专利汇可以提供System and method for automatic determination of a Region Of Interest within an image专利检索,专利查询,专利分析的服务。并且A system and method for automated image analysis selects a Region Of Interest (ROI) within a frame, suitable for subsequent Quantitative Intima Media Thickness (QIMT) analysis. Presently, a time consuming manual analysis by a highly trained technician is required to select the ROI within the frame for subsequent automated analysis. The automated image analysis provides an immediate and accurate identification of the ROI within the selected frame. In one embodiment, the automated image quality analysis may be directed to an ultrasound image of the carotid artery.,下面是System and method for automatic determination of a Region Of Interest within an image专利的具体信息内容。
The present invention relates to automated determination of a Region Of Interest (ROI) within an Image, and more particularly, to an automated determination of an ROI of an ultra sound image of a carotid artery.
According to the U.S. Center for Disease Control (CDC), heart disease is the leading cause of death in the United States, and atherosclerosis (a progressive narrowing and hardening of the arteries over time) is the leading factor in heart disease. Although initial thickening of arterial walls develops over a very long period of time, the complications from atherosclerosis generally occur suddenly and often prove fatal. Further, because atherosclerosis is asymptomatic, the first symptom perceived by a patient may be a life-threatening event.
Several methods are available for measuring the progress of atherosclerosis. These include Electron Beam Computed Tomography (EBCT), Magnetic Resonance Imaging (MRI), Flow Mediated Dilation (FMD), Quantitative Intima Media Thickness (QIMT), and the like. Although MRI may be use for clinical measurements, it remains a costly procedure. FMD appears promising, but outcome studies are not available to validate its predictive ability.
Carotid Intima Media Thickness (IMT) has been shown to be a useful surrogate endpoint. A surrogate endpoint is an indirect measurement which correlates with an actual (i.e., clinical) outcome. A clinical endpoint, for example, could be death, loss of vision, or other major symptomatic event. A corresponding surrogate endpoint could be a measurement which indicates the development of a condition which is correlated with a clinical endpoint event, but which surrogate endpoint may not produce easily observable symptoms. In particular, IMT measurements of the far wall of the carotid artery, about 1 cm below the bulb, have shown high correlation with cardiovascular clinical endpoints.
Carotid IMT provides a suitable surrogate endpoint for all stages of atherosclerosis and is well suited for tracking the progress of atherosclerosis. Carotid IMT has predictive value with respect to cerebrovascular and cardiovascular clinical events in both men and women. It has further been found that providing an ultrasound image of the carotid artery to a patient has a substantial effect of motivating changes to behavior, and the provision of subsequent images showing progress helps patients remain motivated to sustain a healthy life style.
Carotid IMT may be measured by either analog caliper IMT or digital Quantitative IMT (QIMT). Both methods rely on the same image and attempt to measure the same quantity. Caliper IMT is a manual measurement of the IMT visible in an ultrasound image using calipers, and is similar to printing a photo, and then measuring a dimension of a characteristic of the photograph by laying a ruler on the photograph. QIMT is a computer analysis of IMT performed on a digital version of the ultrasound image. Results have shown that caliper IMT has four times the standard deviation of QIMT when performed on the same images. Because of the accuracy of QIMT measurements, QIMT is ideal for both absolute measurements of the IMT, and for tracking small changes in the carotid artery, and thus the progression of atherosclerosis. The ability to track small changes allows the success or failure of a given course of treatment to be analyzed and adjusted if necessary, before a critical state is reached.
Current practice in the field of QIMT is to record carotid artery ultrasound images on SVHS tapes, and later manually review the images to determine a Region Of Interest (ROI) within each frame for subsequent computer analysis using edge detection algorithms. Such manual review requires highly skilled technicians and is time intensive. Because of the present requirement for manual review, IMT presently is only available from specialized laboratories employing highly trained technicians. There is thus a need for an objective system and method to automate the determination of the ROI of carotid ultrasound images, which method does not introduce operator bias.
The present invention addresses the above and other needs by providing a system and method for automated image analysis, which system and method selects a Region Of Interest (ROI) within the image (or frame) suitable for subsequent Quantitative Intima Media Thickness (QIMT) analysis. Presently, a time-consuming manual analysis by a highly trained technician is required to select the ROI within the frame for subsequent analysis. The automated image analysis of the present invention provides an efficient, timely, and accurate identification of the ROI within the frame.
In accordance with one aspect of the invention, there is provided a method for defining an ROI for subsequent processing, comprising specifying a physical characteristic of a morphological structure, providing at least one image of a portion of the morphological structure, said image containing the physical characteristic, and automatically processing the image to identify an ROI within the image, wherein the ROI contains the physical characteristic, and wherein the ROI is smaller than the original image, and the ROI is preferably from approximately 0.25 centimeters to approximately 1.0 centimeters wide by approximately 0.25 centimeters to approximately 1.0 centimeters high, and the ROI is more preferably approximately 0.5 centimeters wide by approximately 0.5 centimeters high.
The physical characteristic is preferably an Intima Media Thickness (IMT) of an artery, more preferably the IMT of a portion of a carotid artery, and most preferably the IMT of a far wall of the Bulb-Common Carotid Artery interface ranging between approximately 1.0 cm and approximately 1.5 cm below the base of the bulb, or most preferably the IMT of the far wall of a portion of the carotid artery ranging between approximately 1.75 cm and approximately 2.25 cm below the bifurcation. The image is preferably processed digitally, and the image may be provided as a digital image, or the image may be provided as an analog image, and processed to generate a digital image, and more preferably, the image is provided as a SVHS image and is processed to generate a digital image.
In one embodiment, the image is provided as a rectangular frame comprising a rectangular array of pixels, which frame is vertically sliced into a multiplicity of slices, which slices are preferably two pixels wide. The slices are processed to identify the presence of stripes. A stripe comprises three bands, a dark band with light bands both above and below the dark band, which light bands correspond to vessel walls, and which dark band corresponds to a vessel lumen. Each slice is processed by image processing software which identifies the stripes with confidence scores above a threshold, preferably a fifty percent confidence threshold.
Pairs of slices may be compared to detect the bifurcation of the carotid artery, wherein the pairs of slices preferably comprise adjacent slices. If for any reason, adjacent slices cannot be singly compared, substantially adjacent slices (e.g., less than 3 slices apart) may be compared, or a combination of substantially adjacent slices maybe compared (i.e., an averaging of features). If a first slice (i.e., a slice above, or after another slice) of a pair of slices has one more stripe than a second slice (i.e., a slice below or before another slice) and if top most and bottom most edges of two adjacent first stripes in the first slice (i.e., the slice with a greater number of stripes) correspond respectively to the top edge and bottom edge of a second stripe in the second slice, a bifurcation is identified. Preferably, the vertical positions of the corresponding edges will be within approximately five percent of each other. If the bifurcation is found, the ROI is horizontally centered approximately 2.0 centimeters before the bifurcation.
In some cases, the bifurcation may not be successfully identified. In these cases, the slices may be processed to attempt to identify the bulb of the carotid artery. The average number of stripes within the frame is computed. If a valid number of stripes is computed, an attempt is made to use the resulting value to determine the position of the bulb of the carotid artery. The portion of the carotid artery below the bulb is referred to as the Common Carotid Artery (CCA). If the number of stripes is three or more, the stripe that is next to the bottom is tagged as a selected vessel. If the number of stripes is two, the bottom stripe is tagged as the selected vessel. If the number of stripes is one, the one stripe is tagged as the selected vessel.
A moving average of the vessel diameter (i.e., lumen diameter) is computed, preferably from ten consecutive slices. The center of the moving average of slices is tagged as the bulb position if the lumen diameter increases sufficiently from one average to the next average, preferably at least ten percent lumen diameter increase. If the bulb is found, the ROI is horizontally centered approximately 1.25 centimeters before the bulb.
In some cases, neither the bifurcation nor the bulb are detectable. In these cases, the automatic processing may select a defined region approximately twenty percent higher and approximately twenty percent wider than the ROI from a previous frame. A pattern match may then be automatically performed within the defined region to determine a new ROI which matches a previous ROI. If all of the automatic processing fails to identify the ROI, the ROI may be a manually generated by a technician.
It is a further feature of the invention to provide a method for determining a Region Of Interest (ROI) for subsequent Intima Media Thickness (IMT) analysis, the method comprising providing at least one digital frame of a portion of a carotid artery, wherein the frame includes at least one of a set consisting of a bulb of the carotid artery and a bifurcation of the carotid artery, and automatically processing the frame to identify an ROI in the frame. The automatically processing comprises at least one of the set consisting of determining the ROI by detecting a bifurcation of the carotid artery, determining the ROI by detecting a bulb of the carotid artery, and determining the ROI by pattern-matching a portion of the frame to a previous ROI.
Determining the ROI by detecting a bifurcation of the carotid artery comprises slicing the frame into vertical slices, automatically comparing at least one of the vertical slices to at least one other of the vertical slices to find the bifurcation of the carotid artery, determining an ROI horizontal extent ranging from approximately 1.75 cm before the bifurcation to approximately 2.25 cm before the bifurcation, identifying a highest point of a far wall segment of the carotid artery within the ROI horizontal extent, and determining an ROI vertical extent ranging from approximately 0.3 cm above the highest point to approximately 0.2 cm below the highest point.
Determining the ROI by detecting a bulb of the carotid artery comprises slicing the frame into the vertical slices, automatically comparing at least one of the vertical slices to at least one other of the vertical slices to find the bulb of the carotid artery, determining the ROI horizontal extent to range from approximately 1.00 cm before the bulb to approximately 1.50 cm before the bulb, identifying the highest point of a far wall segment of the carotid artery within the ROI horizontal extent, and determining the ROI vertical extent to range from approximately 0.3 cm above the highest point to approximately 0.2 cm below the highest point.
Determining the ROI by pattern-matching a portion of the frame to a previous ROI having a previous horizontal extent and a previous vertical extent comprises determining an expanded ROI horizontal extent to the previous horizontal extent expanded horizontally by a factor of approximately 1.2, determining an expanded ROI vertical extent to the previous vertical extent expanded vertically by a factor of approximately 1.2, performing an automatic pattern match comparing the previous ROI to the region defined by the expanded ROI horizontal extent and the expanded ROI vertical extent, and determining the ROI based on the highest correlation of the automatic pattern match.
It is an additional feature of the invention to provide a system for processing ultrasound images of the carotid artery, comprising an ultrasound sensor, an ultrasound apparatus adapted to generate frames, a sensor cable connecting the ultrasound sensor to the ultrasound apparatus, a computer, and a path adapted to carry the ultrasound data from the ultrasound apparatus to the computer, wherein the path is selected from a group consisting of dedicated cables, phone lines, a Local Area Network (LAN), the Internet, and a wireless path.
The ultrasound apparatus is adapted to receive measurements from the ultrasound sensor and to process the measurements to generate ultrasound data. The computer is adapted to pre-process the ultrasound data to generate Regions Of Interest (ROI) within the data for subsequent processing, which ROIs are subsets of the ultrasound data.
The above and other aspects, features and advantages of the present invention will be more apparent from the following more particular description thereof, presented in conjunction with the following drawings wherein:
Corresponding reference characters indicate corresponding components throughout the several views of the drawings.
The following description is of the best mode presently contemplated for carrying out the invention. This description is not to be taken in a limiting sense, but is made merely for the purpose of describing the general principles of the invention. The scope of the invention should be determined with reference to the claims.
The present invention provides a system and method for automated image quality analysis which selects a Region Of Interest (ROI) within the frame for subsequent analysis. In one embodiment, the subsequent analysis comprises computing a Quantitative Intima Media Thickness (QIMT) of a carotid artery. The QIMT is a useful parameter in determining the cardiovascular health of a patient, and particularly the presence of heart disease, and more particularly the presence of atherosclerosis. QIMT is capable of providing very accurate measurements, and as a result QIMT is ideal for both absolute measurements of the IMT, and for tracking small changes in the carotid artery, and thus the progression of atherosclerosis. The ability to track small changes allows the success or failure of a given course of treatment to be analyzed and adjusted if necessary, before a critical state is reached.
QIMT may be performed using, for example, the ARTIS computerized edge contour measurement technique developed by Prevention Concepts Inc., in West Los Angeles, Calif. Although the automated image quality analysis method of the present invention is described herein in the context of pre-processing images for QIMT of the carotid artery, those skilled in the art will recognize that other methods of processing images will similarly benefit from the present invention, and in particular, other methods of processing ultrasound images will similarly benefit from the present invention, and even more particularly, other methods of processing ultrasound images of any artery or of the carotid artery in particular will similarly benefit from the present invention, and the application of the present invention to pre-processing any image is intended to come within the scope of the present invention.
Elements of a typical system according to the present invention is shown in
The computer 20 processes the signal received from the ultrasound apparatus 16 and may provide results on a screen 26, or over an output cable 22 to an output device 24. When the computer 20 is remote from the ultrasound apparatus 16, the results may be transmitted back to the ultrasound apparatus over the path 18, or over a different path, and be displayed on a second screen 28
A more detailed view of the patient 10 and ultrasound sensor 12 is shown in
A method for automated Region Of Interest (ROI) determination according to the present invention is shown in
The image frames are generally provided as rectangular first frame 86a as shown in
The frames 86a, 86b may be vertically sliced into a multiplicity of slices 88 as shown in
The slices may be compared to detect the bifurcation of the carotid artery, preferably adjacent slices are compared. If for any reason, adjacent slices can not be singly compared, substantially adjacent slices (i.e., less than 3 slices away) may be compared, or a combination of substantially adjacent slices may be compared (i.e., an averaging of features). For example, the first slice 92a and second slice 92b maybe compared as shown in
If the bifurcation 91 is identified, the ROI is horizontally centered approximately 2.0 centimeters before the bifurcation, shown by dimension D1 in
In some cases, the bifurcation may not be successfully identified. In these cases, the slices may be processed to attempt to identify the bulb of the carotid artery. The number of stripes (or vessels) in each slice, for example in a third slice 92c as shown in
A rounded average number of stripes is computed by rounding the average to the nearest whole number. The mean and variance of the number of stripes is computed relative to the rounded average number of stripes. The rounded average number of stripes is tagged as valid if the mean is below a mean_threshold, and the variance is below a var_threshold. The mean_threshold is preferably approximately 0.25, and the var_threshold is preferably approximately 1.0.
The minimum confidence level of the stripes within each slice is compared to a min_con_threshold. Slices having one or more stripes with a confidence level below the min_con_threshold are tagged as outliers, and not used in subsequent computations. A second variance is computed based on the remaining slices, and the second variance is compared to a second_var_threshold. An average confidence level is computed for the remaining stripes and compared to an avg_con_threshold. If either the second variance is greater than the second_var_threshold, or the average confidence is less than the avg_con_threshold, the average number of stripes is tagged as invalid. The second_var_threshold is preferably approximately 1.0, and the avg_con_threshold is preferably approximately 75 percent.
If a valid number of stripes is computed, average~stripes is set to the rounded average number of stripes, and an attempt is made to use the resulting value to determine the position of the bulb 34 of the carotid artery 30 (see
In each of the three cases described above, one or more of the slices may have a number of stripes different from the average_stripes. In this case, initially, the stripes are selected from the slices having a number of stripes equal to the average_stripes. Stripe position is set to the average of the positions of the tagged stripes. The stripe in the remaining slices (i.e., slices with more than one stripe) which overlaps the stripe position is tagged as the selected stripe. If no stripe overlaps the stripe position, the slice is not used for subsequent processing.
A moving average of the stripe diameters is computed using several consecutive selected stripes, the number of consecutive selected stripes is preferably between five and fifteen, and. the number of consecutive selected stripes is more preferably ten. Consecutive averages are compared. If the average increases by at least ten percent, the bulb position 112 is set to the position of the center slice from the average, as shown in
If neither the bifurcation nor the bulb are found, a pattern match may be attempted within a failed frame 114 to position the ROI as shown in
If neither the bifurcation, nor the bulb, nor a pattern match are found, and if only a small number of consecutive failures have occurred, the ROI is set to the previous ROI, otherwise, the ROI is manually determined by a technician.
A method for analyzing a frame for bifurcation 44 is shown in
A method for determining an average number of vessels in a frame 84 (see
A method for technician-generated ROI 62 (see
If a match was not found at 126, and more than a small number of frames have failed to result in a matching pattern at 130, and the technician does not reinitialize automode at 134, the ROI is determined by the technician at 129, and the processing above is repeated. If a match was not found at 126, and more than a small number of frames have failed to result in a matching pattern at 130, and the technician does reinitialize automode at 134, the automode is returned to. If a match was not found at 126, and more than a small number of frames have not failed to result in a matching pattern at 130, the ROI is determined to be the ROI from the previous fame at 132, the technician may reinitialize automode at 136 and return to automode, or the technician may not reinitialize automode at 136, and a frame is selected for analysis at 122. And the processing described above is repeated.
The present invention finds industrial applicability in the field of automated determination of a Region Of Interest (ROI) within an Image. More particularly, the present invention finds industrial applicability in the field of automated determination of an ROI of an ultra sound image of a carotid artery.
While the invention herein disclosed has been described by means of specific embodiments and applications thereof, numerous modifications and variations could be made thereto by those skilled in the art without departing from the scope of the invention set forth in the claims.
标题 | 发布/更新时间 | 阅读量 |
---|---|---|
一种具有更新功能的电镀产品的入库方法 | 2020-05-08 | 622 |
基于人脸识别的数据验证方法 | 2020-05-08 | 193 |
基于云服务的广告远程方法、装置、系统、产品及介质 | 2020-05-08 | 784 |
一种用于冷冻电镜分析的单颗粒图像聚类方法 | 2020-05-11 | 840 |
一种检测待测化合物是否调控脂滴生长的方法 | 2020-05-08 | 860 |
一种基于实时视频图像分析的落水检测方法 | 2020-05-11 | 756 |
基于电阻抗成像的三维血液灌注图像产生方法与装置 | 2020-05-08 | 534 |
影像噪声定位方法及系统 | 2020-05-08 | 314 |
一种智能感知的机场保障能力实时评估系统及评估方法 | 2020-05-08 | 468 |
一种肿瘤药物敏感性试验图像分析仪摄像装置 | 2020-05-08 | 666 |
高效检索全球专利专利汇是专利免费检索,专利查询,专利分析-国家发明专利查询检索分析平台,是提供专利分析,专利查询,专利检索等数据服务功能的知识产权数据服务商。
我们的产品包含105个国家的1.26亿组数据,免费查、免费专利分析。
专利汇分析报告产品可以对行业情报数据进行梳理分析,涉及维度包括行业专利基本状况分析、地域分析、技术分析、发明人分析、申请人分析、专利权人分析、失效分析、核心专利分析、法律分析、研发重点分析、企业专利处境分析、技术处境分析、专利寿命分析、企业定位分析、引证分析等超过60个分析角度,系统通过AI智能系统对图表进行解读,只需1分钟,一键生成行业专利分析报告。