "type" : "article-journal", "volume" : "1131" }, "uris" : [ "" ] } ], "mendeley" : { "formattedCitation" : "[4]", "plainTextFormattedCitation" : "[4]", "previouslyFormattedCitation" : "[4]" }, "properties" : { "noteIndex" : 0 }, "schema" : "" }[4] with different intensity non-uniformity levels of 0%, 20% and 40%, respectively. The voxel spacing was 1.0 x 1.0 x 1.0 mm with a volume dimension of 181 x 217 x 181 voxels.Dataset II consists of 93 multi-channel volumes from the human brain of 27 patients which were consecutively acquired in the 3rd quarter of 2013 on a 1.5 Tesla MR Siemens Avanto system during the clinical routine. In particular 24 T1-, 16 T2-, 14 PD-weighted MR images without contrast agent and 21 T1-, 9 T2-, 9 PD-weighted MR images with contrast agent were used. The T1 images were acquired with a spin echo sequence. The common repetition time and echo time for the T1 images was 450 ms and 7.7 ms, respectively, with a flip angle of 90.0 degrees. For the T2 and PD images a turbo spin echo sequence was used with a common repetition time of 3330 ms and a flip angle of 150 degrees. The echo time was 13 ms for T2 and 81 ms for PD images. The common image acquisition matrix was 384 × 512 × 27 voxels with a voxel size of 0.48 x 0.48 x 6.0 mm.Dataset III consists of 27 T1- and 27 T2-weighted MRI volumes from the human brain, acquired on a 3 Tesla MR scanner. This dataset is publicly available and part of the Human Connectome Project consortium ADDIN CSL_CITATION { "citationItems" : [ { "id" : "ITEM-1", "itemData" : { "DOI" : "10.1016/j.neuroimage.2013.05.041", "ISBN" : "1053-8119", "ISSN" : "10538119", "PMID" : "23684880", "abstract" : "The Human Connectome Project consortium led by Washington University, University of Minnesota, and Oxford University is undertaking a systematic effort to map macroscopic human brain circuits and their relationship to behavior in a large population of healthy adults. This overview article focuses on progress made during the first half of the 5-year project in refining the methods for data acquisition and analysis. Preliminary analyses based on a finalized set of acquisition and preprocessing protocols demonstrate the exceptionally high quality of the data from each modality. The first quarterly release of imaging and behavioral data via the ConnectomeDB database demonstrates the commitment to making HCP datasets freely accessible. Altogether, the progress to date provides grounds for optimism that the HCP datasets and associated methods and software will become increasingly valuable resources for characterizing human brain connectivity and function, their relationship to behavior, and their heritability and genetic underpinnings. \u00a9 2013 Elsevier Inc.", "author" : [ { "dropping-particle" : "", "family" : "Essen", "given" : "David C.", "non-dropping-particle" : "Van", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Smith", "given" : "Stephen M.", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Barch", "given" : "Deanna M.", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Behrens", "given" : "Timothy E.J.", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Yacoub", "given" : "Essa", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Ugurbil", "given" : "Kamil", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" } ], "container-title" : "NeuroImage", "id" : "ITEM-1", "issued" : { "date-parts" : [ [ "2013" ] ] }, "page" : "62-79", "publisher" : "Elsevier Inc.", "title" : "The WU-Minn Human Connectome Project: An overview", "type" : "article-journal", "volume" : "80" }, "uris" : [ "" ] } ], "mendeley" : { "formattedCitation" : "[5]", "plainTextFormattedCitation" : "[5]" }, "properties" : { "noteIndex" : 0 }, "schema" : "" }[5]. The common image acquisition matrix was 256 x 320 x 320 voxels with a voxel size of 0.7 x 0.7 x 0.7 mm. Further details are described in ADDIN CSL_CITATION { "citationItems" : [ { "id" : "ITEM-1", "itemData" : { "DOI" : "10.1016/j.neuroimage.2013.05.041", "ISBN" : "1053-8119", "ISSN" : "10538119", "PMID" : "23684880", "abstract" : "The Human Connectome Project consortium led by Washington University, University of Minnesota, and Oxford University is undertaking a systematic effort to map macroscopic human brain circuits and their relationship to behavior in a large population of healthy adults. This overview article focuses on progress made during the first half of the 5-year project in refining the methods for data acquisition and analysis. Preliminary analyses based on a finalized set of acquisition and preprocessing protocols demonstrate the exceptionally high quality of the data from each modality. The first quarterly release of imaging and behavioral data via the ConnectomeDB database demonstrates the commitment to making HCP datasets freely accessible. Altogether, the progress to date provides grounds for optimism that the HCP datasets and associated methods and software will become increasingly valuable resources for characterizing human brain connectivity and function, their relationship to behavior, and their heritability and genetic underpinnings. \u00a9 2013 Elsevier Inc.", "author" : [ { "dropping-particle" : "", "family" : "Essen", "given" : "David C.", "non-dropping-particle" : "Van", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Smith", "given" : "Stephen M.", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Barch", "given" : "Deanna M.", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Behrens", "given" : "Timothy E.J.", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Yacoub", "given" : "Essa", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Ugurbil", "given" : "Kamil", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" } ], "container-title" : "NeuroImage", "id" : "ITEM-1", "issued" : { "date-parts" : [ [ "2013" ] ] }, "page" : "62-79", "publisher" : "Elsevier Inc.", "title" : "The WU-Minn Human Connectome Project: An overview", "type" : "article-journal", "volume" : "80" }, "uris" : [ "" ] } ], "mendeley" : { "formattedCitation" : "[5]", "plainTextFormattedCitation" : "[5]" }, "properties" : { "noteIndex" : 0 }, "schema" : "" }[5].Dataset IV consists of 15 T1- and 14 T2-weighted MRI volumes of the abdomen with and without contrast agent which were acquired within a pancreas MRI study. All data were acquired on a 1.5 Tesla MR Siemens Avanto system. The common image acquisition matrix was 512 x 416 x 80 voxels with a voxel size of 0.87 x 0.87 x 3.0 mm. The common repetition time and echo time for the T1 images was 4.9 ms and 2.2 ms, respectively, with a flip angle of 10.0 degrees. For the T2 images the common repetition time and echo time was 4101 ms and 86 ms, respectively, with a and flip angle of 140 degrees.For dataset I, II and III the tissue classes for grey-matter (gm) and white-matter (wm) as well as cerebro-spinal fluid (csf) were available or have been automatically segmented by the Expectation-Maximisation algorithm of the Slicer Software ADDIN CSL_CITATION { "citationItems" : [ { "id" : "ITEM-1", "itemData" : { "DOI" : "10.1016/j.mri.2012.05.001", "ISBN" : "1873-5894 (Electronic) 0730-725X (Linking)", "ISSN" : "0730725X", "PMID" : "22770690", "abstract" : "Quantitative analysis has tremendous but mostly unrealized potential in healthcare to support objective and accurate interpretation of the clinical imaging. In 2008, the National Cancer Institute began building the Quantitative Imaging Network (QIN) initiative with the goal of advancing quantitative imaging in the context of personalized therapy and evaluation of treatment response. Computerized analysis is an important component contributing to reproducibility and efficiency of the quantitative imaging techniques. The success of quantitative imaging is contingent on robust analysis methods and software tools to bring these methods from bench to bedside.3D Slicer is a free open-source software application for medical image computing. As a clinical research tool, 3D Slicer is similar to a radiology workstation that supports versatile visualizations but also provides advanced functionality such as automated segmentation and registration for a variety of application domains. Unlike a typical radiology workstation, 3D Slicer is free and is not tied to specific hardware. As a programming platform, 3D Slicer facilitates translation and evaluation of the new quantitative methods by allowing the biomedical researcher to focus on the implementation of the algorithm and providing abstractions for the common tasks of data communication, visualization and user interface development. Compared to other tools that provide aspects of this functionality, 3D Slicer is fully open source and can be readily extended and redistributed. In addition, 3D Slicer is designed to facilitate the development of new functionality in the form of 3D Slicer extensions. In this paper, we present an overview of 3D Slicer as a platform for prototyping, development and evaluation of image analysis tools for clinical research applications. To illustrate the utility of the platform in the scope of QIN, we discuss several use cases of 3D Slicer by the existing QIN teams, and we elaborate on the future directions that can further facilitate development and validation of imaging biomarkers using 3D Slicer. \u00a9 2012 Elsevier Inc.", "author" : [ { "dropping-particle" : "", "family" : "Fedorov", "given" : "Andriy", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Beichel", "given" : "Reinhard", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Kalpathy-Cramer", "given" : "Jayashree", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Finet", "given" : "Julien", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Fillion-Robin", "given" : "Jean Christophe", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Pujol", "given" : "Sonia", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Bauer", "given" : "Christian", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Jennings", "given" : "Dominique", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Fennessy", "given" : "Fiona", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Sonka", "given" : "Milan", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Buatti", "given" : "John", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Aylward", "given" : "Stephen", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "V.", "family" : "Miller", "given" : "James", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Pieper", "given" : "Steve", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" }, { "dropping-particle" : "", "family" : "Kikinis", "given" : "Ron", "non-dropping-particle" : "", "parse-names" : false, "suffix" : "" } ], "container-title" : "Magnetic Resonance Imaging", "id" : "ITEM-1", "issue" : "9", "issued" : { "date-parts" : [ [ "2012", "11" ] ] }, "page" : "1323-1341", "title" : "3D Slicer as an image computing platform for the Quantitative Imaging Network", "type" : "article-journal", "volume" : "30" }, "uris" : [ "" ] } ], "mendeley" : { "formattedCitation" : "[6]", "plainTextFormattedCitation" : "[6]", "previouslyFormattedCitation" : "[5]" }, "properties" : { "noteIndex" : 0 }, "schema" : "" }[6] for all modalities. In the case of dataset IV the tissue classes for the kidney, liver and spleen for all volumes were generated semi-automatically by the Slicer Software.Supplement E: Results of correction of simulated Brain-Web datasetSupplementary Table 1 shows the results for dataset I. The first column contains the MRI sequence and simulated bias field strength (%).Supplementary Table 1: Coefficient of variation for simulated MR brain volumes (dataset I).MRI-sequenceBias%OriginalPCMN450N4100T100.390.390.400.38200.420.390.390.38400.480.390.400.3800.420.420.430.42T2200.450.430.430.42400.510.500.450.4400.170.170.200.19PD200.210.170.200.19400.290.190.200.19Supplement F: Example of 3 Tesla MRI brain imageSupplementary Figure 2 shows an example of 3 Tesla MRI volumes acquired by a Philips Ingenuity TF scanner with and without automatic constant level appearance (CLEAR) correction. It is visible that the CLEAR correction (red arrows) leads to an inversion of the physically motivated correction assumption that the image signal emitted by the tissue is slowly decreasing to the coil array centre. In contrast to that, without the CLEAR correction (orange arrows) our assumption is valid. In this case the image quality could be improved by the PCM. Supplementary Figure 2: An example of 3 Tesla volumes with and without automatic constant level appearance (CLEAR) correction as well as the corrected image by the PCM and the correction function. The red and orange arrows show the region of intensity non-uniformity. In the case of the automatic CLEAR correction the physically motivated correction assumption becomes invalid.ReferencesADDIN Mendeley Bibliography CSL_BIBLIOGRAPHY [1]Otsu N. Otsu1975. Automatica 1975.[2]Tustison NJ, Avants BB, Cook PA, Zheng Y, Egan A, Yushkevich PA, et al. N4ITK: Improved N3 bias correction. IEEE Trans Med Imaging 2010;29:1310–20.[3]Brian B. Avants NT and GS. Advanced Normalization Tools (ANTS) 2011.[4]Cocosco CA, Kollokian V, Kwan RK-S, Pike GB, Evans AC. BrainWeb?: Online Interface to a 3D MRI Simulated Brain Database. 3-Rd Int Conf Funct Mapp Hum Brain 1996;1131:1996.[5]Van Essen DC, Smith SM, Barch DM, Behrens TEJ, Yacoub E, Ugurbil K. The WU-Minn Human Connectome Project: An overview. Neuroimage 2013;80:62–79.[6]Fedorov A, Beichel R, Kalpathy-Cramer J, Finet J, Fillion-Robin JC, Pujol S, et al. 3D Slicer as an image computing platform for the Quantitative Imaging Network. Magn Reson Imaging 2012;30:1323–41. ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download