Classifications

• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T11/00—2D [Two Dimensional] image generation
  • G06T11/003—Reconstruction from projections, e.g. tomography
  • G06T11/005—Specific pre-processing for tomographic reconstruction, e.g. calibration, source positioning, rebinning, scatter correction, retrospective gating
• • A—HUMAN NECESSITIES
  • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
  • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
  • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
  • A61B6/50—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment specially adapted for specific body parts; specially adapted for specific clinical applications
  • A61B6/508—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment specially adapted for specific body parts; specially adapted for specific clinical applications for non-human patients

Definitions

• the present invention relates generally to radiation therapy equipment for the
• treatment of tumors and more particularly to methods for reconstructing incomplete patient data for radiation therapy and treatment verification.
• the amount of radiation and its placement must be accurately controlled to ensure both that the tumor receives sufficient radiation to be destroyed, and that the
• External source radiation therapy uses a radiation source that is external to the
• the external source is normally collimated to direct a
• the tumor will be treated from multiple angles with the intensity and shape of the beam adjusted appropriately.
• energy radiation may be x-rays or electrons from a linear accelerator in the range of 2-25
• CT tomography
• This LFON can cause visibility problems with the images, images with artifacts, images with distorted values, and affect applications that use these images, including dose calculations, delivery verification, deformable patient
• Intensity modulated radiation therapy uses intensity modulated radiation beams
• the radiation field is "sculpted" to match the shape of the cancerous tissue and to keep the dose of radiation to healthy tissue near the cancer low.
• a radiation treatment plan may be based on a CT image of the patient.
• a CT image is produced by a mathematical reconstruction of many projection images obtained at different angles about the patient, h a typical CT
• the projections are one-dimensional line profiles indicating the attenuation of the
• the actual CT data is held in sinogram space as a matrix
• each row represents a gantry position, a gantry angle, a ray angle or the like (a first sinogram dimension); each column represents a detector number, a detector distance,
• a third detector angle a detector angle, a ray position, or the like (a second sinogram dimension).
• sinogram dimension is commonly used with multi-row or volumetric detectors
• the matrix of data obtained in a CT image can be
• a physician views the cancerous areas on a CT
• a computer program selects the beam angles and intensities after the physician
• planning CT images are used to create a three-dimensional (3-
• voxels which are defined as volumetric pixels. Each voxel is then assigned a
• the planning CT image of a patient is acquired substantially before the
• the planning CT image can undermine the conformality of the radiation delivery.
• LFON image as shown in FIG. 3, which shows only a portion of the image shown in FIG. 2.
• the FON or image data sets may also be intentionally limited by modulated treatment data or region-of-interest tomography (ROIT) involving reconstruction of treatment data, intentionally only delivered to a specific region(s).
• ROI region-of-interest tomography
• FIG. 3 not only is there a LFON, but the data around the edges contains significant artifacts so that the image has an irregular border and internal values that are distorted.
• the LFON of radiotherapy images creates problems of impaired visibility and degraded dose calculations.
• the most common reasons for impaired visibility are the limited field size of the MLC attached to the linear accelerator and the limited detector size. These limitations prevent the CT imaging system from collecting complete FON data for all sizes of patients at all sites.
• the problem of degraded dose calculations is caused by distorted electron densities and the loss of peripheral information for attenuation and scatter from the LFON images. This distortion of image values and loss of peripheral information can likewise affect other applications that utilize these images.
• the present invention relates to methods by which an incomplete CT patient data
• the present invention provides methods for utilizing complete planning CT data for reconstruction of incomplete CT data with
• the method includes the steps of
• the aligned or "fused” image is reprojected as a sinogram.
• This reprojected sinogram is compared to either the first or second sinogram to determine what data exists beyond the scope of the first or second sinogram.
• This additional data is added to the sinogram to which the reprojected sinogram was compared to obtain an augmented sinogram
• the augmented sinogram is then converted or reconstructed to an image, referred to as a fusion-aligned reprojection (FAR) image.
• FAR fusion-aligned reprojection
• the method of the first embodiment of the present invention is advantageous in that the availability of only one limited data sinogram/image will not affect the ability to perform accurate delivery verification, dose reconstruction, patient setup or the like.
• the previously taken complete image or "second image” is fused, or aligned, to the limited data image or "first image.”
• the sinogram representing the fused image is compared to the limited data sinogram, and the augmented limited data sinogram is prepared therefrom. From the augmented limited data sinogram the FAR image is obtained.
• the FAR image is used to accurately apply radiation to the treatment area, which may be positioned differently or contain anatomical changes as compared to the previously obtained complete image.
• FAR compensates for limited data radiotherapy images by enhancing the conspicuity of structures in treatment images, improving electron density values, and estimating a complete representation of the patient.
• FAR combines the LFOV data with prior information about the patient including CT images used for planning the radiotherapy.
• the method of the first embodiment includes aligning or "fusing" the LFON image and the planning image, converting the images into "sinogram space", merging the images in sinogram space, and reconstructing the images from sinograms into normal images.
• a key step of the FAR method is "fusion" or alignment of the planning image with the LFON image. However, if a patient's treatment position is close to the planning position, explicit fusion under the FAR method may not be necessary. Instead, an implicit fusion may be adequate if the normal setup error is sufficiently small.
• NEAR normal-error-aligned reprojection
• NEAR A benefit of NEAR is that it may enable an iterative (two or more) variation of
• NEAR2FAR FAR
• NEAR2FAR FAR
• NEAR can be followed by FAR iterations, or FAR can be tried multiple times with different registration results.
• the quantitatively improved voxel values in the FON might enable an explicit fusion with the planning image, and a FAR image could be generated.
• NEAR and NEAR2FAR may be particularly beneficial when a LFON causes severe quantitative and qualitative degradation of the images, whether because of a large patient, a small detector or MLC, or because a ROIT strategy is being
• NEAR may also be quicker than FAR, as no time is required to do an explicit
• NEAR, FAR, and NEAR2FAR utilize planning CT data or other images as
• CT images e.g. megavoltage CT acquired at different energies than planning CT images.
• FAR, NEAR and NEAR2FAR may also be used for multi-modality imaging
• image values they may be correctable, or they may show the patient boundary, which
• the methods of the present invention improve the data by aligning the LFON and
• FAR can be implemented using the implicit fusion of NEAR.
• NEAR and/or FAR optional iterative use of NEAR and/or FAR is also possible, as are applications of NEAR and FAR to dose calculations and the compensation of LFON online megavoltage CT
• FIG. 1 an example of a sinogram obtained from the CT image of a patient
• FIG. 2 is an example of a planning image of a patient obtained from a sinogram
• FIG. 3 is an example of a LFON treatment image of a patient
• FIG. 4 is a flow diagram showing the steps involved in creating a FAR treatment image in accordance with a first embodiment of the present invention
• FIG. 5 is a schematic representation of a full image scan of a patient
• FIG. 6 is a schematic representation of FIG. 5 with illustrative "anatomical"
• FIG. 7 demonstrates how the full image of FIG. 5 is aligned to the limited image
• FIG. 8 is a schematic representation of a FAR image
• FIG. 9 is a schematic representation of a full image corresponding to the image of
• FIG. 6; FIG. 10 shows a schematic representation of the actual alignment or "fusion" of
• FIG. 11 is a reconstructed FAR image of FIGS. 2 and 3 aligned in accordance
• FIG. 12 shows a comparison of a planning image, a LFON treatment image, an
• FIG. 13 shows an example FAR sinogram obtained by merging a LFON online
• FIG. 14 shows a comparison of radiotherapy dose calculations for a LFON image
• FIG. 15 A is a flow diagram showing the steps involved in creating an aligned
• FIG. 15B is a flow diagram showing the steps involved in creating an aligned
• FIG. 15C is a flow diagram showing the steps involved in creating an aligned
• FIG. 16 shows examples of LFON images, NEAR images, and FAR images for
• FIG. 17 shows a LFON reconstruction for a 10.5 cm FON, a NEAR
• FIG. 18 shows a comparison of radiotherapy dose calculations for complete FON
• FIG. 19 shows canine CT images from a kilo voltage CT scanner, a megavoltage
• FIG. 1 is an example of a sinogram 10 obtained
• FIG. 2 is an example of a planning CT image obtained
• FIG. 3 is an example of a LFOV
• FIG. 4 represents the first embodiment
• the process begins by obtaining a limited data sinogram 50 typically representing the treatment area from a patient.
• the limited data sinogram 50 is preferably obtained near the time that the patient is receiving his or
• the limited data sinogram 50 is
• FIG. 3 contains a significant
• the methods of the present invention can be applied to images of any part of the body, or be used in other applications, such as veterinary
• image 12 shown by way of example in FIG. 2 as image 12, and represented schematically in FIG. 5 as object 154, is typically obtained prior to obtaining the limited data image 52, image 14
• FIG. 2 there are often inherent differences between the location of certain organs and/or tissue due to motion caused by normal bodily functions as the patient travels from the
• weight loss or growth of certain tissue can also occur.
• image 52, image 14 of FIG. 3 need not be from a CT scanner or imager, and that this
• MRI magnetic resonance imaging
• positron emission positron emission
• PET PET
• SPECT single photon emission tomography
• FIGS. 2 and 3 intestinal gas 16 is shown in FIG. 3, thereby displacing
• object 154 is composed of diagonals 158a and 160a and an inclusion 161a, within a frame 162a. Limited object
• limited object 156 is fused with complete object 154 so that statistically, there is optimal
• FIG. 7 shows how the orientation of object
• FIG. 10 shows diagonal 160c as the
• FAR is not specific to the registration technique. It could be through automatic,
• Image registration or fusion may be
• MI mutual information
• EFF Extracted Feature Fusion
• the bones can in effect be extracted from each image
• diagonal 160a and frame 162 may represent bone or tissue that remains
• time is generally proportional to the number of points selected, so reducing that number from the size of the entire three-dimensional image set to a subset of points meeting
• image registration techniques include manual fusion, alignment using
• geometric features e.g., surfaces
• gradient methods e.g., gradient methods
• voxel-similarity techniques e.g., voxel-similarity techniques
• the aligned or transformed complete image 56 is
• the data for sinogram 58 is once again in a matrix wherein each row represents an angle, and each column represents a distance.
• the reprojected sinogram 58 is compared to the data matrix for limited data sinogram 50
• the augmented limited data sinogram 60 is reconstructed to a FAR image 62 that is an approximation of what the complete image
• image 62 is represented schematically in FIG. 8.
• Frame 162a is the same as in FIG. 5,
• regions 170 of diagonal 158d are not the same as diagonal 158c is not critical to the
• FIG. 11 represents a reconstructed FAR image obtained by combining the
• contouring identifying target regions and sensitive structures, either
• FIG. 12 shows the comparison of a planning image 12', which is equivalent to the
• a LFOV treatment image 14' which is equivalent to the
• treatment image 18 and 18' is substantially similar to the ideal treatment image 20, except for the slight artifact rings 180 and 180' that do not impair the conspicuity of the
• FIG. 4 The completion process of FIG. 4 can be seen in sinogram space in FIG. 13.
• FIG. 13 The completion process of FIG. 4 can be seen in sinogram space in FIG. 13.
• FIG. 13 shows an example FAR sinogram 26 obtained by merging a LFOV sinogram 22 with an aligned planning sinogram 24.
• the truncated limited data sinogram 22 is shown in
• FIG. 13 A The missing data from the LFOV sinogram 22 is estimated from the aligned
• the resulting FAR sinogram 26 shown in FIG. 13C estimates the missing data from the aligned planning sinogram 24 of FIG. 13B.
• FIG. 14 shows a comparison of radiotherapy dose calculations for a LFOV image
• the LFOV image 28 results in substantial dose calculation
• volume histogram 28 shows both overestimation and underestimation between the
• FIGS. 15A, 15B, and 15C represent different embodiments of methods involved in creating an aligned-reprojection image from a limited data image or sinogram and a
• FIG. 15 A a FAR, NEAR, or
• NEAR2FAR image is created by obtaining a limited data sinogram 32 A representing the
• the limited data sinogram is reconstructed to a limited
• a complete planning image 36A of the same patient is typically
• the aligned complete planning image 38 A is reprojected as a sinogram 40A.
• the reprojected sinogram of the aligned planning image 40A is compared to the limited data sinogram 32A.
• the missing sinogram data from the reprojected sinogram 40A is added or merged with the limited data sinogram 32A to create an augmented limited data sinogram 42A.
• the augmented limited data sinogram 42A is
• the aligned-reprojection image may be fed back to the limited data image 34A for a
• image e.g., complete FOV planning image or limited online FOV
• planning image is used to estimate the missing data from the limited data image.
• the complete planning image could be realigned to the LFOV image creating an aligned planning image, reproject the aligned planning image to a sinogram, augment or
• the LFOV image could be realigned to the complete planning image creating an aligned LFOV image, reproject the
• the method of realigning the image and reprojecting it into a sinogram can be mathematically streamlined as shown in FIGS. 15B and 15C.
• the relative alignment between the complete planning image and the limited data image is determined. Then, instead of realigning the complete planning image to the limited data
• the aligned planning sinogram is then used to estimate the missing data
• FIG. 15B illustrates another embodiment of a method for creating an aligned-
• the inputs to the process are a complete planning image 36B or complete
• the LFOV sinogram 32B is
• the complete planning image 36B is
• the aligned planning image 40B is used to estimate the data missing from the LFOV sinogram 32B.
• the limited data sinogram 32B is merged with the aligned planning image sinogram 40B, resulting in an augmented limited data sinogram 42B.
• This augmented limited data sinogram 42B is reconstructed into an aligned-reprojection image 44B.
• the aligned- reprojection image may supersede the original limited data image 34B for a multiple iteration process (NEAR2FAR).
• FIG. 15C illustrates yet another embodiment of the present invention for creating an aligned-reprojection image from a limited data sinogram and a complete planning image or sinogram.
• the inputs to the process are a limited data sinogram 32C and either an optional complete planning image 36C or most preferably a complete planning sinogram 108C. If the process starts with a complete planning image 36C as one of the inputs, then that image is reprojected to sinogram space to yield a complete planning sinogram 108C.
• the limited sinogram 32C is fused in sinogram space (explicit (FAR) or implicit (NEAR)) with the complete planning sinogram 108C.
• the next step involves realigning the complete planning sinogram 108C, or realigning and reprojecting the complete planning image 36C using the same fusion result.
• the resulting aligned plaiming image sinogram 40C is merged with the limited data sinogram 32C to create an augmented limited data sinogram 42C.
• the augmented limited data sinogram 42C is then reconstructed into an aligned-reprojection image 44B.
• the fusions are performed in sinogram-space as the limited data sinogram 32C is fused (implicit or explicit) to the complete data sinogram 108C, unlike the embodiments of FIGS. 15A and 15B that use image fusion.
• the realigned planning sinogram 40C can be created by realigning sinogram 108C, or by realigning planning image 36C and reprojecting into sinogram space. The process is then the same for each case.
• the aligned planning sinogram 40C is merged with the limited data sinogram 32C to create an augmented limited data sinogram 42C.
• the augmented limited data sinogram 42C is then reconstructed into an aligned-reprojection image 44B.
• FIG. 16 shows representative images from a planning CT image 66 and the corresponding online image 64.
• the contours 65 for the planning images are shown in black, while the contours 67 for the online images are shown in white.
• Three different LFOV images 68, 70, 72, NEAR images 74, 76, 78, and FAR images 80, 82, 84 for field- of-view sizes of 38.6, 29.3, and 19.9 cm are shown based upon the online image 64.
• the FOV decreases, the artifacts become more severe in the LFOV images 68, 70, 72, while the NEAR 74, 76, 78 and FAR images 80, 82, 84 are less affected.
• NEAR and FAR are representative of how NEAR and FAR can utilize available information to qualitatively improve the reconstructions for a range of FOV sizes. In this particular case, there is little visual difference between the NEAR and FAR images. The similarity of NEAR and FAR images can occur for several reasons. Where the normal setup error is small, the explicit fusion will generally not improve much upon the normal error, or because the anatomical differences between the planning CT image 66 and the online image 64 are a more significant factor than the alignment between those images, there will also be little improvement. NEAR and FAR can utilize available information to qualitatively improve the
• FAR can produce images that are quantitatively closer to the complete FOV online image
• FAR may not be possible if the distortion of image values preclude a successful fusion. In this case,
• a NEAR image is created, and by fusing or aligning the NEAR image to the planning CT
• NEAR2FAR image is generated, further reducing artifacts and improving
• FIG. 17 shows a LFOV reconstruction 86 for a 10.5 cm FOV, a NEAR
• NEAR2FAR NEAR2FAR
• FIG. 18 shows a comparison of radiotherapy dose calculations for complete FOV
• Histogram are based upon the known contours from the complete FOV online image.
• the LFOV dose calculation overestimates the prostate dose by approximately 15%, and the rectum and bladder doses have areas of both overestimation and underestimation.
• the dose distributions calculated using NEAR and NEAR2FAR produce DVH's indistinguishable from the full FOV dose calculation.
• FIG. 19 shows canine CT images from a kilovoltage CT scanner 98, a megavoltage CT scanner 100, a LFOV version of the megavoltage image 102, and a FAR reconstruction 104 from the LFOV data augmented with planning CT data.
• these data sets were not only acquired on different CT systems but at different energies, requiring that FAR combine megavoltage and kilovoltage data.
• the resulting FAR image 104 includes slight artifacts 106 that can result from this method. However, such artifacts 106 are insignificant because they do not impair the conspicuity of the important structures in the FOV, nor are they noticeably detrimental to dose calculations or other processes that utilize these images.
• the methods of the present invention may be used for purposes beyond radiotherapy in cases where potentially imperfect prior information is available. While the present description has primarily disclosed use of prior information in the form of a planning CT, it is feasible to apply NEAR and FAR to multi-modality images, such as creating a FAR image by combining an online CT (megavoltage or kilovoltage) data set with a planning MRI image. In such cases, the MRI or other- modality image needs to be converted to values compatible with the LFOV data set. A complex mapping of values will provide the best results, but even using the alternate modality image to describe the patient's outer contour and using a water-equivalency
• FAR can also combine megavoltage and kilovoltage CT
• ROI region-of-interest tomography
• the limited data is not necessarily LFOV, but can also be more complex patterns of missing data, such as modulated treatment data.
• NEAR and FAR may also be extensible to other types of limited data situations, such as limited slice or limited-projection images.

Landscapes

• Physics & Mathematics (AREA)
• General Physics & Mathematics (AREA)
• Engineering & Computer Science (AREA)
• Theoretical Computer Science (AREA)
• Apparatus For Radiation Diagnosis (AREA)
• Magnetic Resonance Imaging Apparatus (AREA)
• Image Generation (AREA)

Applications Claiming Priority (3)

Publications (2)

Family

ID=29732442

Family Applications (1)

Country Status (5)

Cited By (1)

Families Citing this family (11)

Citations (3)

Family Cites Families (3)

• 2003
  • 2003-06-10 JP JP2004512066A patent/JP2005529658A/ja active Pending
  • 2003-06-10 WO PCT/US2003/018229 patent/WO2003105069A1/en active Application Filing
  • 2003-06-10 EP EP03757458A patent/EP1514228A4/de not_active Withdrawn
  • 2003-06-10 CA CA002489157A patent/CA2489157A1/en not_active Abandoned
  • 2003-06-10 AU AU2003243469A patent/AU2003243469A1/en not_active Abandoned

Patent Citations (3)

Non-Patent Citations (3)

Also Published As

Similar Documents

Legal Events