Publications

2006
Ion-Florin Talos, Kelly H Zou, Lucila Ohno-Machado, Jui G Bhagwat, Ron Kikinis, Peter M Black, and Ferenc A Jolesz. 2006. “Supratentorial low-grade glioma resectability: statistical predictive analysis based on anatomic MR features and tumor characteristics.” Radiology, 239, 2, Pp. 506-13.Abstract
PURPOSE: To retrospectively assess the main variables that affect the complete magnetic resonance (MR) imaging-guided resection of supratentorial low-grade gliomas. MATERIALS AND METHODS: Institutional review board approval was obtained for this retrospective HIPAA-compliant study, with the requirement for informed consent waived. Data from 101 patients (61 men, 40 women; mean age, 39 years; age range, 18-72 years) who had nonenhancing supratentorial mass lesions that were histopathologically diagnosed as low-grade (World Health Organization grade II) gliomas and consecutively underwent surgery with intraoperative MR imaging guidance were analyzed. There were 21 low-grade astrocytomas, 64 oligodendrogliomas, and 16 mixed oligoastrocytomas. Initial and residual tumor volumes were measured on intraoperative T2-weighted MR images and three-dimensional spoiled gradient-echo MR images. The anatomic relationships between the tumor and eloquent cortical and/or subcortical regions and the influence of these relationships on the extent of resection were analyzed on the basis of preoperative MR imaging findings. Summary measures, univariate Fisher exact test and t test, and multivariate logistic regression analyses were performed. RESULTS: Tumor volume ranged from 2.7-231.0 mL. Univariate analyses revealed the following tumor characteristics to be significant predictive variables of incomplete tumor resection: diffuse tumor margin on T2-weighted MR images, oligodendroglioma or oligoastrocytoma histopathologic type, and large tumor volume (P < .05 for all). Tumor involvement of the following structures was associated with incomplete resection: corpus callosum, corticospinal tract, insular lobe, middle cerebral artery, motor cortex, optic radiation, visual cortex, and basal ganglia (P < .05 for all). Multivariate analyses revealed that incomplete tumor resection was due to tumor involvement of the corticospinal tract (P < .01), large tumor volume (P < .01), and oligodendroglioma histopathologic type (P = .02). CONCLUSION: The main variables associated with incomplete tumor resection in 101 patients were identified by using statistical predictive analyses.
Manabu Kinoshita, Nathan McDannold, Ferenc A Jolesz, and Kullervo Hynynen. 2006. “Targeted delivery of antibodies through the blood-brain barrier by MRI-guided focused ultrasound.” Biochem Biophys Res Commun, 340, 4, Pp. 1085-90.Abstract
The blood-brain barrier (BBB) is a persistent obstacle for the local delivery of macromolecular therapeutic agents to the central nervous system (CNS). Many drugs that show potential for treating CNS diseases cannot cross the BBB and there is a need for a non-invasive targeted drug delivery method that allows local therapy of the CNS using larger molecules. We developed a non-invasive technique that allows the image-guided delivery of antibody across the BBB into the murine CNS. Here, we demonstrate that subsequent to MRI-targeted focused ultrasound induced disruption of BBB, intravenously administered dopamine D(4) receptor-targeting antibody crossed the BBB and recognized its antigens. Using MRI, we were able to monitor the extent of BBB disruption. This novel technology should be useful in delivering macromolecular therapeutic or diagnostic agents to the CNS for the treatment of various CNS disorders.
Nathan McDannold, Clare M Tempany, Fiona M Fennessy, Minna J So, Frank J Rybicki, Elizabeth A Stewart, Ferenc A Jolesz, and Kullervo Hynynen. 2006. “Uterine leiomyomas: MR imaging-based thermometry and thermal dosimetry during focused ultrasound thermal ablation.” Radiology, 240, 1, Pp. 263-72.Abstract
PURPOSE: To retrospectively evaluate magnetic resonance (MR) imaging-based thermometry and thermal dosimetry during focused ultrasound treatments of uterine leiomyomas (ie, fibroids). MATERIALS AND METHODS: All patients gave written informed consent for the focused ultrasound treatments and the current HIPAA-compliant retrospective study, both of which were institutional review board approved. Thermometry performed during the treatments of 64 fibroids in 50 women (mean age, 46.6 years +/- 4.5 [standard deviation]) was used to create thermal dose maps. The areas that reached dose values of 240 and 18 equivalent minutes at 43 degrees C were compared with the nonperfused regions measured on contrast material-enhanced MR images by using the Bland-Altman method. Volume changes in treated fibroids after 6 months were compared with volume changes in nontreated fibroids and with MR-based thermal dose estimates. RESULTS: While the thermal dose estimates were shown to have a clear relationship with resulting nonperfused regions, the nonperfused areas were, on average, larger than the dose estimates (means of 1.9 +/- 0.7 and 1.2 +/- 0.4 times as large for areas that reached 240- and 18-minute threshold dose values, respectively). Good correlation was observed for smaller treatment volumes at the lower dose threshold (mean ratio, 1.0 +/- 0.3), but for larger treatment volumes, the nonperfused region extended to locations within the fibroid that clearly were not heated. Variations in peak temperature increase were as large as a factor of two, both between patients and within individual treatments. On average, the fibroid volume reduction at 6 months increased as the ablated volume estimated by using the thermal dose increased. CONCLUSION: Study results showed good correlation between thermal dose estimates and resulting nonperfused areas for smaller ablated volumes. For larger treatment volumes, nonperfused areas could extend within the fibroid to unheated areas.
Michael Greenspan, Liping Ingrid Wang, and Randy Ellis. 2006. “Validation and improved registration of bone segmentation using contour coherency.” Conf Proc IEEE Eng Med Biol Soc, 1, Pp. 244-7.Abstract
A method is presented to validate the segmentation of computed tomography (CT) image sequences, and im prove the accuracy and efficiency of the subsequent registration of the 3D surfaces that are reconstructed from the segmented slices. The method compares the shapes of contours extracted from neighborhoods of slices in CT stacks of tibias. The bone is first segmented by an automatic segmentation technique, and the bone contour for each slice is parameterized as a 1-D function of normalized arc length versus inscribed angle. These functions are represented as vectors within a K-dimensional space comprising the first K amplitude coefficients of their Fourier Descriptors. The similarity or coherency of neighboring contours is measured by comparing statistical properties of their vector representations within this space. Experimentation has demonstrated this technique to be very effective at automatically identifying low coherency segmentations, the removal of which significantly improved the accuracy and time efficiency of the registration of 3-D bone surface models.
Liping Ingrid Wang, Michael Greenspan, and Randy Ellis. 2006. “Validation of bone segmentation and improved 3-D registration using contour coherency in CT data.” IEEE Trans Med Imaging, 25, 3, Pp. 324-34.Abstract
A method is presented to validate the segmentation of computed tomography (CT) image sequences, and improve the accuracy and efficiency of the subsequent registration of the three-dimensional surfaces that are reconstructed from the segmented slices. The method compares the shapes of contours extracted from neighborhoods of slices in CT stacks of tibias. The bone is first segmented by an automatic segmentation technique, and the bone contour for each slice is parameterized as a one-dimensional function of normalized arc length versus inscribed angle. These functions are represented as vectors within a K-dimensional space comprising the first K amplitude coefficients of their Fourier Descriptors. The similarity or coherency of neighboring contours is measured by comparing statistical properties of their vector representations within this space. Experimentation has demonstrated this technique to be very effective at identifying low-coherency segmentations. Compared with experienced human operators, in a set of 23 CT stacks (1,633 slices), the method correctly detected 87.5% and 80% of the low-coherency and 97.7% and 95.5% of the high coherency segmentations, respectively from two different automatic segmentation techniques. Removal of the automatically detected low-coherency segmentations also significantly improved the accuracy and time efficiency of the registration of 3-D bone surface models. The registration error was reduced by over 500% (i.e., a factor of 5) and 280%, and the computational performance was improved by 540% and 791% for the two respective segmentation methods.
Simon K Warfield, Kelly H Zou, and William M Wells. 2006. “Validation of image segmentation by estimating rater bias and variance.” Med Image Comput Comput Assist Interv, 9, Pt 2, Pp. 839-47.Abstract
The accuracy and precision of segmentations of medical images has been difficult to quantify in the absence of a "ground truth" or reference standard segmentation for clinical data. Although physical or digital phantoms can help by providing a reference standard, they do not allow the reproduction of the full range of imaging and anatomical characteristics observed in clinical data. An alternative assessment approach is to compare to segmentations generated by domain experts. Segmentations may be generated by raters who are trained experts or by automated image analysis algorithms. Typically these segmentations differ due to intra-rater and inter-rater variability. The most appropriate way to compare such segmentations has been unclear. We present here a new algorithm to enable the estimation of performance characteristics, and a true labeling, from observations of segmentations of imaging data where segmentation labels may be ordered or continuous measures. This approach may be used with, amongst others, surface, distance transform or level set representations of segmentations, and can be used to assess whether or not a rater consistently over-estimates or under-estimates the position of a boundary.
2005
Olivier Clatz, Hervé Delingette, Ion-Florin Talos, Alexandra J Golby, Ron Kikinis, Ferenc A Jolesz, Nicholas Ayache, and Simon K Warfield. 2005. “Hybrid Formulation of the Model-based Non-rigid Registration Problem to Improve Accuracy and Robustness.” Med Image Comput Comput Assist Interv, 8, Pt 2, Pp. 295-302.Abstract
We present a new algorithm to register 3D pre-operative Magnetic Resonance (MR) images with intra-operative MR images of the brain. This algorithm relies on a robust estimation of the deformation from a sparse set of measured displacements. We propose a new framework to compute iteratively the displacement field starting from an approximation formulation (minimizing the sum of a regularization term and a data error term) and converging toward an interpolation formulation (least square minimization of the data error term). The robustness of the algorithm is achieved through the introduction of an outliers rejection step in this gradual registration process. We ensure the validity of the deformation by the use of a biomechanical model of the brain specific to the patient, discretized with the finite element method. The algorithm has been tested on six cases of brain tumor resection, presenting a brain shift up to 13 mm.
Steven Haker, William M Wells, Simon K Warfield, Ion-Florin Talos, Jui G Bhagwat, Daniel Goldberg-Zimring, Asim Mian, Lucila Ohno-Machado, and Kelly H Zou. 2005. “Combining Classifiers using their Receiver Operating Characteristics and Maximum Likelihood Estimation.” Med Image Comput Comput Assist Interv, 8, Pt 1, Pp. 506-14.Abstract

In any medical domain, it is common to have more than one test (classifier) to diagnose a disease. In image analysis, for example, there is often more than one reader or more than one algorithm applied to a certain data set. Combining of classifiers is often helpful, but determining the way in which classifiers should be combined is not trivial. Standard strategies are based on learning classifier combination functions from data. We describe a simple strategy to combine results from classifiers that have not been applied to a common data set, and therefore can not undergo this type of joint training. The strategy, which assumes conditional independence of classifiers, is based on the calculation of a combined Receiver Operating Characteristic (ROC) curve, using maximum likelihood analysis to determine a combination rule for each ROC operating point. We offer some insights into the use of ROC analysis in the field of medical imaging.

Yan Yang, Lei Zhu, Steven Haker, Allen R Tannenbaum, and Don P Giddens. 2005. “Harmonic Skeleton Guided Evaluation of Stenoses in Human Coronary Arteries.” Med Image Comput Comput Assist Interv, 8, Pt 1, Pp. 490-7.Abstract

This paper presents a novel approach that three-dimensionally visualizes and evaluates stenoses in human coronary arteries by using harmonic skeletons. A harmonic skeleton is the center line of a multi-branched tubular surface extracted based on a harmonic function, which is the solution of the Laplace equation. This skeletonization method guarantees smoothness and connectivity and provides a fast and straightforward way to calculate local cross-sectional areas of the arteries, and thus provides the possibility to localize and evaluate coronary artery stenosis, which is a commonly seen pathology in coronary artery disease.

Steven J Haker, Robert V. Mulkern, Joseph R Roebuck, Agnieszka Szot Barnes, Simon DiMaio, Nobuhiko Hata, and Clare M Tempany. 2005. “Magnetic Resonance Guided Prostate Interventions.” Top Magn Reson Imaging, 16, 5, Pp. 355-68.Abstract

We review our experience using an open 0.5-T magnetic resonance (MR) interventional unit to guide procedures in the prostate. This system allows access to the patient and real-time MR imaging simultaneously and has made it possible to perform prostate biopsy and brachytherapy under MR guidance. We review MR imaging of the prostate and its use in targeted therapy, and describe our use of image processing methods such as image registration to further facilitate precise targeting. We describe current developments with a robot assist system being developed to aid radioactive seed placement.

Lei Zhu, Steven Haker, and Allen Tannenbaum. 2005. “Mass Preserving Registration for Heart MR Images.” Med Image Comput Comput Assist Interv, 8, Pt 2, Pp. 147-54.Abstract

This paper presents a new algorithm for non-rigid registration between two doubly-connected regions. Our algorithm is based on harmonic analysis and the theory of optimal mass transport. It assumes an underlining continuum model, in which the total amount of mass is exactly preserved during the transformation of tissues. We use a finite element approach to numerically implement the algorithm.

Delphine Nain, Steven Haker, Aaron Bobick, and Allen R Tannenbaum. 2005. “Multiscale 3D Shape Analysis using Spherical Wavelets.” Med Image Comput Comput Assist Interv, 8, Pt 2, Pp. 459-67.Abstract

Shape priors attempt to represent biological variations within a population. When variations are global, Principal Component Analysis (PCA) can be used to learn major modes of variation, even from a limited training set. However, when significant local variations exist, PCA typically cannot represent such variations from a small training set. To address this issue, we present a novel algorithm that learns shape variations from data at multiple scales and locations using spherical wavelets and spectral graph partitioning. Our results show that when the training set is small, our algorithm significantly improves the approximation of shapes in a testing set over PCA, which tends to oversmooth data.

Pages