Multiple Image Sets Research Articles

Measurements of intrafraction motion and interfraction and intrafraction rotation of prostate by three-dimensional analysis of daily portal imaging with radiopaque markers

Purpose To measure the interfraction and intrafraction motion of the prostate during the course of external beam radiotherapy using a video electronic portal imaging device and three-dimensional analysis. Methods and materials Eighteen patients underwent implantation with two or three gold markers in the prostate before five-angle/11-field conformal radiotherapy. Using CT data as the positional reference, multiple daily sets of portal images, and a three-dimensional reconstruction algorithm, intrafraction translations, as well as interfraction and intrafraction rotations, were analyzed along the three principal axes (left–right [LR], superoinferior [SI], and AP). The overall mean values and standard deviations (SDs), along with random and systematic SDs, were computed for these translations and rotations. Results For 282 intrafraction translational displacements, the random SD was 0.8 mm (systematic SD, 0.2) in the LR, 1.0 mm (systematic SD, 0.4) in the SI, and 1.4 mm (systematic SD, 0.7) in the AP axes. The analysis of 348 interfraction rotations revealed random SDs of 6.1° (systematic SD, 5.6°) around the LR axis, 2.8° (systematic SD, 2.4°) around the SI axis, and 2.0° (systematic SD, 2.2°) around the AP axis. The intrafraction rotational motion observed during 44 fractions had a random SD of 1.8° (systematic SD, 1.0°) around the LR, 1.1° (systematic SD, 0.8°) around the SI, and 0.6° (systematic SD, 0.3°) around the AP axis. Conclusion The interfraction rotations observed were more important than those reported in previous studies. Intrafraction motion was generally smaller in magnitude than interfraction motion. However, the intrafraction rotations and translations of the prostate should be taken into account when designing planning target volume margins because their magnitudes are not negligible.

Deformable image registration for the use of magnetic resonance spectroscopy in prostate treatment planning

Purpose There is now convincing evidence that prostate cancer cells lack the ability to produce and accumulate citrate. Using magnetic resonance spectroscopy imaging (MRSI), regions of absent or low citrate concentration in the prostate can be visualized at a resolution of a few mm. This new advancement provides not only a tool for early diagnosis and screening but also the opportunity for preferential targeting of radiation to regions of high tumor burden in the prostate. The differences in the shape and location of the prostate between MRSI imaging and treatment have been the major obstacle in integrating MRSI in radiation therapy treatment planning. The purpose of this study is to develop a reliable method for deforming the prostate and surrounding regions from the geometry of MRSI imaging to the geometry of treatment planning, so that the regions of high tumor burden identified by the MRSI study can be faithfully transferred to the images used for treatment planning. Methods and materials Magnetic resonance spectroscopy imaging studies have been performed on 2 prostate cancer patients using a commercial MRSI system with an endorectal coil and coupling balloon. At the end of each study, we also acquired the MRI of the pelvic region at both the deformed state where the prostate is distorted by the endorectal balloon and the resting state with the endorectal balloon deflated and removed. The task is to find a three-dimensional matrix of transformation vectors for all volume elements that links the two image sets. We have implemented an optimization method to iteratively optimize the transformation vectors using a Newton-Ralphson algorithm. The objective function is based on the mutual information. The distorted images using the transformation vectors are compared with the images acquired at the resting conditions. Results and discussion The algorithm is capable of performing the registration automatically without the need for intervention. It does not require manual contouring of the organs. By applying the algorithm to multiple image sets of different patients, we found a good agreement between the images transformed from those acquired at the deformed state and those acquired at resting conditions. The computation time required for achieving the registration is in the range of a half-hour (for image size: 256 pixels × 256 pixels × 25 slices). However, the space of registration can be restricted to speed up the process. Conclusion In this article, we described a three-dimensional deformable image registration method to automatically transform images from the deformed imaging state to resting state. Our examples show that this method is feasible and useful to the treatment planning system.

Automated performance evaluation of range image segmentation algorithms.

Previous performance evaluation of range image segmentation algorithms has depended on manual tuning of algorithm parameters, and has lacked a basis for a test of the significance of differences between algorithms. We present an automated framework for evaluating the performance of range image segmentation algorithms. Automated tuning of algorithm parameters in this framework results in performance as good as that previously obtained with careful manual tuning by the algorithm developers. Use of multiple training and test sets of images provides the basis for a test of the significance of performance differences between algorithms. The framework implementation includes range images, ground truth overlays, program source code, and shell scripts. This framework should a) make it possible to objectively and reliably compare the performance of range image segmentation algorithms; b) allow informed experimental feedback for the design of improved segmentation algorithms. The framework is demonstrated using range images, but in principle it could be used to evaluate region segmentation algorithms for any type of image.

Determination of 3D positions of pacemaker leads from biplane angiographic sequences.

In vitro and in vivo analyses of stress on pacemaker leads and their components during the heart cycle have become especially important because of incidences of failure of some of these mechanical components. For stress analyses, the three-dimensional (3D) position, shape, and motion of the pacemaker leads must be known accurately at each time point during the cardiac cycle. We have developed a method for determination of the in vivo 3D positions of pacemaker leads during the entire heart cycle. Sequences of biplane images of patients with pacemakers were obtained at 30 frames/s for each projection. The sequences usually included at least two heart cycles. After patient imaging, biplane images of a calibration object were obtained from which the biplane imaging geometry was determined. The centerlines of the leads and unique, identifiable points on the attached electrodes were indicated manually for all acquired images. Temporal interpolation of the lead and electrode data was performed so that the temporal nonsynchronicity of the image acquisition was overcome. Epipolar lines, generated from the calculated geometry, were employed to identify corresponding points along the leads in the pairs of biplane images for each time point. The 3D positions of the lead and electrodes were calculated from the known geometry and from the identified corresponding points in the images. Using multiple image sets obtained with the calibration object at various orientations, the precision of the calculated rotation matrix and of the translation vector defining the imaging geometry was found to be approximately 0.7 degree and 1%, respectively. The 3D positions were reproducible to within 2 mm, with the error lying primarily along the axis between the focal spot and the imaging plane. Using data obtained by temporally downsampling to 15 frames/s, the interpolated data were found to lie within approximately 2 mm of the true position for most of the heart cycle. These results indicate that, with this technique, one can reliably determine pacemaker lead positions throughout the heart cycle, and thereby it will provide the basis for stress analysis on pacemaker leads.

Multimedia image display: a view to the future.

The development of multimedia image review stations offers the cardiologist the ability to review multiple image sets in a single setting rather than moving from one review station to another. To accomplish this goal there are three essential barriers: 1) memory requirements for digital images; 2) standardization of data format; and 3) practical user interfaces. Image compression algorithms can be used to overcome the digital barrier; however, these must be used cautiously so as to not adversely affect image quality. The lack of standardization is being addressed by the implementation of the Digital Communication in Medicine (DICOM) standard, and practical user interfaces are being developed every day with the widespread implementation of World Wide Web technology. These solutions will allow the clinician to review all of a patient's image data in one location, such as in the office, on the nursing unit, or at home.

Effects of Sampling Parameters on Principal Components Analysis of Raman Line Images

Raman images of the distribution of materials in a sample prepared from 10-μm-diameter polystyrene spheres embedded in epoxy were reconstructed from sets of line-scanned images, with the use of univariate and multivariate processing of the spectral data. Multiple sets of microscopic Raman spectral line images were acquired by using line-focused illumination with a cylindrical lens, a motorized translation stage to move the sample perpendicular to the illumination line, and a holographic imaging spectrograph equipped with a 2D charge-coupled device (CCD) detector. Repeat sets of data were obtained at different spectrometer slit width settings and different magnification. The raw spectral data were processed by using both a simple univariate method (single-band integration) and a more sophisticated multivariate method [principal components analysis (PCA) with eigenvector rotation] to generate 2D Raman images representing spatial distribution of the individual polymeric constituents. The repeat data sets were compared to ascertain the effects of the sampling parameters on the PCA method. The results indicated that spectrometer slit width and magnification affect the sampling depth and spatial resolution, but have little effect on the PCA. Moreover, digital sampling (i.e., number of PCA wavelength variables) could be significantly reduced with little or no degradation of the PCA-generated images, particularly if key bands were represented.

A digital brain atlas and its application to the visceral neuraxis

We describe the details and application of a digital brain atlas for the comparison and integration of graphical neurobiological data. The atlas consists of multiple sets of high-resolution video images acquired from histological tissue sections representing a 3-dimensional (3D) volume of an exemplar rat brain. Through an interactive graphical interface running on a standard computer workstation, experimental data is brought into register with the atlas. Once in the atlas, coordinate reference frame data can be compared, analyzed, and visualized in 3 dimensions. We demonstrate the validity and usefulness of the digital brain atlas with a series of results on the visceral neuraxis in the rat.

Design of ternary correlation filters to reduce probability of error

The problem of designing ternary phase and amplitude filters (TPAF's) that reduce the probability of image misclassification for a two-class image set is studied. The Fisher ratio is used as a measure of the correct classification rate, and an attempt is made to maximize this quantity in the filter designs. Given the nonanalytical nature of the design problem, a simulated annealing optimization technique is employed. Computer simulation results are presented for several cases including single in-class and out-of-class image sets and multiple image sets corresponding to the design of synthetic discriminant function filters. Significant improvements are found in expected rates of correct classification in comparison to binary phase-only filters and other TPAF designs. Approaches to accelerate the filter design process are also discussed.

A fast 3D-imaging technique for performing dynamic Gd-enhanced MRI of breast lesions

The characterization of breast lesions by their Gd-enhancement profiles has been proposed as a method for differentiating benign from malignant breast lesions. The limitations of dynamic contrast enhanced 2D imaging of the breast are the low number of slices that can be acquired, and the need to know the location of the lesion a priori to correctly select the noncontiguous 2D slice locations. These problems are exacerbated when multi-focal disease is present but not anticipated. Standard fast 3D gradient-echo imaging has a variable delay between successive acquisitions. We have developed a fast 3D gradient-echo imaging technique for dynamic Gd-DTPA enhanced breast imaging which obtains multiple 3D image sets of 32 contiguous images at 44 s intervals without an interscan delay time. This rapid 3D imaging technique achieves good temporal resolution and reduces patient motion between pre- and postcontrast images while covering a much larger portion of the breast and eliminating the need for a priori knowledge concerning the location of the lesion(s) when performing Gd-enhanced dynamic MR imaging.

A high performance Interactive Image SpreadSheet (IISS)

New methods are needed for visualizing, interpreting, comparing, organizing, and analyzing immense multispectral satellite datasets. The traditional numerical spreadsheet paradigm has been extended to develop a new scientific visualization approach for processing multisensor image datasets interactively. Exploring the advantages of extending the powerful spreadsheet style of computation to multiple sets of images and organizing image processing tasks is the objective of the Interactive Image SpreadSheet (IISS) project at Goddard Space Flight Center. In the IISS each cell can display any portion of an original or calculated image, a projection of a multidimensional dataset such as a 3‐D surface, a glyph (graphic symbol) representing an image, digitized maps, digital terrain models, graphs, or vector drawings. The term image is used in a general sense to refer to any 2‐D multisource dataset. The IISS typically contains an array of image cells of arbitrary size each of which can contain one or more frames (images). The user can scroll or page through this multidimensional cube of frames along any dimension. The IISS emphasizes an immediate visual approach to interacting with data. A unique capability that the IISS provides are the highly interactive browsing tools, accessible through a graphical user interface, for effectively inspecting large sets of image arrays using synchronized cell level operations such as zoom, roam, animation, and function execution.The IISS combines the quantitative aspects of a numerical spreadsheet with powerful visualization tools to enable an investigator to easily experiment with various combinations of multispectral image data using a library of standard algorithms and to interactively develop custom algorithms. Remotely sensed datasets from multispectral instruments on satellites such as GOES, NOAA, Nimbus, DMSP, and Landsat have been used to develop and evaluate the functionality of the IISS. Formula expressions for creating color composites, implementing image enhancements, calculating vegetation indices, viewing perspective and stereo imagery have been developed using multispectral data. The IISS can also be used in a variety of imaging disciplines that routinely need to organize and manipulate large volumes of visual data including numerical simulation data, observational field data, astronomical imaging, biomedical imaging, computer vision and manufacturing robotics, business document imaging, and multimedia. The practical realization of the computationally challenging IISS project relies on the fact that personal superworkstations have become inexpensive enough that one can extend the interactive scalar spreadsheet concept to the image processing field. The hardware features that make this possible include multiple processors, large amounts of general purpose memory, high‐performance data buses, large mass storage, and pipelined or other advanced architectures for graphics and image operations. The need for increasingly more interactive imaging and visualization applications using high definition displays in collaborative environments will continue to drive the demand for more powerful hardware features and information network capabilities that are widely accessible.

Reconstruction of intracellular and extracellular networks from stereo IVEM images using computer graphic methods

High and intermediate voltage electron microscopes can be used to get good images of many types of thick biological specimens, including sections of embedded material up to several micrometers thick, whole cells, or isolated cellular components mounted on support films, stereoscopic pairs of images can be obtained by tilting the specimen between exposures, and this requires only relatively small tilt angles, often less than 15°. With a high tilt specimen holder, images covering a wider range of tilt angles can be obtained, increasing the amount of 3D information. Computer graphic methods for extracting quantitative 3D information from such images have been under development and use in this laboratory for several years.Many cellular structures take the form of networks. Examples that we study are transverse tubular networks in muscle, fibrin clots, and the microtrabecular lattice of the cytoplasm. We perform quantitative, 3D reconstruction and analysis of such networks from multiple stereoscopic image sets obtained using a JEOL 4000-EX TEM.