- Email: [email protected]
Quantification of power Doppler energy and its future potential
FERTILITY AND STERILITY威 VOL. 76, NO. 3, SEPTEMBER 2001 Copyright ©2001 American Society for Reproductive Medicine Published by Elsevier Science Inc. Printed on acid-free paper in U.S.A.
Quantification of power Doppler energy and its future potential Nazar N. Amso, Ph.D., Sean R. Watermeyer, M.B.B.Ch., Neil Pugh, Ph.D., Sharon O’Brien, D.M.U., and Arianna D’Angelo, M.D. Department of Obstetrics and Gynecology, University of Wales College of Medicine and the University Hospital of Wales, Cardiff, Wales, United Kingdom
Objective: To test a new software package (Color Quantifier, Kinetic Imaging, Liverpool, United Kingdom) that quantifies power Doppler energy and to determine its reproducibility. Design: Intraobserver and interobserver reproducibility study. Setting: University tertiary referral center. Patient(s): Transvaginal power Doppler ultrasound images were recorded from women taking part in a study evaluating the physiological vascular changes in the uterus and ovaries during the normal menstrual cycle. Intervention(s): Nineteen consecutive frames of regions of interest in the uterus, ovary, and follicle, respectively, were analyzed by each of four observers on 10 occasions. Main Outcome Measure(s): Analysis of variance to determine the image and observer effect as well as the intraobserver and interobserver coefficients of variation. Result(s): Significant image and observer effects were found. However, the image effect was by far the largest component of the total variation. The large image-to-image variability was expected because the cardiac cycle was included within the 19 frames (images) analyzed. The combined intraobserver and interobserver variation, expressed as the coefficients of variation, was found to be small for the above indices (as low as 1.9%), particularly for total ovary and endometrium. Conclusion(s): The indices obtained with this color quantification software are reproducible in an in vitro setting using prerecorded images. Its applicability as a useful assay in the clinical setting requires further evaluation. (Fertil Steril威 2001;76:583–7. ©2001 by American Society for Reproductive Medicine.) Key Words: Power Doppler energy, objective quantification
Received September 22, 2000; revised and accepted March 2, 2001. Supported by the National Assembly, Wales, United Kingdom. Presented at the annual research registrars meeting at the Royal Society of Medicine, London, United Kingdom, February 16, 2001. Reprint requests: Nazar N. Amso, Ph.D., Department of Obstetrics and Gynecology, University Hospital of Wales, Heath Park, Cardiff, CF14 4XW United Kingdom (FAX: 4429-20743722; E-mail: [email protected]). 0015-0282/01/$20.00 PII S0015-0282(01)01940-9
Power Doppler energy (PDE) is a relatively new modality in ultrasound technology that is being used in a number of specialties and has important diagnostic and therapeutic potential (1). PDE uses the amplitude of the Doppler signal as opposed to the mean frequency shift (2). It is advantageous over traditional color Doppler insofar as it is angle independent and there is no problem with aliasing. Background noise, which can present difficulties in evaluation using traditional color Doppler, is easily distinguishable from true flow when using power Doppler (2, 3). PDE has enhanced sensitivity and can therefore be employed to detect image areas of low blood flow that are currently undetectable by color Doppler imaging (2, 3). Such areas include the microvasculature to be found in the kidney, liver, ovary, endometrium, and other tissues of interest. PDE results from the scattering of ultrasound waves
from red blood cells in arterial, arteriolar, venular, and venous blood vessels. Hence it is generally agreed that PDE reflects the vascularity of the tissue in question, at any given time. Quantification of PDE in two-dimensional ultrasound machines has been a subject of interest in recent years and has ranged from subjective methods (4) to semiquantitative techniques (5). We describe and report our assessment of new software (Color Quantifier [CQ], Kinetic Imaging, Liverpool, United Kingdom) that is a personal computer– based system and can be used for the capture and objective quantification of the PDE image. The aim of this study was to determine the intraobserver and interobserver variability for the analysis of color Doppler energy images of pelvic structures (uterus, ovaries, and follicles) 583
FIGURE 1 Ultrasound color power Doppler images. Top left: transvaginal gray-scale image of endometrium. Top right: ROI drawn around endometrium. Bottom left: ROI drawn around ovarian follicle. Bottom right: ROI drawn around ovary. Center: color reference bar.
Amso. Quantification of power Doppler energy. Fertil Steril 2001.
using this new software. This study was approved by the hospital’s ethics committee, and all patients gave written consent before the study was performed.
MATERIALS AND METHODS Software The CQ interprets the scanner’s Doppler color scale and converts the color image data to a numerical array of power values. This new software is capable of both online or deferred analysis from recorded video images. It can be interfaced to videocassette recorders and ultrasound scanners and operates in Microsoft Windows NT so that data can be exported to a statistical package for further analysis. For the purposes of this study, a Toshiba Corevsion (model SSA350A) ultrasound machine was used, together with a Super VHS video tape recorder. Power Doppler settings that 584
Amso et al.
Quantification of power Doppler energy
achieved optimal power Doppler color images with minimal artifact were determined at the outset of the study and remained unaltered throughout all scans during the study period. High-resolution video images were “grabbed” and stored on the hard drive of a personal computer (Pentium II processor at 333 MHz, operating on Windows NT Workstation V4). A calibration process was performed before image analysis. During this process, the following areas were selected: area 1, the reference color bar (Fig. 1); area 2, an area of gray-scale imaging only; area 3, an area incorporating gray scale and power Doppler energy. The software analyzes areas 2 and 3 to assign a value to the gray scale pixels, as well as interpreting the color pixels within area 3 in accordance with the reference color bar. This information was then stored as a calibration file and Vol. 76, No. 3, September 2001
TABLE 1 F Values and adjusted coefficient of variation (includes interobserver and intraobserver variation) after a two-way (image and operator) analysis of variance. Parameter Ovary CDA MCE PCE ICE/mm2 Endometrium CDA MCE PCE ICE/mm2 Follicle CDA MCE PCE ICE/mm2
Fvalue image (Fi)
Fvalue operator (Fo)
Ratio FI: Fo
Adjusted coefficient of variation (%)
187.1 1068.7 145.5 733.0
19.7 13.3 10.8 211.4
9.5 80.4 13.5 3.46
2.5 2.3 1.9 2.6
394.7 668.9 922.8 941.5
159.0 92.9 2.8 79.72
2.48 7.2 327.6 11.8
9.5 8.1 6.0 5.3
364.7 376.6 106.7 407.8
76.6 44.9 18.5 7.7
4.8 8.4 5.8 53.1
13.0 17.7 16.4 13.0
Note: All F values are significant at 5% significance level (P ⬍ .05). Amso. Quantification of power Doppler energy. Fertil Steril 2001.
used for subsequent image analyses. After this, regions of interest (ROIs) were identified and selected. Regions of interest may be of any shape and were literally drawn around the areas for analysis (Fig. 1). One or more ROIs may be drawn at any one time. It was also possible to exclude areas within a ROI from the analysis, such as fluid areas within cysts, follicles or tumors. The software has been developed to calculate the following indices; [1] the area of the ROI in square millimeters, [2] the area within the ROI occupied by the color pixels—the color Doppler area (CDA), [3] the mean color energy (MCE), [4] the peak color energy (PCE), and [5] the integrated color energy per square millimeter (ICE/mm2). The ICE/mm2 is a measure of the overall intensity of the PDE signal that incorporates not only the CDA but also the summated energy within the CDA in any given region of interest. This represents a true reflection of vascular density or vascularity within the tissue being analyzed at any given time. All of the above indices were calculated for the ROI studied.
Subjects The software analysis was evaluated for its reproducibility within and between observers of a multidisciplinary team (gynecologists, physicist, sonographer) working in the assisted reproduction unit within our institution. Video recordings of endometrium, ovary, and ovarian follicle were chosen for the analysis. The images were recorded from transvaginal power Doppler ultrasound scans of women taking part in a study evaluating the physiological vascular changes in the uterus and ovaries during the normal menstrual cycle. The transvaginal route was chosen because of its superior image quality. FERTILITY & STERILITY威
From each recording, 19 consecutive frames (images) of approximately 5 seconds’ duration were selected on account of the gray scale and power Doppler clarity. Each of the four analysts then independently evaluated these same images on 10 occasions, attempting to redraw the same ROI on each occasion. The ROIs and the methodology in drawing the ROIs were determined before commencing the study. The drawn ROI remained constant throughout the 19 frames of each single analysis. To ensure the appropriateness of the ROI positioning, it was possible to preview the drawn ROI superimposed on all frames to be analyzed.
Statistical Analysis After analysis, the numerical data of the above power Doppler indices (CDA, MCE, PCE, and ICE/mm2) were exported to a statistical package, namely Microsoft Excel and SPSS version 10 (SPSS, UK Limited, Surrey, UK). An analysis of variance, in which the effects were assessed by the F value (6), was carried out to determine variation between and within observers and variation between images for each of the indices (CDA, MCE, PCE, and ICE/mm2). This was repeated for the endometrium, ovary, and ovarian follicle. After adjusting for the image effect, the intraobserver and interobserver coefficient of variation (CV) was calculated.
RESULTS Analysis of variance for all observers’ raw data showed a significant interobserver variation and image effect. Table 1 depicts the F values for images (Fi) and operators (Fo), their ratios, and adjusted coefficients of variation for all indices. 585
The image to image effect was by far the greatest component of the overall variability as demonstrated by the Fi/Fo ratio (range, 2.5 to 327.6). The data were then adjusted for the image effect so that only between- and within-operator variability remained and was reported as the adjusted CV. For the ovary and endometrium, the adjusted CV over the four indices (CDA, MCE, PCE, and ICE/mm2) ranged from 1.9% to 9.5%, whereas for the follicle, the range was 13.0% to 17.7%.
DISCUSSION We believe that this is the first report describing color quantification software for two-dimensional ultrasound machines that measures the above-described power Doppler indices calculated in a specified unit area (square millimeters). This facilitates comparison of vascularity at different stages of the menstrual cycle within and between individuals for a variety of regions of interest. Our results show that although the ROI remained constant over a series of frames, there was a significant image-toimage variation. This is expected in view of the vascular changes taking place during the cardiac cycle over a number of consecutive frames. When the above image effect was adjusted for, there remained a significant effect of intraobserver and interobserver variation. However, the contribution of image effect was by far the greatest. Despite the significant variations, the results showed a low adjusted CV for the endometrial and total ovary indices and an acceptable adjusted CV for follicular indices. This indicates that this technique has the potential to study vascular changes and patterns in both the normal and abnormal state. In addition, we believe that the ICE/mm2 may ultimately prove to be the most useful index because it provides an overall assessment of vascularity incorporating both the power Doppler area and the energy and brightness of the signal within a defined area. We are currently conducting a study to determine the vascular changes during the menstrual, early follicular and mid-follicular, periovulatory, and luteal phases of normally cycling fertile women. After taking into account the intraobserver and interobserver variability, preliminary and limited analysis from the above ongoing study suggests that the changes in vascularity at different stages of the menstrual cycle are of a sufficient magnitude to be distinguished by this technique. Full analysis should enable us to establish any potential clinical value of these indices and their possible use in disease states. It is interesting that our results of combined intraobserver and interobserver variability (adjusted CV) compare favorably with those reported in a similar study (7) assessing intraobserver and interobserver variability of Doppler measurements, which involved observers drawing around the cross-sectional area of the common hepatic artery lumen. 586
Amso et al.
Quantification of power Doppler energy
Future applications may include studies of corpus luteum perfusion in continuing or failing early pregnancy or monitoring placental perfusion in normal or growth-restricted fetuses. In reproductive medicine, vascular studies may include the monitoring of follicular and endometrial changes in women with infertility before and during their treatment. Poor vascularity of mature follicles may be associated with poor oocyte quality, and equally poor vascularity of the endometrium may result in poor receptivity, thus compromising pregnancy rates. There is also the potential to differentiate between benign and malignant ovarian, renal, and hepatic lesions, as well as other conditions in which neovascularization or its absence constitutes a hallmark of the disease process. These potential benefits do not necessarily preclude subjective assessment of these images for features such as those associated with vascular morphology that are not evaluated by this approach of quantification. A different application is the potential for using the software to compare different ultrasound machines with each other. Machines may differ in their sensitivity and accuracy of exhibiting true power Doppler energy, potentially causing differences of opinion between observers using different machines. Video recordings of the same specified area or areas of interest in an individual may be made using two different probes or machines within minutes of each other, and the images are then analyzed as described above. Such comparison may facilitate standardization of this new technology and may have implications for clinical services and research. In summary, we evaluated a computer software program for the quantification of vascularization (amplitudes of the reflected ultrasound waves) in specified unit areas in the uterus and ovary. Calculated indices showed low intraobserver and interobserver CV for the total ovary and endometrium and acceptable levels for small structures such as the follicle. This method of quantification may have the potential to be of clinical use, and its relevance to our practice is currently being explored.
Acknowledgment: The authors thank Mr. A. Barry Nix, Ph.D., medical statistician, University Hospital of Wales, Cardiff, United Kingdom, for his help with the statistical analyses. They also thank Mr. Mark Browne, Ph.D., Kinetic Imaging, Liverpool, United Kingdom, for his continuing assistance with the development of the Color Quantifier Software.
References 1. Murphy KJ, Rubin JM. Power Doppler: it’s a good thing. Semin Ultrasound CT MR 1997;18:13–21. 2. Rubin JM, Bude RO, Carson PL, Bree RL, Adler RS. Power Doppler US: a potentially useful alternative to mean frequency-based color Doppler US. Radiology 1994;190:853– 6. 3. Sonn C, Weskott HP. The sensitivity of new color systems in blood flow
Vol. 76, No. 3, September 2001
diagnosis. The maximum entropy method and angio-color-comparative in vitro flow measurements to determine sensitivity. Surg Endosc 1997; 11:1040 – 4. 4. Bhal PS, Pugh N, Chui D, Gregory L, Walker SM, Shaw RW. The use of transvaginal power Doppler ultrasonography to evaluate the relationship between perifollicular vascularity and outcome in in-vitro fertilisation treatment cycles. Hum Reprod 1999;14:939 – 45. 5. Yang JH, Wu MY, Chen CD, Jiang MC, Ho HN, Yang YS. Association
FERTILITY & STERILITY威
of endometrial blood flow as determined by a modified colour Doppler technique with subsequent outcome of in-vitro fertilization. Hum Reprod 1999;14:1606 –10. 6. Altman DG. Practical statistics for medical research. London, UK: Chapman and Hall, 1999:325–30. 7. Oppo K, Leen E, Angerson WJ, Cooke TG, McArdle CS. Doppler perfusion index: an interobserver and intraobserver reproducibility study. Radiology 1998;208:453–7.
587
Quantification of power Doppler energy and its future potential Nazar N. Amso, Ph.D., Sean R. Watermeyer, M.B.B.Ch., Neil Pugh, Ph.D., Sharon O’Brien, D.M.U., and Arianna D’Angelo, M.D. Department of Obstetrics and Gynecology, University of Wales College of Medicine and the University Hospital of Wales, Cardiff, Wales, United Kingdom
Objective: To test a new software package (Color Quantifier, Kinetic Imaging, Liverpool, United Kingdom) that quantifies power Doppler energy and to determine its reproducibility. Design: Intraobserver and interobserver reproducibility study. Setting: University tertiary referral center. Patient(s): Transvaginal power Doppler ultrasound images were recorded from women taking part in a study evaluating the physiological vascular changes in the uterus and ovaries during the normal menstrual cycle. Intervention(s): Nineteen consecutive frames of regions of interest in the uterus, ovary, and follicle, respectively, were analyzed by each of four observers on 10 occasions. Main Outcome Measure(s): Analysis of variance to determine the image and observer effect as well as the intraobserver and interobserver coefficients of variation. Result(s): Significant image and observer effects were found. However, the image effect was by far the largest component of the total variation. The large image-to-image variability was expected because the cardiac cycle was included within the 19 frames (images) analyzed. The combined intraobserver and interobserver variation, expressed as the coefficients of variation, was found to be small for the above indices (as low as 1.9%), particularly for total ovary and endometrium. Conclusion(s): The indices obtained with this color quantification software are reproducible in an in vitro setting using prerecorded images. Its applicability as a useful assay in the clinical setting requires further evaluation. (Fertil Steril威 2001;76:583–7. ©2001 by American Society for Reproductive Medicine.) Key Words: Power Doppler energy, objective quantification
Received September 22, 2000; revised and accepted March 2, 2001. Supported by the National Assembly, Wales, United Kingdom. Presented at the annual research registrars meeting at the Royal Society of Medicine, London, United Kingdom, February 16, 2001. Reprint requests: Nazar N. Amso, Ph.D., Department of Obstetrics and Gynecology, University Hospital of Wales, Heath Park, Cardiff, CF14 4XW United Kingdom (FAX: 4429-20743722; E-mail: [email protected]). 0015-0282/01/$20.00 PII S0015-0282(01)01940-9
Power Doppler energy (PDE) is a relatively new modality in ultrasound technology that is being used in a number of specialties and has important diagnostic and therapeutic potential (1). PDE uses the amplitude of the Doppler signal as opposed to the mean frequency shift (2). It is advantageous over traditional color Doppler insofar as it is angle independent and there is no problem with aliasing. Background noise, which can present difficulties in evaluation using traditional color Doppler, is easily distinguishable from true flow when using power Doppler (2, 3). PDE has enhanced sensitivity and can therefore be employed to detect image areas of low blood flow that are currently undetectable by color Doppler imaging (2, 3). Such areas include the microvasculature to be found in the kidney, liver, ovary, endometrium, and other tissues of interest. PDE results from the scattering of ultrasound waves
from red blood cells in arterial, arteriolar, venular, and venous blood vessels. Hence it is generally agreed that PDE reflects the vascularity of the tissue in question, at any given time. Quantification of PDE in two-dimensional ultrasound machines has been a subject of interest in recent years and has ranged from subjective methods (4) to semiquantitative techniques (5). We describe and report our assessment of new software (Color Quantifier [CQ], Kinetic Imaging, Liverpool, United Kingdom) that is a personal computer– based system and can be used for the capture and objective quantification of the PDE image. The aim of this study was to determine the intraobserver and interobserver variability for the analysis of color Doppler energy images of pelvic structures (uterus, ovaries, and follicles) 583
FIGURE 1 Ultrasound color power Doppler images. Top left: transvaginal gray-scale image of endometrium. Top right: ROI drawn around endometrium. Bottom left: ROI drawn around ovarian follicle. Bottom right: ROI drawn around ovary. Center: color reference bar.
Amso. Quantification of power Doppler energy. Fertil Steril 2001.
using this new software. This study was approved by the hospital’s ethics committee, and all patients gave written consent before the study was performed.
MATERIALS AND METHODS Software The CQ interprets the scanner’s Doppler color scale and converts the color image data to a numerical array of power values. This new software is capable of both online or deferred analysis from recorded video images. It can be interfaced to videocassette recorders and ultrasound scanners and operates in Microsoft Windows NT so that data can be exported to a statistical package for further analysis. For the purposes of this study, a Toshiba Corevsion (model SSA350A) ultrasound machine was used, together with a Super VHS video tape recorder. Power Doppler settings that 584
Amso et al.
Quantification of power Doppler energy
achieved optimal power Doppler color images with minimal artifact were determined at the outset of the study and remained unaltered throughout all scans during the study period. High-resolution video images were “grabbed” and stored on the hard drive of a personal computer (Pentium II processor at 333 MHz, operating on Windows NT Workstation V4). A calibration process was performed before image analysis. During this process, the following areas were selected: area 1, the reference color bar (Fig. 1); area 2, an area of gray-scale imaging only; area 3, an area incorporating gray scale and power Doppler energy. The software analyzes areas 2 and 3 to assign a value to the gray scale pixels, as well as interpreting the color pixels within area 3 in accordance with the reference color bar. This information was then stored as a calibration file and Vol. 76, No. 3, September 2001
TABLE 1 F Values and adjusted coefficient of variation (includes interobserver and intraobserver variation) after a two-way (image and operator) analysis of variance. Parameter Ovary CDA MCE PCE ICE/mm2 Endometrium CDA MCE PCE ICE/mm2 Follicle CDA MCE PCE ICE/mm2
Fvalue image (Fi)
Fvalue operator (Fo)
Ratio FI: Fo
Adjusted coefficient of variation (%)
187.1 1068.7 145.5 733.0
19.7 13.3 10.8 211.4
9.5 80.4 13.5 3.46
2.5 2.3 1.9 2.6
394.7 668.9 922.8 941.5
159.0 92.9 2.8 79.72
2.48 7.2 327.6 11.8
9.5 8.1 6.0 5.3
364.7 376.6 106.7 407.8
76.6 44.9 18.5 7.7
4.8 8.4 5.8 53.1
13.0 17.7 16.4 13.0
Note: All F values are significant at 5% significance level (P ⬍ .05). Amso. Quantification of power Doppler energy. Fertil Steril 2001.
used for subsequent image analyses. After this, regions of interest (ROIs) were identified and selected. Regions of interest may be of any shape and were literally drawn around the areas for analysis (Fig. 1). One or more ROIs may be drawn at any one time. It was also possible to exclude areas within a ROI from the analysis, such as fluid areas within cysts, follicles or tumors. The software has been developed to calculate the following indices; [1] the area of the ROI in square millimeters, [2] the area within the ROI occupied by the color pixels—the color Doppler area (CDA), [3] the mean color energy (MCE), [4] the peak color energy (PCE), and [5] the integrated color energy per square millimeter (ICE/mm2). The ICE/mm2 is a measure of the overall intensity of the PDE signal that incorporates not only the CDA but also the summated energy within the CDA in any given region of interest. This represents a true reflection of vascular density or vascularity within the tissue being analyzed at any given time. All of the above indices were calculated for the ROI studied.
Subjects The software analysis was evaluated for its reproducibility within and between observers of a multidisciplinary team (gynecologists, physicist, sonographer) working in the assisted reproduction unit within our institution. Video recordings of endometrium, ovary, and ovarian follicle were chosen for the analysis. The images were recorded from transvaginal power Doppler ultrasound scans of women taking part in a study evaluating the physiological vascular changes in the uterus and ovaries during the normal menstrual cycle. The transvaginal route was chosen because of its superior image quality. FERTILITY & STERILITY威
From each recording, 19 consecutive frames (images) of approximately 5 seconds’ duration were selected on account of the gray scale and power Doppler clarity. Each of the four analysts then independently evaluated these same images on 10 occasions, attempting to redraw the same ROI on each occasion. The ROIs and the methodology in drawing the ROIs were determined before commencing the study. The drawn ROI remained constant throughout the 19 frames of each single analysis. To ensure the appropriateness of the ROI positioning, it was possible to preview the drawn ROI superimposed on all frames to be analyzed.
Statistical Analysis After analysis, the numerical data of the above power Doppler indices (CDA, MCE, PCE, and ICE/mm2) were exported to a statistical package, namely Microsoft Excel and SPSS version 10 (SPSS, UK Limited, Surrey, UK). An analysis of variance, in which the effects were assessed by the F value (6), was carried out to determine variation between and within observers and variation between images for each of the indices (CDA, MCE, PCE, and ICE/mm2). This was repeated for the endometrium, ovary, and ovarian follicle. After adjusting for the image effect, the intraobserver and interobserver coefficient of variation (CV) was calculated.
RESULTS Analysis of variance for all observers’ raw data showed a significant interobserver variation and image effect. Table 1 depicts the F values for images (Fi) and operators (Fo), their ratios, and adjusted coefficients of variation for all indices. 585
The image to image effect was by far the greatest component of the overall variability as demonstrated by the Fi/Fo ratio (range, 2.5 to 327.6). The data were then adjusted for the image effect so that only between- and within-operator variability remained and was reported as the adjusted CV. For the ovary and endometrium, the adjusted CV over the four indices (CDA, MCE, PCE, and ICE/mm2) ranged from 1.9% to 9.5%, whereas for the follicle, the range was 13.0% to 17.7%.
DISCUSSION We believe that this is the first report describing color quantification software for two-dimensional ultrasound machines that measures the above-described power Doppler indices calculated in a specified unit area (square millimeters). This facilitates comparison of vascularity at different stages of the menstrual cycle within and between individuals for a variety of regions of interest. Our results show that although the ROI remained constant over a series of frames, there was a significant image-toimage variation. This is expected in view of the vascular changes taking place during the cardiac cycle over a number of consecutive frames. When the above image effect was adjusted for, there remained a significant effect of intraobserver and interobserver variation. However, the contribution of image effect was by far the greatest. Despite the significant variations, the results showed a low adjusted CV for the endometrial and total ovary indices and an acceptable adjusted CV for follicular indices. This indicates that this technique has the potential to study vascular changes and patterns in both the normal and abnormal state. In addition, we believe that the ICE/mm2 may ultimately prove to be the most useful index because it provides an overall assessment of vascularity incorporating both the power Doppler area and the energy and brightness of the signal within a defined area. We are currently conducting a study to determine the vascular changes during the menstrual, early follicular and mid-follicular, periovulatory, and luteal phases of normally cycling fertile women. After taking into account the intraobserver and interobserver variability, preliminary and limited analysis from the above ongoing study suggests that the changes in vascularity at different stages of the menstrual cycle are of a sufficient magnitude to be distinguished by this technique. Full analysis should enable us to establish any potential clinical value of these indices and their possible use in disease states. It is interesting that our results of combined intraobserver and interobserver variability (adjusted CV) compare favorably with those reported in a similar study (7) assessing intraobserver and interobserver variability of Doppler measurements, which involved observers drawing around the cross-sectional area of the common hepatic artery lumen. 586
Amso et al.
Quantification of power Doppler energy
Future applications may include studies of corpus luteum perfusion in continuing or failing early pregnancy or monitoring placental perfusion in normal or growth-restricted fetuses. In reproductive medicine, vascular studies may include the monitoring of follicular and endometrial changes in women with infertility before and during their treatment. Poor vascularity of mature follicles may be associated with poor oocyte quality, and equally poor vascularity of the endometrium may result in poor receptivity, thus compromising pregnancy rates. There is also the potential to differentiate between benign and malignant ovarian, renal, and hepatic lesions, as well as other conditions in which neovascularization or its absence constitutes a hallmark of the disease process. These potential benefits do not necessarily preclude subjective assessment of these images for features such as those associated with vascular morphology that are not evaluated by this approach of quantification. A different application is the potential for using the software to compare different ultrasound machines with each other. Machines may differ in their sensitivity and accuracy of exhibiting true power Doppler energy, potentially causing differences of opinion between observers using different machines. Video recordings of the same specified area or areas of interest in an individual may be made using two different probes or machines within minutes of each other, and the images are then analyzed as described above. Such comparison may facilitate standardization of this new technology and may have implications for clinical services and research. In summary, we evaluated a computer software program for the quantification of vascularization (amplitudes of the reflected ultrasound waves) in specified unit areas in the uterus and ovary. Calculated indices showed low intraobserver and interobserver CV for the total ovary and endometrium and acceptable levels for small structures such as the follicle. This method of quantification may have the potential to be of clinical use, and its relevance to our practice is currently being explored.
Acknowledgment: The authors thank Mr. A. Barry Nix, Ph.D., medical statistician, University Hospital of Wales, Cardiff, United Kingdom, for his help with the statistical analyses. They also thank Mr. Mark Browne, Ph.D., Kinetic Imaging, Liverpool, United Kingdom, for his continuing assistance with the development of the Color Quantifier Software.
References 1. Murphy KJ, Rubin JM. Power Doppler: it’s a good thing. Semin Ultrasound CT MR 1997;18:13–21. 2. Rubin JM, Bude RO, Carson PL, Bree RL, Adler RS. Power Doppler US: a potentially useful alternative to mean frequency-based color Doppler US. Radiology 1994;190:853– 6. 3. Sonn C, Weskott HP. The sensitivity of new color systems in blood flow
Vol. 76, No. 3, September 2001
diagnosis. The maximum entropy method and angio-color-comparative in vitro flow measurements to determine sensitivity. Surg Endosc 1997; 11:1040 – 4. 4. Bhal PS, Pugh N, Chui D, Gregory L, Walker SM, Shaw RW. The use of transvaginal power Doppler ultrasonography to evaluate the relationship between perifollicular vascularity and outcome in in-vitro fertilisation treatment cycles. Hum Reprod 1999;14:939 – 45. 5. Yang JH, Wu MY, Chen CD, Jiang MC, Ho HN, Yang YS. Association
FERTILITY & STERILITY威
of endometrial blood flow as determined by a modified colour Doppler technique with subsequent outcome of in-vitro fertilization. Hum Reprod 1999;14:1606 –10. 6. Altman DG. Practical statistics for medical research. London, UK: Chapman and Hall, 1999:325–30. 7. Oppo K, Leen E, Angerson WJ, Cooke TG, McArdle CS. Doppler perfusion index: an interobserver and intraobserver reproducibility study. Radiology 1998;208:453–7.
587