Patent Grant
8036730
The present invention relates to imaging techniques and apparatus for performing such techniques.
In magnetic resonance imaging (“MRI”), a strong, uniform magnetic field is applied to the region of the patient to be imaged. Radio frequency (“RF”) energy is applied to this region of the patient by a transmitter and antenna. The RF energy excites atomic nuclei within the patient's tissues. The excited nuclei spin at a rate dependent upon the magnetic field. As they spin, they emit faint RF signals, referred to herein as magnetic resonance signals. By applying small magnetic field gradients so that the magnitude of the magnetic field varies with location within the patient's body, the magnetic resonance phenomenon can be limited to only a particular region or “slice” of the patient's body, so that all of the magnetic resonance signals come from that slice. Moreover, by applying additional magnetic field gradients, the frequency and phase of the magnetic resonance signals from different locations within the slice can be made to vary in a predictable manner depending upon the position within the slice. Stated another way, the magnetic resonance signals are “spatially encoded,” so that it is possible to distinguish between signals from different parts of a slice.
If this process is repeated numerous times to elicit signals using different gradients, it is possible to derive a set of information which indicates one or more characteristics of magnetic resonance signals from particular locations within the patient's body. Such a set of information is referred herein to as an image data set. Because the characteristics of the magnetic resonance signals vary with the concentration of different chemical substances and other chemical characteristics of the tissues, different tissues provide different magnetic resonance signal characteristics. When a magnetic resonance signal image data set is displayed in a visual format, such as on a computer screen or printed image, the information forms a picture of the structures within the patient's body, with different tissues having different intensities or colors.
Typically, a magnetic resonance image data set is stored as a set of individual data elements. The data in each element represents one or more characteristics of magnetic resonance signals from a small volume element or “voxel.” For example, the map can be stored as a three-dimensional array of data elements, the dimensions of the array corresponding to three-dimensional space. Data elements corresponding to a given plane in three-dimensional space can be selected for display in a two-dimensional picture such as a screen display or printed image. Each small area element on the surface of the picture, commonly referred to as a “pixel,” is assigned an intensity or color value based on the numerical values of the data element for the corresponding voxel.
MRI has been widely adopted in the medical arts. Because MRI does not use X-rays or other ionizing radiation, it offers safety advantages over techniques such as conventional X-ray imaging, fluoroscopy and CAT imaging.
Moreover, MRI allows visualization of tissues which are difficult or impossible to depict using other techniques. Magnetic resonance imaging can show abnormal tissues in contrast to surrounding normal tissues. For example, as disclosed in U.S. Pat. No. 3,789,832 of Raymond V. Damadian, magnetic resonance signals from malignant tumors have a characteristic referred to as the spin-lattice relaxation time or “T1” different from the T1 of normal tissues. If a magnetic resonance image is taken so that the data in each data element depends at least in part on the T1 of the tissue at the corresponding location, a picture showing malignant tumor tissue in contrast to normal tissue can be displayed.
MRI is also particularly useful in imaging the spine. MRI can depict the vertebrae in conjunction with related tissues such as the lamina or “discs,” as well as nerves, muscles and other neighboring tissues.
In magnetic resonance angiography, the magnetic field gradients applied during imaging, and the characteristics of the magnetic resonance signals which are translated into the image, are selected according to principles well-known in the art so that the data in voxels within arteries differs from the data for voxels in other structures, so that the arteries can be depicted in contrasting color or density to surrounding tissues. For example, arterial blood has a significant velocity and the surrounding tissues are nearly stationary. A so-called “motion-sensitive” MRI technique can be used so that a characteristic of the magnetic resonance signals from each voxel depends on the velocity of matter within the voxel. Magnetic resonance angiography yields images directly analogous to those obtained by conventional angiography, without the need for X-ray exposure. In some cases, MRI angiography can be performed without injection of a contrast medium. Moreover, MRI angiography can provide three-dimensional imaging information, so that images from any desired perspective can be displayed.
However, magnetic resonance imaging procedures have suffered from significant limitations. Conventional MRI equipment requires the patient in a supine position on a horizontal bed which fits with the patient receiving space of the static field magnet. Some medical conditions have effects which change with posture. For example, a spinal disc may impinge on a nerve or other surrounding structure only when the patient is in an upright posture so that the disk is compressed by the patient's weight. Various proposals have been advanced to allow MRI procedures to be performed on patients in a posture other than the conventional supplying of posture. For example, Japanese published Patent Application No. 1-242056 published Sep. 27, 1989 depicts a magnetic resonance imaging unit with a tilting bed for supporting the patient in a supine position or in a standing position. Yoshida, U.S. Pat. No. 5,008,624 depicts a magnetic resonance imaging instrument with movable static field magnet in conjunction with a patient carrier which supports the patient in “various postures.” Palkovich et al., U.S. Pat. No. 5,779,637 discloses a system in which the patient lies supine within the static field magnet during one imaging procedure. The entire system, including the static field magnet and the patient can be pivoted so as to swing the magnet, the patient bed and the patient as a unit to a different position in which the patient bed extends vertically and the patient in an upright posture. A further image is taken in this position. None of these systems have been widely adopted.
Copending, commonly assigned U.S. patent application Ser. No. 09/718,946, filed Nov. 22, 2000 (“the '946 application”), now U.S. Pat. No. 6,677,753, the disclosure of which is hereby incorporated by reference herein and copending commonly assigned U.S. patent application Ser. No. 09/789,460 (“the '460 application”), now U.S. Pat. No. 6,414,490, the disclosure of which is also incorporated by reference herein, describe additional MRI magnet structures and patient handling devices as well as additional imaging methods. As disclosed for example in certain embodiments of the '946 application, a patient support such as a bed which can both tilt and elevate can be used in conjunction with a static field magnet to allow imaging of a patient in various orientations and to position various portions of the patient's anatomy in the appropriate location relative to the magnet for imaging. Discussion of the '946 and '460 applications in this background section of the present application should not be taken as an admission that the same constitute legally available prior art with respect to the present invention.
Most commonly, pictures derived from MRI images are read by a physician visually examining the picture to diagnose disease which may be present or to evaluate the progress of a known disease. Such evaluation may involve, for example, a mental comparison by the physician with his or her memory of pictures the physician has previously seen of normal and other diseased patients or pictures taken in the past of the same patient. This task requires careful examination and considerable professional judgment. Even with the capabilities achievable in MRI imaging, it is not always easy to spot disease states or changes in the patient's condition. Lemelson et al., U.S. Pat. No. 5,878,746 proposes an automated process in which the computer examines a new image to extract “features relating to particular disease states” using a pattern recognition technique and stores signals descriptive of these features in a “fact database.” These “feature signals” are compared with similar “feature signals” extracted from previously acquired images and the resulting comparison information is subjected to artificial intelligence rules to provide “a diagnostic assessment.” Perhaps because of the extraordinary difficulty of developing appropriate automated tools for finding features relating to disease states and rules for deriving diagnostic assessments from the compared features, this approach has not been adopted widely, if at all, in practice. Thus, the physician still generally faces the task of visually observing a picture of a patient derived from a particular MRI imaging session and mentally comparing that picture with either a prior picture of the same patient or a mental image of a “normal” anatomy. In this process, the physician typically attempts to discern the outlines of body structures in the picture.
As disclosed in Apicella et al., U.S. Pat. No. 5,273,040, the volume of blood contained within the ventricles of the heart can be determined from an MRI image. To do this, the physician, or an automated system must accurately identify the boundary of the blood-containing ventricle. To enhance the accuracy of an automated process for detecting where the boundary lies, two MRI images taken through the patient's heart in rapid succession, as, for example, during successive heartbeats, are mathematically superimposed and subtracted from one another to yield a “difference image.” Because the two MRI images are taken at two slightly different points in the cardiac cycle, the ventricles will be of slightly different sizes in the two images. Subtraction of the images will eliminate essentially all of the data, leaving only a small border or line having a width corresponding to the change in size between the first and second images. The difference image thus provides a clear line indicating the border of the ventricle which can be recognized in an automated system.
As disclosed in Bani-Hashemi et al., U.S. Pat. No. 5,647,360, digital subtraction angiography or “DSA” typically is performed using a computerized tomography or “CT” x-ray scan. A first set of CT scan data is obtained before injection of a contrast agent into the blood vessels, whereas a second scan is obtained a few minutes later, after injection of the contrast agent. The two data sets are registered with one another and subtracted from one another to yield a difference image which shows the blood vessels in high contrast. The registration procedure used to correlate data elements at corresponding locations within the patient in the two images with one another for subtraction operates by applying a mathematical transformation to one or another of the data sets. Different transformations are used for different parts of the transformed image so as to compensate for warping or non-uniform motion of different parts of the patient. The transformation for each portion of the image is determined using a pattern recognition procedure to match corresponding features shown in each portion of the image with corresponding features shown in the other image. The Bani-Hashemi patent suggests briefly that the registration technique “can be useful in various imaging systems, such as CT, MRI, PET, etc.”
Despite all of the effort devoted in the art heretofore to development of imaging systems and techniques, still further enhancements would be desirable.
One aspect of the invention provides methods of monitoring changes in the body of a patient over time. A method according to this aspect of the invention desirably includes the steps of imaging a part of a patient by magnetic resonance in a first imaging step at a first time so as to provide a first image data set, and imaging at least that part of the patient by magnetic resonance in a second imaging step at a later time. The method desirably includes the step of comparing the two image data sets so as to detect a non-cyclical change in the patient's anatomy, i.e., a change other than a temporary deformation of anatomical structures caused by the cardiac or respiratory cycles. For example, methods according to this aspect of the invention can detect changes due to response to a therapeutic agent or the progression or regression of disease over time. The second imaging step optionally may be performed days, weeks or months later than the first imaging step.
The first and second data may be provided in a common frame of reference and compared by computer so as to provide a comparison image data set. The method desirably includes the step of displaying a visually perceptible image based at least in part on the comparison image data set. For example, the displaying step may include the step of automatically highlighting portions of the visually perceptible image representing at least some regions of the body where the comparison image data set indicates the presence of differences between the first and second image data set. The preferred methods according to this aspect of the present invention can greatly simplify the task of the physician in spotting changes in anatomy over time as, for example, the growth of abnormal tissue. According to a further aspect of the invention, at least one of the magnetic resonance imaging steps is performed so as to acquire data, referred to herein as “tissue data” indicative of the presence or absence of a predetermined characteristic in a body tissue. The predetermined characteristic may be, for example, an abnormal characteristic or a characteristic of a particular tissue type. For example, the magnetic resonance signal characteristics incorporated in the first or second image data sets may include characteristics such as T1 indicative of malignancy. The highlighting step may be performed so as to highlight those portions of the visually perceptible image which represent at least some of the regions of the body where two conditions apply: first, that the tissue data indicates the presence of the predetermined characteristic and second, that the difference data set indicates the presence of differences. In this approach, the displayed image combines the results of two techniques which help to isolate regions of interest for further study.
Preferably, each of the first and second image data sets includes data elements each having coordinates. The data in each data element of each data set represents at least one characteristic of magnetic resonance signals from tissue at a physical location in a frame of reference associated with that data set, such physical location being specified by the coordinates. The method may further include the step of specifying a particular anatomical structure and the comparing step may include registering the first and second data sets with one another in a common frame of reference so that first data elements representing the specified anatomical structure in the first image data set have the same coordinates, in the common frame of reference, as second data elements representing the specified anatomical structure in the second data set.
The comparing step desirably includes comparing the data in one or more groups of first elements with the data in one or more groups of second elements having the same coordinates in the common frame of reference. Each group of data elements may include a single data element having data representing properties of magnetic resonance signals from a single volume element. For example, where each group includes a single data element, the data in a first data element may be subtracted from the data in a second data element having the same coordinates in the common frame of reference, or vice-versa. Other types of comparison, such as division or a determination as to equality or inequality between the data in these elements, may be used. In the alternative, each group of data elements may include a plurality of data elements having data representing properties of magnetic resonance signals from a plurality of contiguous volume elements or from a plurality of non-contiguous volume elements. Where each group of data elements includes a plurality of data elements, a composite property of the group, such as the sum or mean of the data in the various data elements in the group or a measure of variance among the data in the various data elements such as the standard deviation can be determined. In this case, the step of comparing the data in first and second groups of data elements may be performed by comparing the composite properties of these groups to one another.
As further discussed below, it is relatively simple to register the data sets at a particular, readily identifiable anatomical structure. If the data sets are registered at a particular anatomical structure, the comparison image set will accurately reflect meaningful anatomical changes in the vicinity of that anatomical structure.
These and other aspects, features and advantages of the present invention will be more readily apparent from the detailed description of the preferred embodiments set forth below, taken in conjunction with the accompanying drawings.
Although many different types of magnetic resonance imaging apparatus can be used in accordance with the present invention, one apparatus used in one embodiment of the present invention is depicted in
Two cylindrical ferromagnetic poles 56, extend into the interior of the frame from side walls 12 and 13. The poles extend on a common horizontal polar axis 60 and define a patient receiving space 67 between them. The apparatus also includes a source of magnetic flux such as electromagnet coils 57 encircling the poles for providing a constant, substantially uniform static magnetic field within patient-receiving space 67.
A carriage 116 has a patient positioning assembly mounted thereon. The patient positioning assembly includes an elongated elevator frame pivotally mounted to the carriage for movement about a horizontal pivot axis 119. The patient positioning assembly further includes an elongated patient support or bed 120 with a footrest 124 at one end. The patient support is slidably mounted on the carriage. Appropriate actuators (not shown) are provided for driving the carriage along rails 117; for moving the patient support along the elevator frame 118 and for tilting the elevator frame about axis 119 relative to the carriage, so as to position patient support 120 in a generally vertical orientation as seen in
A set of gradient coils 106 is physically mounted within the magnet frame. The gradient coils are arranged in the conventional manner to apply magnetic field gradients within the patient-receiving space. The gradient coils in turn are connected to a gradient coil driver 107 which is controlled by computer 126. In the conventional manner, the computer 126 can control the gradient coil driver to apply appropriate currents to the various gradient coils so as to provide gradients in the desired direction within patient-receiving space 67 and to vary these gradients with time.
A conventional RF antenna 108 and RF transmitting and receiving apparatus 109 are also associated with the control and reconstruction computer 126. The antenna may include one or more elements positioned in the conventional manner within the magnetic frame, on the patient support 120 or even carried by the patient. The transmitting and receiving apparatus can be actuated by the computer to apply RF excitation signals and to receive the magnetic resonance signals emitted by the patient. The apparatus may use the same or different antenna elements for transmitting and receiving.
The apparatus further includes a system controller, also referred to as a control and reconstruction computer 126, linked to a display 101 such as a CRT display or printer and input/output devices 103 for entry of data and control commands into the computer. The control and reconstruction computer includes the conventional elements of a general-purpose computer, including a programmable processor and conventional memory devices for storing data and programs. The input/output devices 103 may include conventional elements such as a keyboard, as well as a conventional pointing device such as a mouse, touchpad or trackball, and may also include specialized command entry devices such as switches or pushbuttons used to control at least some aspects of movement of patient support 120. The system controller or computer 126 is linked through an interface (not shown) to the patient positioning assembly 116, so that the computer can control positioning of patient support 120. The positioning arrangement is further described in the aforementioned U.S. Provisional Patent Application Ser. No. 60/373,789 (“the '789 provisional application”) and in the commonly assigned, co-pending non-provisional U.S. patent application entitled POSITIONAL MAGNETIC RESONANCE IMAGING” filed of even date herewith and claiming benefit of said '789 provisional application. The disclosure of such non-provisional application is incorporated by reference herein.
In a method according to one embodiment of the invention, a patient P is disposed on patient support 120 so that the patient's back rests against the support and the patient's feet rest on footrest 124. To facilitate loading, the system may command the actuator to move the patient support to a pre-selected loading position (not shown) in which the patient support is generally vertical with footrest 124 near the level of the surrounding floor and rails 117, so that the patient can readily walk across the floor and step onto the footrest. This provides an easy and non-threatening loading procedure for ambulatory patients. Preferably, in the loading position the patient support is tilted back so that the plane of the support is at an angle of about 5-10 degrees from vertical. Thus, when the patient is positioned on the support, gravity will tend to hold the patient's back against the support.
When the patient is loaded on the support and the support and patient are in the desired position shown in
While the patient P is positioned as shown in
The gradients along the different axes vary with time in a pre-selected manner. One or more RF excitation pulses are applied by the RF transmitter. The cycle of operations required to elicit a single set of RF signals from the patient is commonly referred to as a “pulse sequence.” One example of a pulse sequence is shown in
During application of the slice select gradient, a pulse of RF energy is applied as indicated schematically at 134. The RF energy excites nuclei within a slice of tissue transverse to the slice select axis. The receiver in unit 109 is operated to receive magnetic resonance signals from the tissue during a so-called receive window or interval indicated schematically by an elongated bar 136 in
The magnetic resonance signals 138 received during the various repetitions of the cycle are digitized and stored in a memory (not shown) incorporated in the control and reconstruction computer 126. Using conventional techniques, these digitized signals are processed mathematically so as to recover information from the signals for individual voxels within the selected slice. The information recovered thus provides an array of data elements, each such data element having coordinates along the A1, A2 and A3 axes of the first frame of reference F1 defining the location of the voxel in that frame of reference and also having data defining one or more parameters related to the chemical composition and physical state of the matter within the voxel. The computer 126 reconstructs the data constituting the first, or coronal scout image into a visible image SC1C (
In exactly the same manner, the control computer acquires another scout image data set in the same frame of reference F1 but defining an additional scout slice in a plane normal to the plane of the first scout slice. In this process, the slice select gradient may be directed along axis A2, into and out of the plane of the drawing in
One or both of the scout images can be used in the conventional manner to select and specify the planes for one or more slices, typically many slices, useful in evaluation of anatomical structures, referred to herein as the “operative” slices. For example, the operator may command the control computer to display a line or plane representing the orientation and position of a proposed slice on monitor 101 in conjunction with either scout image SC1C or SC1S. Using controls 140 associated with the monitor and linked to the control computer, the operator instructs the computer to reorient or move the slice. The computer moves the line or plane representation of the slice on the image to represent the moved position and orientation. The operator continues this process until the representation of the slice is in the desired relationship to the anatomical structures of the patient as shown on the scout image. For example, in
Once the operator is satisfied with the disposition of the proposed slices relative to the patient, the operator signals the computer to acquire imaging data in the slices. The computer applies a slice selection gradient in a direction normal to the plane of each proposed slice and readout and phase encoding gradients in directions within that plane and applies RF signals in the manner discussed above with reference to acquisition of the scout images. In the same manner as discussed above, the system acquires magnet resonance data for voxels lying within each operative slice. The positions and orientations of the operative slices are known in the first frame of reference F1. Stated another way, position in a coordinate system CS0S (
In the next stage of the method, the patient is imaged again at a later time to acquire a second image data set. Preferably, the patient is positioned in the apparatus in the same orientation relative to gravity as was used during acquisition of the first data set. In exactly the same manner as discussed above, scout images SC2C and SC2S are acquired. Preferably, the second scout images are generally similar to the first scout images, i.e., if the first scout images were sagital and coronal images, then the second scout images also should be sagital and coronal images. However, the frame of reference F2 of the second scout images, indicated by coordinate system B1, B2, B3, is not in precisely the same relationship to the patient's body as the frame of references of the first scout images, inasmuch as the patient may not be in exactly the same position relative to the apparatus during acquisition of the second scout images.
In the next stage of operation, the operator instructs the computer to display the coronal second scout SC2C image on display unit 101 and also to display a cursor 144 overlayed on the images. Using controls 140, the operator maneuvers the cursor on the monitor until the cursor is aligned with a well-defined point P1, referred to herein as a “fiducial point,” as, for example, the tip of the sternum or other structure having a well defined, readily identifiable shape. Once the cursor is aligned, the operator enters a signal indicating such alignment. The computer records the current position of the cursor as coordinates along axes B2 and B3 of the second frame of reference. In a similar fashion, the operator aligns the cursor with other points P2 and P3 in the coronal scout image SC2C so as to provide coordinates for all of points P1, P2 and P3 along axes B2 and B3. The operator then commands the computer to display the sagital second scout image SC2S and maneuvers the cursor 144 and again aligns the cursor with the identified fiducial points P1, P2 and P3. As the cursor is aligned with each point, the operator again signals the computer which records the coordinate of each fiducial point along axis B1 of the second scout frame of reference. The operator then calls up the previously recorded first scout images SC1C and SC1S associated with the first image data set (
Once the transformation between the first and second frames of reference has been established, the computer transforms the data defining the position and orientation of the operative slices OSA1, OSB1 in the first frame of reference (
Once the second image data set has been recorded, the two data sets are compared with one another. In a simple comparison step, visually perceptible images of corresponding operative slices constructed from the first and second image data sets may be displayed side-by-side so that the physician may compare them visually. Such a comparison allows the physician to detect changes in the patient's anatomy occurring during the time between the first and second imaging procedures. For example, as seen in
Methods according to this aspect of the invention can detect changes such as long-term growth or regression of normal or abnormal tissues occurring over periods of, for example, days, weeks, months or years. The same methods can also detect changes occurring over shorter times. For example, methods according to this aspect of the invention can be used to detect changes in the anatomy occurring over a few minutes following administration of a drug, radiant energy or other therapeutic agent. For example, some procedures for monitoring chemotherapy use invasive evaluation techniques as, for example, catheterization of the bladder, so that the physician can visually examine a bladder cancer. This catheterization typically is performed at relatively infrequent intervals because it is invasive and painful. Thus, in common protocols for monitoring the response of bladder cancer to chemotherapy, a chemotherapeutic agent is administered for six weeks and then the patient is catheterized to determine whether the tumor has shrunk or grown. Using the preferred techniques according to the present invention, however, the patient can be examined more frequently by magnetic resonance imaging as, for example, at intervals of a week or so. This more frequent examination can provide a faster indication as to whether the particular agent being administered is or is not effective, so that the physician can begin therapies if warranted. Moreover, magnetic resonance imaging can provide a more accurate indication as to the overall growth or shrinkage of the tumor than visual examination through a catheter. In a therapeutic monitoring method according to this preferred aspect of the invention, the second imaging step is performed after administration of a therapeutic agent such as an anti-tumor drug or radiation regimen and the step of comparing the images is performed so as to determine the presence or degree of response to the therapeutic regimen as, for example, by evaluating the size of a tumor, evaluating the appearance of a tumor such as its contrast to surrounding tissue, granularity or other characteristic. Preferably, the changes detected are real changes in the anatomy, as distinguished from temporary distortions of the anatomy induced by the cardiac or respiratory cycle. The term “non-cyclical” change is used herein to describe a real change in the anatomy, as distinguished from a distortion due to cardiac or respiratory cycling. Also, changes in the anatomy other than changes in the size or shape of particular tissues can be monitored. Thus, changes in metabolic or physiologic processes occurring within tissues which change the chemical nature of the tissue can be seen in magnetic resonance images and can be monitored using procedures according to this aspect of the invention.
Most preferably, the two sets of image data are compared automatically by the computer. To perform such a comparison, the two sets of image data must be in a common frame of reference. If the region of the patient imaged in the imaged data is a substantially rigid portion of the patient's anatomy, the registration achieved by registration of the scout images as discussed above will suffice to place all of the data for the operative slices of the second set in the same frame of reference as the operative slices of the first set. That is, a voxel having particular coordinates within the plane of slice OSA2 in the second set theoretically lies at the same position within the patient as a voxel having the same coordinates in the corresponding slice OSA1 of the first imaged data set. Alternatively, the data constituting the operative slices themselves may be used to register or be a part of the data constituting one data set with the corresponding data in another data set. This alternative procedure can be employed if the scout image registration steps discussed above are not used and can also be employed to provide more accurate registration for local areas within slices as discussed below even if the scout image registration steps are employed.
In one registration procedure using the operative slice data, the operator commands the system controller 126 to display on monitor 101 a visually-perceptible first image 202 (
In the manual and automatic registration procedures discussed herein, fiducial markers can be used in place of anatomical features. Thus, fiducial markers can be physically attached to the patient's body prior to acquisition of the first image data set and remain in place during acquisition of the second image data set. The fiducial markers preferably are formed from a material which will provide contrast when seen against the background of the surrounding tissue in an MRI image. For example, the fiducial markers may be hollow objects filled with oil or other material having different magnetic resonance properties than the surrounding tissue, such as different spin-spin or spin-lattice relaxation time. Alternatively, the fiducial markers may be formed from a solid polymer or metal. The image data sets acquired during the procedure will include data representing the fiducial markers, and the displayed images will include depictions of the markers. In the cursor-based registration procedures discussed above, the operator may align the cursor with depictions of the fiducial markers, rather than with depictions of fiducial points in the anatomical structure. In an automated pattern-matching scheme, the pattern-matching algorithm may be arranged to search for data representing the known shapes of the fiducial markers. Also, if the fiducial markers have magnetic resonance properties differing from those occurring in other substances present in the imaging volume, the system may search through the image data set for data elements having those properties to locate the fiducial markers without use of pattern recognition.
Using the image data sets in a common frame of reference, and hence in registration with one another, the data in each data element of the second image data set is compared with the data in the data element of the first image data set having the same location. In a simple, single-element subtraction comparison scheme, the value specified by the data in each element of the second image data set may be subtracted from the value specified by the data in the corresponding element of the first image data set, or vice-versa. Other forms of comparison between such values may be used. For example, the value specified by the data in each element of the second image data set can be divided by the value specified by the data in the corresponding elements of the first image data set. Alternatively, the values specified by the data in the corresponding elements of the first and second data sets can be compared with one another so as to determine whether or not they are equal to one another within a preselected limit, and to provide a first result if they are equal and a second result if they are not. The result of the comparison is a comparison image data set representing the differences between the two images. In the subtraction scheme, the comparison image data set would have zero data at all locations if anatomical structures of the patient have not changed. In this scheme, non-zero values in elements of the comparison image data set indicate that a structure of the patient has changed between the time when the first image data set was acquired and the time the second image data set was acquired. In the thresholding scheme, each element of the comparison data set would contain either the first result or the second result. Elements having the second result would indicate a change in the anatomy.
In a variant of this approach, the elements of each data set may be combined into groups of plural data elements. For example, a plurality of first data elements having data representing magnetic resonance properties in a plurality of contiguous volume elements may be treated as a group. In effect, such a group forms a single, larger data element representing a single larger volume element. Alternatively, a plurality of first data elements representing a plurality of non-contiguous volume elements in a preselected spatial relationship to one another may be treated as a group. For example, a group of data elements representing volume elements disposed at a given radius from a central point, such as on a circle or sphere surrounding the central point, may be treated as a group. In a further variant, a the group may also include the data element representing the volume element at the central point. The elements of any such group have data representing magnetic resonance characteristics in the neighborhood of a particular location. The data in the data elements of a plural-element group can be combined in various ways to provide one or more composite properties for the group. For example, the values represented by the data in the various elements can be summed or averaged. Also, a measure of the diversity among values represented by the various elements in the group as, for example, the standard deviation or range of such values, can be calculated. The elements of the second image data set can be combined into groups in the same manner, and a composite property can be calculated for each such group. The composite properties of each group of first data elements can be compared with the composite properties of the corresponding group of second data elements, representing the same location, in the same ways as the data in individual first data elements can be compared to the data in individual second data elements. Here again, the result is a comparison image data set, with a plurality of individual comparison data elements each having a value which represents the degree or existence of differences between data for the same location.
To minimize spurious data in the difference set, the comparison image data set can be subjected to an image processing procedure commonly referred to as “erosion” which removes non-zero data for particular voxels unless a certain number of neighboring voxels also have non-zero data.
The elements of the comparison image data set inherently have locations corresponding to the locations in the image data sets. Thus, a data element of the comparison data set derived by comparing data from voxels at a given location in two image data sets represents comparison data for the given location. The comparison data set thus may be displayed as one or more separate visual images. Alternatively or additionally, the comparison data set may be used to select regions within one or both image data sets corresponding to locations where differences exist as, for example, where the comparison image set has non-zero data or, more preferably, data which is of a magnitude greater than a specified threshold. The frame of reference of the comparison image data set is the same as the frame of reference of the first and second image data sets which have been registered with one another. Thus, a visually-perceptible image may be formed from either the first or second image data set and displayed. Those pixels corresponding to voxels where the comparison image data set has values above the specified threshold are highlighted in the display, as by displaying these pixels in a contrasting color, different intensity, flashing or other visually-perceptible indication. For example, the image 204 from the second data set may be displayed and pixels of the image representing voxels where changes have occurred between the two sets of image data may be shown darkened as seen at 217 in
The highlighting procedure may also introduce artifacts into the displayed image. In the example depicted in
One particularly useful application of the techniques discussed above is in monitoring the growth or shrinkage of particular tissues, which may be abnormal tissues. In a further variant, the first and second image data sets are acquired using a magnetic resonance pulse sequence which yields tissue data for each voxel indicative of the presence or absence of a particular tissue characteristic within the voxel. The characteristic may be a characteristic indicative of an abnormality For example, the data in each data element may include the value of spin-lattice relaxation time or T1 for the tissue in such voxel. T1 values above a threshold are indicative of malignancy. In this instance, the comparison data set can be combined with the T1 values or tissue data to yield a revised comparison image data set indicating only those voxels where two conditions apply: (1) the difference between data values for the same voxel in the two image sets is above a set threshold; and (2) the T1 value in a selected one of the first and second image data sets is above another threshold. The displayed visible image may be highlighted only at pixels corresponding to voxels of the revised comparison image data set. This procedure alerts the physician to the most critical anatomical information: where the malignancy is growing (or shrinking), in a simple and effective manner, which does not depend upon complex artificial-intelligence inferences. Similar techniques can be used with other abnormalities or with normal tissues of a specified type. For example, in monitoring development of a muscle in an athletic training program or physical rehabilitation, the signal characteristic may be a signal characteristic associated with muscle tissue. This characteristic can be combined with the difference between image data sets in the same way as the T1 values discussed above to yield a revised comparison image data set indicating where changes in muscle tissue have occurred.
In the foregoing discussion, the individual being examined is referred to as a “patient.” This does not imply that an individual being examined must be known or suspected to have a disease or abnormality. Thus, the word “patient” as used herein includes individuals who are not known or suspected to have any disease or abnormality, as well as those who have a known or suspected disease or abnormality. According to a further embodiment of the invention, the temporal comparison methods discussed above are applied as a screening technique to a series of patients having no known or suspected disease or abnormality. Thus, magnetic resonance image data sets are acquired from each patient at intervals of weeks, months or years. An image data set acquired at one time is automatically compared with an earlier image data set acquired from the same patient at a later time to provide a comparison image data set. In one embodiment, the first and second image data sets are T1-weighted image data sets and, hence, the comparison image data set shows changes in the patient's body which may be indicative of a developing malignancy. Here again, one or more of the image data sets may be displayed as a visually-perceptible image. The comparison image data set itself may be displayed as a visually-perceptible image.
Techniques according to this aspect of the present invention are particularly useful in the case of screening procedures because they permit a human evaluator to focus on those aspects of the images which show some change. For example, a screening center may employ 100 slices per patient each time a patient is scanned. If such a center processes three patients per hour, it will generate data corresponding to 2400 visible images per day. The burden associated with reviewing all of these images may cause a reviewing radiologist to miss significant findings. However, where changes over time are highlighted, the radiologist can focus his or her attention on those images having highlighting. In a variant of the procedure discussed above, less than all of the data is displayed as visually perceptible images. If the automatic comparison between image data sets indicates a difference between successive data sets, the computer issues a difference signal. One or both of the data sets, most preferably the later-acquired data set involved in the comparison, is selected for inclusion in a subset of the generated data sets. By reviewing visually perceptible images in that subset, the radiologist will concentrate his or her review efforts on the data which is most likely to indicate development of an abnormality. Other factors may be considered in a scheme for deciding which data sets should be included in a subset to be reviewed by a human as, for example, the age of the patient, occupational data, medical history data, genetic data, or other data which tend to indicate the likelihood that the patient will develop an abnormality. In a variant of this approach, less than all of the data in some or all data sets is reproduced in the form of visible images for human inspection. For example, where the comparison step indicates a difference between data for particular corresponding slices in successive data sets from the same patient, only those slices, and optionally adjacent slices are displayed as visually perceptible images.
It is not essential to include the same number of slices in every data set. For example, in a screening procedure, a comprehensive base line data set incorporating data for numerous slices may be acquired during the first imaging procedure and subsequent data sets may include data for fewer slices. In a further variant, the automatic comparison step can be performed concomitantly with acquisition of a second or subsequent data set incorporating only a few slices. If comparison of this data with previously acquired data indicates that a difference exists, additional slices are acquired.
As mentioned above, use of the comparison data set in a visual display, either by display of the comparison image data set itself or by using the comparison data set to highlight a visually-perceptible image generated from another data set allows use of the comparative data without reliance on artificial intelligence or automated pattern recognition schemes. However, such schemes may be applied to the image data set. Merely by way of example, the system may find the dimensions of any region of contiguous voxels in the comparison image data set having non-zero values, or values above a selected threshold. Similarly, the comparison image data set can be processed to provide additional information. For example, the comparison image data set may be subjected to an automated feature-extraction process to find characteristics of the comparison image data set or of particular portions of the difference, features such as the ratio of dimensions can be found for each such region. These and other features can be extracted and compared with known disease patterns either in a rule-based system or by a neural network or other system capable of learning by exposure to a known learning set of comparison images.
To minimize or eliminate artifacts, the magnetic resonance pulse sequences may be gated to a particular time within the cardiac cycle, so that each repetition of the pulse sequence commences at the same phase of the cardiac cycle, as, for example, at a pre-selected time after end-systole or at a pre-selected time after end-diastole. The magnetic resonance pulse sequences also may be gated to the respiratory cycle, so that so that each repetition of the pulse sequence commences at the same phase of the respiratory cycle as, for example, at end-inspiration or end-expiration.
The disclosure of U.S. patent application Ser. No. 10/200,943, filed Jul. 23, 2002 is also incorporated by reference herein.
As these and other variations and combinations can be employed, the foregoing description of the preferred embodiments should be taken by way of illustration rather than by way of limitation of the invention as defined by the claims.
The present application claims benefit of U.S. Provisional Patent Application Ser. No. 60/373,789, filed Apr. 19, 2002, the entire disclosure of which is hereby incorporated by reference herein.
| Number | Name | Date | Kind |
|---|---|---|---|
| 3789832 | Damadian | Feb 1974 | A |
| 4350998 | Verhoeven | Sep 1982 | A |
| 4394684 | Verhoeven | Jul 1983 | A |
| 4411270 | Damadian | Oct 1983 | A |
| 4534358 | Young | Aug 1985 | A |
| 4608991 | Rollwitz | Sep 1986 | A |
| 4629989 | Riehl et al. | Dec 1986 | A |
| 4770182 | Damadian et al. | Sep 1988 | A |
| 4829252 | Kaufman | May 1989 | A |
| 4871966 | Smith et al. | Oct 1989 | A |
| 4875485 | Matsutani | Oct 1989 | A |
| 4908844 | Hasegawa | Mar 1990 | A |
| 4924198 | Laskaris | May 1990 | A |
| 4968937 | Akgun | Nov 1990 | A |
| 4985678 | Gangarosa et al. | Jan 1991 | A |
| 5008624 | Yoshida | Apr 1991 | A |
| 5065761 | Pell | Nov 1991 | A |
| 5153546 | Laskaris | Oct 1992 | A |
| 5162768 | McDougall et al. | Nov 1992 | A |
| 5197474 | Englund et al. | Mar 1993 | A |
| 5207224 | Dickinson et al. | May 1993 | A |
| 5250901 | Kaufman et al. | Oct 1993 | A |
| 5259011 | Petro | Nov 1993 | A |
| 5273040 | Apicella et al. | Dec 1993 | A |
| 5289374 | Doi et al. | Feb 1994 | A |
| 5305365 | Coe | Apr 1994 | A |
| 5305749 | Li et al. | Apr 1994 | A |
| 5315248 | Yamaguchi | May 1994 | A |
| 5349956 | Bonutti | Sep 1994 | A |
| 5386447 | Siczek | Jan 1995 | A |
| 5436607 | Chari et al. | Jul 1995 | A |
| 5490513 | Damadian et al. | Feb 1996 | A |
| 5515863 | Damadian | May 1996 | A |
| 5519372 | Palkovich et al. | May 1996 | A |
| 5520181 | Kreidler et al. | May 1996 | A |
| 5577502 | Darrow et al. | Nov 1996 | A |
| 5592090 | Pissanetzky | Jan 1997 | A |
| 5606970 | Damadian | Mar 1997 | A |
| 5615430 | Nambu et al. | Apr 1997 | A |
| 5640958 | Bonutti | Jun 1997 | A |
| 5647360 | Bani-Hashemi et al. | Jul 1997 | A |
| 5735278 | Hoult et al. | Apr 1998 | A |
| 5743264 | Bonutti | Apr 1998 | A |
| 5779637 | Palkovich et al. | Jul 1998 | A |
| 5810727 | Groen | Sep 1998 | A |
| 5873824 | Doi et al. | Feb 1999 | A |
| 5878746 | Lemelson et al. | Mar 1999 | A |
| 5997883 | Epstein et al. | Dec 1999 | A |
| 6009341 | Edelman | Dec 1999 | A |
| 6011396 | Eckels et al. | Jan 2000 | A |
| 6023165 | Damadian et al. | Feb 2000 | A |
| 6097977 | Collick et al. | Aug 2000 | A |
| 6112112 | Gilhuijs et al. | Aug 2000 | A |
| 6134464 | Tan et al. | Oct 2000 | A |
| 6141579 | Bonutti | Oct 2000 | A |
| 6198957 | Green | Mar 2001 | B1 |
| 6241671 | Ritter et al. | Jun 2001 | B1 |
| 6246239 | Krogmann et al. | Jun 2001 | B1 |
| 6249695 | Damadian | Jun 2001 | B1 |
| 6280383 | Damadian | Aug 2001 | B1 |
| 6335623 | Damadian et al. | Jan 2002 | B1 |
| 6411088 | Kuth et al. | Jun 2002 | B1 |
| 6414486 | Koellner et al. | Jul 2002 | B1 |
| 6414490 | Damadian et al. | Jul 2002 | B1 |
| 6509735 | Mueller et al. | Jan 2003 | B2 |
| 6534982 | Jakab | Mar 2003 | B1 |
| 6541973 | Danby et al. | Apr 2003 | B1 |
| 6556008 | Thesen | Apr 2003 | B2 |
| 6559641 | Thesen | May 2003 | B2 |
| 6567684 | Chenevert et al. | May 2003 | B1 |
| 6586934 | Biglieri et al. | Jul 2003 | B2 |
| 6667618 | Thesen | Dec 2003 | B2 |
| 6677753 | Danby et al. | Jan 2004 | B1 |
| 6690962 | Schmitz et al. | Feb 2004 | B2 |
| 6725078 | Bucholz et al. | Apr 2004 | B2 |
| 6828792 | Danby et al. | Dec 2004 | B1 |
| 6844884 | Balloni et al. | Jan 2005 | B2 |
| 6909792 | Carrott et al. | Jun 2005 | B1 |
| 6975896 | Ehnholm et al. | Dec 2005 | B2 |
| 7020578 | Sorensen et al. | Mar 2006 | B2 |
| 7151816 | Maier et al. | Dec 2006 | B2 |
| 7239906 | Green et al. | Jul 2007 | B1 |
| 7375521 | Damadian et al. | May 2008 | B1 |
| 20030009098 | Jack et al. | Jan 2003 | A1 |
| 20030016850 | Kaufman et al. | Jan 2003 | A1 |
| 20030055331 | Kotmel et al. | Mar 2003 | A1 |
| 20050187459 | Trequattrini et al. | Aug 2005 | A1 |
| Number | Date | Country |
|---|---|---|
| 1-242056 | Sep 1989 | JP |
| H01-242056 | Sep 1989 | JP |
| Number | Date | Country | |
|---|---|---|---|
| 60373789 | Apr 2002 | US |
