Patent Grant
6901277
The present invention relates generally to systems, methods, software, and graphical user interfaces for displaying and analyzing body images and for generating organ reports. More particularly, the present invention relates to graphical user interfaces and systems for analyzing one or more thoracic CT datasets to track and analyze lung nodules and other lung parameters.
Lung cancer is one of the most common forms of cancer among both men and women. Advances in medical imaging, such as CT and MRI scanning, have made it possible to localize and track early stage nodules that were previously non-detectable. However, such scanning protocols on a CT or MRI scanner typically generate no less than 40 images during a thoracic exam, while multi-slice protocols may generate 300 or more axial images. In order to analyze the dataset for lung nodules, the radiologist must review all of the slice images to localize the lung nodules. If a nodule is found in one slice image, the radiologist must then attempt to locate the nodule in the adjacent slices.
Unfortunately, such large amounts of data for each patient increases the probability that the radiologist will miss a potential nodule in their analysis of the image dataset, i.e., a “false negative.” Tumors may be too small to be reliably detected, or their appearance may be obscured by the surrounding tissues such as vessels. Missed tumors may be detected months or years later in a follow-up examination. During this interval the tumor may grow larger and, in the worst case, metastasize.
In their early stages of development, malignant lung tumors may not be detected even upon careful inspection of the image dataset. The early detection of lung cancer is of particular importance because the overall survival rate from the disease is very low. It is generally believed that early detection of cancer is beneficial, but in the case of lung cancer this is not established because of the recentness of the technique. The present invention provides tools which will help to elucidate this question.
To improve detection of lung nodules manual and semiautomatic pixel-based methods for segmenting CT images have been developed. One such method is manually creating a region of interest (ROI) delineating a nodule. A semiautomatic method requires a single, operator-defined seed point in which a computer algorithm will select similar contiguous gray-scale pixels that surround the seed point as the potential nodule. In another case it may be an operator-placed region of interest (such as a rectangle or ellipse) around the nodule.
While the proposed imaging methods offer significant potential to locate early stage nodules, still further improvements are desirable. In particular, if a small nodule or nodules are located in a first imaging scan, the radiologist will usually recommend that the patient return for a second, follow-up imaging scan. When the patient returns for the follow-up scan, the radiologist must relocate the nodules in the image scan and analyze the parameters of the nodules. Conventional imaging systems and methods do not provide an efficient way to determine if the nodules have increased in size, stayed the same, or the like.
Therefore, what is needed are reliable methods and devices which allow the radiologist to quickly and accurately localize and track any changes in nodules found in an imaging scan. Furthermore, what is needed is an improved method for visualization and characterization of small malignant lung tumors on thoracic image scan that would enable earlier detection of these tumors or nodules so as to enable earlier detection.
The present invention relates to methods and software for analyzing and comparing thoracic images of a baseline scan with thoracic images of follow-up scan(s) and for generating a report.
The present invention provide an efficient method of displaying and analyzing thoracic image scans and localizing potential nodules in the images. The present invention can be used to localize lung nodules and to determine the lung nodules relative position in the patient's lung, dimensions of the lung nodule, and other morphological parameters of a baseline image scan taken at t1. After the nodules are located in the baseline scan, through manual, semiautomatic, or automatic localization techniques, the nodule information derived from the images can be analyzed by the operator to determine if the potential nodule is an a benign or malignant nodule or merely a part of the patient's vasculature.
The nodule information can be stored and later transferred to a follow up image scan taken at t2 to ease the localization of the previously localized nodules in the follow-up scans. The stored nodule information provides a method that allows the operator to quickly determine if the nodules localized in the baseline scan have increased in size, the amount of increase, and the like. Additionally, the present invention can also allow the operator to quickly locate nodules—either manually or automatically with an analysis algorithm—that were not located in the baseline scan.
In exemplary embodiments, the present invention displays a first image along a body slice taken at t1 and a follow-up image of the same body slice taken at t2 on a user output device, such as a computer monitor. Typically, the first and second images are displayed adjacent to each other so as to allow the operator to visually assess changes in the nodules.
In order to compare the first image and second image the first and second images can be aligned. In some embodiments, a computer algorithm can automatically align the images. In other embodiments, the operator can manually pan the either the first or second image to align the images.
Typically, each of the nodules can be localized on the image with a marker to indicate its lesion number and its relative position in the patient's lung. By choosing the marker, the operator can view the statistics of the marked nodule which can include, but is not limited to, volume, surface area, major and minor axis, CT density or MRI signal intensity, signal standard deviation, signal histogram, roundness criteria, and the like.
Optionally, to improve the visualization of the nodules, the nodules in the baseline and follow-up images are colored differently from the surrounding tissue. Typically, the nodules displayed in the first image are displayed a different color than the nodules in the second image. Thus, if the first image and second image are superimposed over each other, the operator can visually assess the change in size (if any) of the superimposed nodules.
A third comparison image can be displayed on the user input device to illustrates changes (if any) between the first and second images. In particular, the comparison image can be used to illustrate the change in size of the nodules and any development of new nodules. Additionally, in some methods, a comparison chart can be displayed which quantitatively illustrates any change in size, volume, etc., of the previously localized nodules.
In another aspect, the present invention provides software code that allows a radiologist to compare a first body image taken along a body slice with a second image taken a long the same body slice. The first image will typically be a “baseline” image which the second image is compared with to determine if there have been any changes or additional nodules to the nodules in the lungs. The present invention further provides software code that allows the two images to be manipulated so that the two images are substantially aligned along three axis so as to ease the process of comparison. Software code calculates a third image to show the differences between the first and second image. The third image can be a sum of the first and second image, a subtraction of the first and second image which illustrates the differences between the first and second images, or the like.
In exemplary embodiments the software of the present invention includes software code for registering or aligning the first image with the second image. The software code can include an alignment algorithm that automatically aligns the baseline image with the follow up image. Alternatively, the software code can allow the user to manually align the baseline image and baseline image by physically selecting and moving one of the images into alignment with the other image.
Typically, the software includes a comparison algorithm that produces a third image for comparing the baseline image and the follow up image. The comparison image can produce a composite image that is either a sum or subtraction of the baseline and follow-up image. Such images provide a visual indication to the operator of any changes or additions to the nodules localized in the baseline image.
In another aspect, the software includes code that automatically generates a report comprising selected and annotated baseline and comparison images, lists of nodule number, size, roundness criteria and the like, and recommendations for patient follow-up care and studies. The software includes a decision tree for tailoring the report to the specific patient demographic information, medical history, and the results of the nodule analysis.
In yet another aspect, the code allows the physician operator to customize the elements and appearance of the report through the selection of text and graphics and through modification of the decision tree.
A further understanding of the nature and advantages of the invention will become apparent by reference to the remaining portions of the specification and drawings.
The present invention provides systems, software code, graphical user interfaces, and methods for displaying and analyzing lung CT or MRI image datasets of a patient. The lung datasets can be analyzed to map, track, and analyze the nodules in a series of lung slice images or image scans, as well as record other lung and chest abnormalities.
A lung slice image can be displayed on a user interface display for analysis by a radiologist or other operator. The methods of the present invention allows the radiologist to locate and map out the tumors, nodules, or lesions (hereinafter referred to as “nodules”) that are both manually localized and/or automatically localized by the software of the present invention. The mapped nodules can be segmented and have its volume and other dimensions ascertained. Such nodule information can then be transferred onto a lung report, if desired.
Exemplary embodiments of the present invention may allow an operator to compare a first, baseline scan taken at t1 with one or more follow-up dataset scans taken at t2, t3, etc. If an operator locates one or more nodules in the baseline scan, the operator can use the methods and software of the present invention to relocate the previously located nodules and to compare the nodule parameters from the baseline scan with the nodule parameters from the follow up scan(s). The operator can compare changes in volume, changes in surface area, changes in other morphological parameters of the nodules, and the like. The present invention also can also be used to locate additional nodules that appear in the follow up scan that were not assessed (or present) in the baseline scan.
In order to compare the baseline image with the follow up image, the imaging technical parameters (e.g., slice thickness, beam collimation, kV, mAs, table incrementation, slice overlap, reconstruction parameters) of the baseline dataset scan and the follow-up dataset scan should substantially correspond.
If the imaging parameters of the two images are not consistent, it may be difficult to align the baseline and follow up images for comparison. Nevertheless, methodology exists for correcting for some differences. For instance, images can be rescaled to adjust for spatial resolution, slice thickness, slice overlap and rotation. While changes in imaging parameters such as kV in CT or timing parameters in MRI would impede comparison of tissue characteristics, they would not affect size and volume measurements. It should be appreciated however, that there will inherently be some differences in the baseline and follow-up images due to external factors such as the cardiac cycle or breathing pattern of the patient. Some of these may be ameliorated by retrospective or prospective gating techniques. A more complete disclosure of Retrospective Gating can be found in co-pending Provisional Patent Application Ser. No. 60/306,311, entitled “Retrospective Gating,” filed concurrently herewith, the complete disclosure of which is incorporated herein by reference. Thus, the images will typically not be exactly the same. Such deviations will generally not prevent a comparison of the baseline and follow up images as long as they are not so large as to affect the individual measurement itself in a significant manner.
If the imaging parameters are substantially the same, the images can be “aligned” to allow for comparison. Aligning can include manually aligning the baseline and follow up image, or automatically aligning the baseline and follow-up images using an aligning algorithm.
The present invention further provides a lung report generator for producing customizable body reports that can incorporate information derived from the operator's analysis of the image dataset(s). The lung report generator can analyze a single image scan or can compare a plurality of body scans and produce a customizable lung report. The lung report generator can use a customizable decision tree to analyze the information from the images and determine what information is displayed on the lung report. Typically, the decision tree will generate information related to the nodules, patient demographic information, medical history, a comparison to population data, radiologist recommendations, standards of practice of the community, or the like. The customization aspect allows different imaging sites or each physician to customize the lung report to meet their own needs. This customization reflects community standards, personal standing in areas where consensus is not universal, a different knowledge base, and intellectual biases.
The present invention also provides graphical user interfaces for display on a user interface output device, such as a computer monitor, for displaying and analyzing the body image dataset(s). The graphical user interfaces provides windows for showing one or more image planes and inputs that allow the operator localize and analyze lesions in the body image(s). For comparison studies, the graphical user interface further allows the operator to view and compare the images of a baseline scan with the images in the follow-up scan so as to track and analyze potential lung nodules. Such information will allow the operator to assess any changes (visually and/or quantitatively) in the nodules over the time period between the baseline scan and follow up scan.
The locations of the nodules on the graphical user interface can be indicated with a marker, such as a number, symbol, shape, color, or the like. The map of the lesions can then be automatically transferred to a follow up image dataset to indicate where the potential nodules are, or should be, located in the follow up image dataset. In other embodiments, the potential nodules can be automatically colored differently from the surrounding tissue to allow the operator to more easily visualize the nodules.
The graphical user interface generally allows the operator to display the baseline image side-by-side with a follow up image to allow the operator to visually assess the changes in the size or shape of the nodules. In exemplary configurations, the graphical user interface provides a third, comparison image that is a combination of the baseline and follow up images. The combination image can be a sum image, a subtraction image, or a superimposed image of the baseline images. In one configuration, the lesions in the baseline and follow up images can be displayed in different colors so as to provide improved visual assessment of the change in size and/or shape of the lesion. Thus, when the baseline and follow up images are superimposed over each other, the corresponding nodules having different colors provide a visual indication for the purpose of matching nodules, and also indicates to the operator changes in size of the nodule (if any) in the various image planes, even though, the results of quantitative measures are more robust for this latter purpose.
In many embodiments, a chart or panel can be displayed on the graphical user interface to quantitatively compare the nodules in the baseline and follow up images. The panel can provide numerical indications of the change in volume, roundness, surface area, HU, mean, standard deviation, density (e.g. HU in x-ray CT or signal intensity in MRI), and the like. The chart will generally indicate which nodule is attached to the numerical information through the use of the marker which is superimposed over the nodule(s).
While the remaining discussion will focus on analyzing lung images and generating a lung report, it should be appreciated that the present invention is not limited to analyzing lung images and producing lung reports. For example, the present invention can also be used to analyze other body organs such as for use in the colon, the vascular tree, brain, Calcium Scoring, the whole body, or the like.
Referring now to
In a particular embodiment, as illustrated in
Communication network 20 may be comprised of many interconnected computer systems and communication links. Communication links may be hardwire links, optical links, satellite or other wireless communication links, wave propagation links, or any other mechanisms for communication of information. While in one embodiment, communication network 20 is the Internet, in other embodiments, communication network 20 may be any suitable computer network, such as an intranet, a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), or the like.
User interface input devices 34 may include a keyboard, pointing devices such as a mouse, trackball, touch pad, or graphics tablet, a scanner, foot pedals, a joystick, a touchscreen incorporated into the output device 32, audio input devices such as voice recognition systems, microphones, and other types of input devices. In general, use of the term “input device” is intended to include a variety of conventional and proprietary devices and ways to input information into computer system 24 or onto computer network 20.
User interface output devices 32 may include a display subsystem, a printer, a fax machine, or non-visual displays such as audio output devices. The display subsystem may be a cathode ray tube (CRT), a flat-panel device such as a liquid crystal display (LCD), a projection device, or the like. The display subsystem may also provide non-visual display such as via audio output devices. In general, use of the term “output device” is intended to include a variety of devices and ways to output information from computer system 24 to an operator or to another machine or computer system.
Storage subsystem 36 stores the basic programming and data constructs that provide the functionality of the various embodiments of the present invention. For example, database and modules implementing the functionality of the present invention may be stored in storage subsystem 36. These software modules are generally executed by processor 28. In a distributed environment, the software modules may be stored on a plurality of computer systems and executed by processors of the plurality of computer systems. Storage subsystem 36 typically comprises memory subsystem 38 and file storage subsystem 44.
Memory subsystem 38 typically includes a number of memories including a main random access memory (RAM) 42 for storage of instructions and data during program execution and a read only memory (ROM) 40 in which fixed instructions are stored. File storage subsystem 44 provides persistent (non-volatile) storage for program and data files, and may include a hard disk drive, a floppy disk drive along with associated removable media, a Compact Digital Read Only Memory (CD-ROM) drive, an optical drive, or removable media cartridges. One or more of the drives may be located at remote locations on other connected computers at other sites coupled to communication network 20. The databases and modules implementing the functionality of the present invention may also be stored by file storage subsystem 44.
Bus subsystem 26 provides a mechanism for letting the various components and subsystems of computer system 22 communicate with each other as intended. The various subsystems and components of computer system 22 need not be at the same physical location but may be distributed at various locations within distributed network 10. Although bus subsystem 26 is shown schematically as a single bus, alternate embodiments of the bus subsystem may utilize multiple busses.
Computer system 22 itself can be of varying types including a personal computer, a portable computer, a workstation, a computer terminal, a network computer, a module in the imaging unit, a mainframe, or any other data processing system. Due to the ever-changing nature of computers and networks, the description of computer system 24 depicted in
As shown in
When an operator desires to analyze the image data and/or produce a lung report, the operator will activate the software using conventional methods, such as clicking on an icon on the desktop with a cursor that is controlled by a mouse or other user input device 34 or voice recognition devices. The operator can request the patient image data via user interface 34 and the analysis module 46 can retrieve the requested information from database 14 and display the selected image(s) on user interface display 32. Generally, the image data will be displayed on a window 102 in graphical user interface 100. (FIG. 9). In some embodiments, the lung image will be shown as a single axial view. In other embodiments, three viewing planes (e.g., axial, sagittal, and coronal) of the image will be shown. In such embodiments, the coronal and sagittal view will have arbitrary locations, until the operator chooses, e.g., clicks on a point on one of the three views to set the point in which the image planes will correspond to. It should be appreciated, however, that it may be possible to display an oblique and/or reformatted image plane, if desired.
To improve the analysis of nodules in the lung image, the image data can then optionally be edited, either automatically by the analysis module 46 or manually by the operator, to set the thresholds of the image to separate vascular, interstitial, and emphasemic tissue from the image. (Step 54). Based on known tissue attenuation ranges, different tissue types (e.g., bone, tumors, etc.) can be identified by pixels within a specified gray level range. Setting the thresholds can be set by interactively moving corresponding bars on a histogram 104 to detect and filter out specific gray scale pixel intensity levels or using a Histogram Preset Button to remove the non-target tissue from the image(s). Typically, such Preset Buttons are configurable to allow the operator to customize the ranges of lung attenuation (HU) that is filtered out. It should be appreciated that gray level thresholding is merely one method of segmenting the lung image dataset and that other methods can be incorporated into the teachings of the present invention. For example, other methods of filtering out known tissue structures in the displayed image include the use of automated edge tracking algorithms, knowledge-based methods, and patient specific models, among others. Such methods are further described in Brown et al., “Method for Segmenting Chest CT Image Data Using an Anatomical Model: Preliminary Results,” IEEE Transaction on Medical Imaging, Vol. 16, No. 6, December 1997, and Brown et al., “Patient-specific models for lung nodule detection and surveillance in CT images,” SPIE Medical Imaging 2001, the complete disclosure of which is incorporated herein by reference.
Once the thresholding of the images is completed, the nodules in the image dataset can be selected for analysis. Such selection can be accomplished manually or automatically via a selection algorithm in the analysis module 46. One pixel-based method for selecting the lesion comprises manually tracing an outline of the potential nodule by manually tracing a region of interest (ROI) around each of the potential nodules by moving a mouse, joystick, or the like. Such tracing can require that the radiologist perform the same tracing function for each of the nodules in a series of the slice images. Such selection methods will eventually allow the computer software to log the anatomic position of the nodule, calculate the volume, surface area, and other morphological parameters of the potential nodule. In some semi-automated methods, the operator can select a potential nodule on one slice and analysis module 46 can automatically select the corresponding nodules either on the basis of spatial overlap or proximity or of gray-scaling in adjacent slice images. In other fully automated methods, the analysis module can use a selection algorithm to automatically select and map out the potential nodules in each of the slices. A further description of methods of selecting nodules can be found in Brown M. S. et al. “Knowledge-based segmentation of thoracic computed tomography images for assessment of split lung function,” Med. Phys. 27 (3) pp. 592-598 March 2000, Brown, M. S. et al. “Model-Based Segmentation Architecture for Lung Nodule Detection in CT” Radiology 217(P):207, 2000 (Abstract), Zhao, B. et al. “Two-dimensional multi-criterion segmentation of pulmonary nodules on helical CT images,” Med. Phys. 26 (6) pp. 889-895 June 1999, the complete disclosures of which are herein incorporated by reference. After the potential nodules have been selected, the operator can review the potential nodules selected. The operator will then have the option to accept or reject nodules that the operator does not believe is a nodule. Moreover, the operator further has the option to select other potential nodules that were not automatically chosen by analysis module 46.
Since nodules will often have the same signal intensity as blood vessels, chest wall or the like, problems in delineating the nodule may occur when the nodule or lesion is directly adjacent blood vessels, chest wall, or the like. In such cases analysis module 46 may have difficulty picking out the nodule from the adjacent body organs. Thus, if a non-selected image element appears to be a nodule that was not automatically selected by the analysis module 46, the operator can manually select a region of interest (ROI) around the suspected nodule in one or more of the image slices. The analysis module can then perform its analysis of the suspected nodule and provide roundness information, volume information, or the like. Likewise, if a selected nodule appears to be tubular in shape, such data would tell the operator that the nodule is likely part of the patient's vasculature, and not a nodule. If desired, the operator may then de-select the suspected nodule from further tracking for follow up body scans.
Once the nodules are selected for analysis, a grow algorithm in the analysis module 46 can calculate the parameter for each of the selected nodules (Step 56) and the statistics for each of the nodules will be logged as a file into database 14 (Step 58). The algorithm can calculate 3D data for the selected lesion such as volume, major and minor dimensions, density, roundness parameter, gray-level features such as mean and standard deviation, maximum and minimum signal intensity, and the like. The data derived from the image can include histogram statistics for each of the slices separately and for the entire slab/volume of the images. Analysis module 46 of the present invention can allow the user to move between the images slices to allow the operator to determine if the nodule has been properly selected in each of the slices.
In addition to calculating the nodule characteristics, the analysis module 46 can also assign each potential nodule a number or marker. Selecting the marker, typically by clicking on the marker will display the nodule statistics in a panel or chart on the graphical user interface. Such a panel allows the operator to “de-select” a localized nodule if it is determined that the selected element is not a lesion or nodule. The panel displayed on the graphical user interface will typically be only for the slice that is shown in the display 34. In order to view the data for all of the slices, the operator can scroll through all of the slices. If desired, the analysis module can be customizable to display a 3D rendering of the entire volume and the statistics for the entire volume.
If desired, in addition to statistics for lung nodules, the operator can also generate statistics for the lung and other body organs imaged during the scan. For example, it may be desirable to generate statistics for the patient's heart, mediastinum, heart, aorta, chest walls, or the like. The statistics can include data on volume, calcium scoring, surface area, density, or other visual impressions from the operator.
Based on the operator's analysis of the statistics produced by analysis module 46, the operator can enter a diagnosis or recommendation (Step 60). Such information can be entered into the database 14 via user input devices 34 such as a keyboard and mouse. In some embodiments, a report generator module 50 can automatically provide recommendations based on the parameters of the nodules in the images and on patient demographic and medical history information. For example, if a large number of nodules are found, the report generator module 50 can automatically recommend that the patient return in three months for a follow up scan.
If the image dataset is a first scan for the patient, the operator will have the option to create a lung report (Step 62). The report generation module 50 can produce a lung report that shows statistics of the lung images, some lung image cross-sections, recommendations, and the like. A sample lung report is provided in
If the image dataset is a second, follow-up scan for the patient, the operator will have the option to compare the follow up image dataset with a baseline image dataset of the patient. Using the data logged from the baseline image scan as a starting “map” of the potential nodules in the follow up scan, compare module 48 can automatically relocate the nodules in the follow up images. Automatically localizing the previously assessed nodules in the follow up scan, allows the operator to quickly determine if the previously located nodule(s) are growing or static.
It should be appreciated that in other methods of the present invention, instead of manually aligning the first and second image, the compare module 48 may automatically align the first and second images by matching the major axis of the lung using least square fits or other methodology. It is to be noted that this alignment need not be perfect, since matching criteria can include overlap of any one pixel in each nodule, or even matching by overlap of a “halo” region around each nodule.
If the two images aren't registered, the operator will not be able to properly compare the baseline images with the follow up images. In order to compare the baseline image with the follow up image, the technical imaging parameters of the two images should substantially correspond or be adjusted for such variations. In order to compare the baseline image with the follow up image, the imaging technical parameters (e.g., slice thickness, beam collimation, kV, mAs, Table incrementation, slice overlap, reconstruction parameters) for acquiring the images in the baseline dataset scan and the follow-up image dataset scan should substantially correspond. Nevertheless, methodology exists for correcting for some differences. For instance, images can be rescaled to adjust for spatial resolution, slice thickness, slice overlap and rotation.
One aspect of registering the images is aligning the lung nodules in the two images. One method of matching lung nodules in the two image is implementing software code that requires that at least one voxel in the nodules overlap. Another method of matching nodules is requiring that the edges of the lesions be no more than a fixed number of pixels apart. In exemplary embodiments, the software can require that the edges of the nodules (or other lesion) be no more than one pixel apart, two pixels apart, or three pixels apart. It should be appreciated however, that there are a variety of other methods for registering the images and nodules in the two images.
While changes in imaging parameters such as kV in CT or timing parameters in MRI would impede comparison of tissue characteristics, they would not affect size and volume measurements. It should be appreciated however, that there will inherently be some differences in the baseline and follow-up images due to external factors such as the cardiac cycle or breathing pattern of the patient. Some of these may be ameliorated by retrospective or prospective gating techniques, but, the images will typically not be exactly the same, and matching criteria will be such as to allow some resiliency in the process.
In exemplary embodiments, comparison module 48 illustrates to the operator any changes that may have occurred between the baseline scan and follow up scan. In some embodiments, illustration of the changes in the image datasets includes displaying a comparison image on the user interface display 34 (Step 72). The comparison image will typically be a composite image that can be a subtraction of the first and second image—i.e., all common image characteristics between the first and second image will not be shown and only the differences will be shown. Alternatively, the comparison image can be a sum of the first and second image which shows overlap as brighter than the non-overlapped areas of a nodule. If displayed in color, the baseline scan may mark the nodules in one color (e.g. blue), the follow up scan in another (e.g., yellow) and the overlapped region show in a third color (e.g., green). Alternatively, or in addition to the comparison image, comparison module 48 can provide a statistical chart that shows the changes to the nodules found in the first image dataset. In addition to comparing the nodules in the first and second image dataset, the software of the present invention can also automatically analyze the image datasets to determine if new nodules have formed in the lung tissue since the first image dataset.
After the nodules have been localized and/or compared, the operator can generate a lung report. The lung reports can include data comprising the patient's history, the image data, comparison data to national statistics, radiologist recommendations, prognosis for the number, size and location of nodules found in the scan, or the like. Typically, the software of the present invention can be configured to allow each of the lung reports to be customized. The lung reports can be customizable for each patient, if desired. Alternatively, each of the imaging facilities (or each of the radiologists) can configure the software to produce customized lung reports that are tailored to their needs.
Typically, the operator will be given a choice to choose images for a report. Three dimensional images can be selected prior to generating a report. In addition to displaying the selected images from the scan, if desired, the statistical data can be compared to population data. The other patient data can be derived from database 14 that stores the national percentile data. The patient's position relative to the national percentile can be established with a marker within a chart, as illustrated in FIG. 18C. Criteria for comparison can include smoking history, family history, gender, age, occupational history, and other factors that may be known at the time as influencing the risk factors of a particular individual or his/her prognosis.
There are several aspects to report customization. The first customizable aspect of the body report is the report layout. The layout affects the look and feel of the report. Fields can be laid out in particular locations. To accomplish the report layouts, a list of fields will be created. The list can be tagged with a position marker and the marker can direct the field to a particular place on a page of the report. In this way, the location, number, and order of the field can be chosen by the physician. The library of choices for report elements can be quite large and can include both graphic and text elements. Graphic elements can be fixed (e.g. a histogram of population statistics) or particular to the patient (e.g. an annotated image from the patient set.) Similarly, the text elements can be fixed (e.g. site name and address, or an explanation of the test) or related to the specific results of the study.
Another aspect of the customization is a decision tree or set of logic steps which can connect the fixed information results from the tests (e.g., demographic data) and patient history data. The program code can provide an interface comprising four sections on the computer screen display. In exemplary embodiments, one section can be a list of all the output fields (e.g., recommendations such as “urgent to visit a physician”, “need to return in six months”, “results are normal for subject's age, gender and medical factors”). The second section can be a list of all the input fields (e.g., subject age, gender, occupational status, number of nodules, size of nodules, and so on).
If the lists are longer than can be displayed at one time on the output display, a scroll bar can be used to cycle through the list. The operator can choose by a click an alphabetical or thematic ordering of each list. The third section contains buttons for all needed logic and arithmetic operations (e.g., AND, OR, NOT, GREATER THAN, LESS THAN, EQUAL TO, THEN, NEXT OUTPUT FIELD, FINISH OUTPUT FIELD, SUM, MINUS, DIVIDE, MULTIPLY, etc.) The fourth section contains a graphical representation of the logic tree as it is generated.
In exemplary embodiments, the operator can generate the logic or decision tree without any typing. For example, an output field can first be created by clicking or otherwise selecting the first section. Then, for that output field, a decision tree will be created by clicks on the input fields interspersed with clicks on the logic or arithmetic operators. In essence, the report can be constructed by stringing together various input elements like pearls in a necklace. Editing code can allow the operator to edit the string of fields.
Another customization possibility is the incorporation of information gathered from the aggregation and statistical analysis of previous patient results and outcomes. These previous patients can be those at the client imaging center, as well as those from other imaging centers with which the client center is sharing results data.
Further customization can occur through incorporation of the decision tree to analyze all of the statistical data collected, so as to give an automated recommendation to the patient. The automated recommendations can be personalized by each radiologist or imaging facility. For example, some radiologists may choose to incorporate data regarding age, race, gender, or the like, while other radiologists may choose to not choose to incorporate such information into his recommendations.
1. Has the never smoked, moderate smoker, heavy smoker?
2. How old is the patient? below 30, below 50, below 70?
3. Is the patient male or female?
4. What is the patient's race?
5. Does the patient's family have a history of lung cancer?
6. How many nodules does the patient have? less than 5 or more than 5?
7. Are any of the nodules over 1 cm in major axis? 3 cm in axis?
8. Has more than one study been performed? If so, have any of the nodules shown an increase in size?
Based on a combination of such questions or other sets of questions, the report generator module 50 will generate a different recommendation. Some hypothetical examples are as follows:
If the subject:
1. never smoked.
2. age 70.
3. male.
4. has 5 nodules.
5. no nodules over 1 cm in major axis.
6. recommend follow up in 3 years.
If the subject
1. heavy smoker.
2. age 50.
3. male.
4. has 5 modules.
5. one nodule over 3 cm in major axis.
6. recommend urgent follow up with primary physician.
If the subject:
1. never smoked.
2. age 70.
3. male.
4. has 5 nodules.
5. no nodule over 1 cm in major axis.
6. second study shows significant size increase in at least one nodule.
7. recommend urgent follow-up with primary physician.
Typically, the operator will be given the choice to choose between a regular report and a comparison report. If the report generated is a comparison report, the operator must select the first and second datasets that are to be compared. After the dataset is selected, the operator (or decision tree) can choose which information can be incorporated into the lung report. Typically, the information will include statistics for the nodules assessed during analysis, physician data inputs, such as recommendations, or other comments, findings regarding the surrounding organs such as the lungs, mediastinum, heart, thoracic aorta, and chest wall, patient data, such as age, gender, medical history, race, family medical history, smoking history, occupational history, asthma, allergies and other factors that may be known at the time as influencing the risk factors of a particular individual or his/her prognosis, or the like can be incorporated into the report. In some embodiments, the report generator module 48 can use the patient data to tailor any recommendations and conclusions. For example, if a patient has a family history or lung cancer, report generator module 48 can be customized to automatically recommend a follow up scan at a later time, even if no nodules were localized during the first scan.
As shown in
One exemplary graphical user interface 100 is illustrated in FIG. 9. In a particular embodiment, toolbar 108 is positioned along a top portion of the display and window 102 is positioned in a lower right portion of the screen. The body image slice illustrated is an axial CT image of the patient's lung. At the bottom of the graphical user interface are icons MPR1, MPR2, and MPR3 icons 110, 112, 114 that allow the user to toggle between different image planes (i.e. axial, coronal, and sagittal).
Graphical user interface toolbar 108 of the present invention will typically have a plurality of buttons or icons that upon activation can manipulate the image displayed in window 102. The icons can include a combination of, but is not limited to, a crossbar button 116, an overlay color button 118, a Draw Region of Interest (ROI) Button 120, a Block selection 122, an Auto Contour Button 124, a Select Lesion 126, an Edit Lesion 128, a 3D button 130, a Frames button 132, Edit Text 134, a Draw Line button 136, a Select Layout button 138, and a Report button 140.
As illustrated in
The operator can select the buttons on toolbar 108 to perform various tasks with the image. To begin obtaining data for the nodules in the lung images, the operator can activate the Draw ROI button 120 or the Auto Contour Button 124. Selecting the Draw ROI button 120 will allow the operator to manually draw a region of interest around the potential nodules. If the operator wishes to deselect a ROI, the operator can activate the Block Selection Button 122 to delete the previously selected potential lesion. Activation of Auto Contour Button 124 will activate analysis module 46 so as to automatically draw a ROI around potential nodules that are displayed within the slice displayed in window 102.
After the lesions have been selected in the displayed slice, the operator can repeat the selection process in the remaining slices, until all of the nodules have been selected. It should be appreciated however, that in some embodiments of the software of the present invention, the software may automatically select the nodules in the adjacent image slices or all of the image slices. After the lesions have been selected, the operator may optionally activate the Overlay Color button 118 so that the selected nodules will be displayed in a different color to differentiate the selected nodules from the surrounding tissue. Overlay color button activates analysis module 46 to automatically highlight the selected nodules in the window by changing the colors or intensity of the nodules or other selected tissue. Determination of which color to display the image elements can be based on the grayscale intensity of the selected elements.
In order to display statistics for an individual nodule, the operator can activate the Select Lesion button 126 and move the cursor over the desired nodule. By clicking on the desired nodule, chart 106 will display various statistics of the selected nodule. Some statistics that can be displayed are the anatomic position, number of pixels, area, volume, Value (HU), including minimum, maximum, median, mean, total sum, roundness, standard deviation, diameter 1, and diameter 2.
As shown in
As illustrated in
To identify the image planes displayed in each window, each of windows 102, 146, 148, 152 can be labeled with a title of the viewing plane, such as “Axial,” “Sagittal,” and “Coronal.”
In an exemplary embodiment, graphical user interface 100 can include a “Frames” button 132 in the toolbar 108 that allows the operator to toggle between concurrent and separate panning and zooming of image windows 102, 124, 126, 128. In a preferred embodiment, the images should start at the same scan FOV and then have concurrent zoom as the default. Activation of Frames button 132, will provide concurrent, proportional zooming and panning between the three image planes. Activating the Frames button 132 a second time will return the software to its default position in which each of windows will have independent zooming and panning in all of windows 102, 146, 148, 152.
If the operator wishes to view only a single pane view, the operator can double click within the selected window that the operator wishes to view. Alternatively, the operator can activate the Select Layout button 138 to toggle between a single plane view and a multiple plane view.
Toolbar 108 can further include a 3D button 140. 3D button toggles between 3D/2D mode for the statistics. When the 3D button is activated, analysis module 46 will display three dimensional statistics for the nodules. For example, in each plane, only partial information on spatial characteristics is available. The 3D button will display the nodule as fully rotatable axes which are controlled by the direction of mouse movement. The user can then see in real time the details of the shape in all axes of the nodule.
Referring now to
In some embodiments, the operator can click on the lesions to add it to list 174. In other configurations, analysis module 48 can automatically add the selected lesions to list 174. The operator can revise the list of lesions by clicking on the potential lesion to select the lesion and click on the delete button 176 or edit button 128 on toolbar 108. The lesion list 174 will characterize the lesions by its number, location, pixel size, surface area, volume, two diameters (longest and shortest), mean value, min/max, and median. The list of all of the tagged nodules provides the operator with the ability to review and edit the selected nodules, based on his observation of the statistics. A total list 178 can be used to show the total statistics for all of the tagged lesions.
If the operator chooses to compare two image dataset, the operator must select the two studies for comparison from database 14. Once the operator has selected the two studies, the software of the present invention will generate graphical user interface 200, as shown in FIG. 17. After the two studies are chosen, graphical user interface 200 will display at least two windows 201, 203 for displaying images from the first and second study. In exemplary embodiments, graphical user interface will include a lung nodule comparison panel 205 for displaying the statistics for the nodules in the first and second studies. As shown, the panel will include a portion for displaying the tagged nodules and their statistics for the first study 207, a portion for displaying the old tagged nodules and their statistics 209, and a portion for displaying any newly located nodules 211.
Lung nodule comparison panel 205 can include an axial button 213, coronal button 215, and sagittal button 217 for displaying corresponding images of the first and second study in their respective image planes in windows 201, 203.
After the first and second studies have been opened, the operator can align the first and second studies in all three planes. Alignment can be carried out manually by panning at least one of the images until the first and second images are aligned. Alternatively, the comparison module 48 can be activated, typically via an input button on the graphical user interface, to automatically align the first and second images.
In exemplary embodiments, activating the align button 219 starts a calculation process in the comparison module 48 to list all of the known nodules in the first study and to determine if any of the nodules localized in the second study are new or old. Criteria for determining whether a localized nodule in the second study is new or old can take a variety of forms. One example, one method of determining of a nodules is new or old is to determine there are N amount of common pixels of the nodules found in the first study and a nodule in the second study.
After the calculating process has been completed, the comparison data is posted into portions 207, 209, 211, respectively. If the nodule was incorrectly placed in the wrong portion of panel 205, the operator can choose the nodule and click on arrows 221 to move the chosen nodule to the correct portion. The comparison data can thereafter be saved as a separate file in database 14.
Once the operator has reviewed and tagged all of the nodules, and in the case of comparison studies analyzed the old nodules and located all of the new nodules, the operator can activate the Report button 180 to generate the lung report for the patient. An exemplary lung report 200 is illustrated in
The lung reports of the present invention will vary depending on whether the image dataset is a first study for the patient or a comparison study between a baseline and follow-up study. As illustrated in
In cases where the lung report is a follow up study, the lung report 200 can include a comparison chart 212 which lists the nodules statistics from the baseline study—typically in the form of volume, the number of nodules etc. The statistics from the baseline study can then be compared with the statistics from the follow up scan so as to display new nodules and any percentage change in size of the old nodules.
As illustrated in
In another aspect, the present invention provides a database 14 of lung nodule statistics (FIG. 1). Information from the database can be used to guide the software decision tree to customize the lung report and to guide a treatment recommendation.
In exemplary databases 14 of the present invention, the database can be accessed over a communication network, like the internet, and can be used to store lung information of the population. The population lung information can be used to determine an individual patient's standing relative to other people of a similar demographic so as to provide a customized recommendation based on the patient's percentile ranking.
In exemplary embodiments, database 14 can store population information in the database for every person scanned regarding each nodule. Such information can include, but is not limited to, nodule information such as the location of the nodule with respect to anatomic landmarks, volume of the nodule, major and minor axis of the nodule, an index of the nodule's roundness, surface area, average signal intensity, standard deviation, maximum and minimum pixel intensity and patient information, such as age, race, gender, smoking history, demographics, geographic location, diet, or the like.
In addition to the above nodule information, the database of the present invention can further store the number of nodules, total nodule mass, and demographics of each of the particular patients. From this database, one can build an expectation value for people in the population with similar demographics (e.g., age, gender, race, smoking pattern, diet, geographic location, and the like).
The population lung information can be searched and sorted using known methods. Typically, the database can be searchable based on at least one of the nodule statistics and/or patient fields. As the database is developed, it will be possible to allow imaging facilities from around the world to access the database and to compare the individual patient's lung nodule information with the population data of people of similar demographics. Typically, the information from each of the individual data scans can also be logged into the database so as to update the database and allow radiologists from around the country to compare the outcome of the patient's case study with other similar case studies.
When survival rates for different populations and nodule burdens become known through development of the database, the same database can be used to provide the lung information based on demographics and nodule burden. As can be understood, as this information is accumulated there will be additional information generated, for instance, whether the location of the nodules in the lung, or their proximity, may change risk factors and recommendations.
As can be appreciated, the stored information in the database can be used to guide the decision tree of the present invention regarding the customization of the lung report generator. It is understood that the risk and recommendations are dynamic and may change in response to the information accumulated in the database, or elsewhere. For example, a male of 50 years of age, with a history of smoking having a burden of five nodules under 3 mm maximum dimension which would be in a high risk group if the database information shows that the norm for this age and smoking history is 1 nodule not to exceed 5 mm.
After the patient lung information has been compared, the software of the present invention can provide a customized recommendation in the lung report 300 indicating the recommendations of the physician based on any combination of the patient's age, race, gender, smoking history, demographics, nodule information, and the like. Additionally, the lung report can provide a graph or chart that illustrates where the patient is relative to national rankings. (FIG. 18C).
While the above is a complete description of the preferred embodiments of the inventions, various alternatives, modifications, and equivalents may be used. For example, while the above methods recite methods of comparing lesions in a first image with the corresponding lesions in a second image, it should be appreciated that the present invention can be used to compare a “clean” baseline image with a second, follow up image to determine if any lesions have formed since the baseline image scan was taken. Although the foregoing has been described in detail for purposes of clarity of understanding, it will be obvious that certain modifications may be practiced within the scope of the appended claim.
The present application is related to U.S. patent application Ser. No. 09/908,432; filed Jul. 17, 2001, and entitled “Customizable Lung Report Generator” and U.S. patent application Ser. No. 09/908,508, filed Jul. 17, 2001, and entitled “Systems and Graphical User Interface for Analyzing Body Images,” each of which is incorporated by reference.
| Number | Name | Date | Kind |
|---|---|---|---|
| 4551800 | Riederer et al. | Nov 1985 | A |
| 5325293 | Dorne | Jun 1994 | A |
| 5359513 | Kano et al. | Oct 1994 | A |
| 5528492 | Fukushima | Jun 1996 | A |
| 5549117 | Tacklind et al. | Aug 1996 | A |
| 5570404 | Liang et al. | Oct 1996 | A |
| 5581460 | Kotake et al. | Dec 1996 | A |
| 5655084 | Pinsky et al. | Aug 1997 | A |
| 5664177 | Lowry | Sep 1997 | A |
| 5704371 | Shepard | Jan 1998 | A |
| 5713358 | Mistretta et al. | Feb 1998 | A |
| 5729620 | Wang | Mar 1998 | A |
| 5807256 | Taguchi et al. | Sep 1998 | A |
| 5832450 | Myers et al. | Nov 1998 | A |
| 5832504 | Tripathi et al. | Nov 1998 | A |
| 5895461 | De La Huerga et al. | Apr 1999 | A |
| 5911133 | Soble | Jun 1999 | A |
| 5917929 | Marshall et al. | Jun 1999 | A |
| 5950207 | Mortimore et al. | Sep 1999 | A |
| 5971767 | Kaufman et al. | Oct 1999 | A |
| 5986662 | Argiro et al. | Nov 1999 | A |
| 5987094 | Clarke et al. | Nov 1999 | A |
| 6014452 | Zhang et al. | Jan 2000 | A |
| 6029138 | Khorasani et al. | Feb 2000 | A |
| 6049622 | Robb et al. | Apr 2000 | A |
| 6061695 | Slivka et al. | May 2000 | A |
| 6075879 | Roehrig et al. | Jun 2000 | A |
| 6083162 | Vining | Jul 2000 | A |
| 6122351 | Schlueter, Jr. et al. | Sep 2000 | A |
| 6133918 | Conrad et al. | Oct 2000 | A |
| 6161080 | Aouni-Ateshian et al. | Dec 2000 | A |
| 6182029 | Friedman | Jan 2001 | B1 |
| 6190334 | Lasky et al. | Feb 2001 | B1 |
| 6205871 | Soloner et al. | Mar 2001 | B1 |
| 6206829 | Iliff | Mar 2001 | B1 |
| 6226352 | Salb | May 2001 | B1 |
| 6240201 | Xu et al. | May 2001 | B1 |
| 6253214 | Hall et al. | Jun 2001 | B1 |
| 6304848 | Singer | Oct 2001 | B1 |
| 6317617 | Gilhuijs et al. | Nov 2001 | B1 |
| 6348793 | Balloni et al. | Feb 2002 | B1 |
| 6549646 | Yeh et al. | Apr 2003 | B1 |
| 6560476 | Pelletier et al. | May 2003 | B1 |
| 6640001 | Roehrig et al. | Oct 2003 | B2 |
| Number | Date | Country | |
|---|---|---|---|
| 20030018245 A1 | Jan 2003 | US |
