- Email: [email protected]
Effects of Lesion Positioning on Digital Magnification Mammography Performance
Technical Report
Effects of Lesion Positioning on Digital Magnification Mammography Performance Franklin Liu, MD, Kalpana M. Kanal, PhD, Brent K. Stewart, PhD, Constance D. Lehman, MD, PhD Rationale and Objectives: We undertook this study to determine whether differences in detector-lesion distance resulted in appreciable effects on digital magnification mammography performance as measured using the American College of Radiology (ACR) mammography phantom and a line pair test pattern. Materials and Methods: Images of the standard 42-mm thick standard ACR mammography phantom with a wax insert on one side containing simulated fibers, calcifications, and masses were obtained on a Senographe Essential digital mammography system with the phantom in upright and inverted positions. The process was repeated with a line pair test pattern for measuring resolution. All images were obtained in contact mode, and with 1.5 and 1.8 magnification, and evaluated on a GE PACS monitor. Results: Overall, changing lesion-detector distance using standard versus inverted positioning did not appreciably increase the number of objects seen on the ACR phantom under all modes. No greater than one line pair difference was seen in standard versus inverted positioning. At 1.8 magnification mode, no difference was detected in line pair resolution with a change in positioning. Conclusion: Differences in lesion-detector distance as modeled using both the ACR mammography phantom and a line pair test pattern did not make an appreciable difference in digital magnification mammography performance. Key Words: Breast imaging; mammography; lesion positioning. ªAUR, 2010
M
agnification mammography works through the principle of geometric magnification, wherein the breast is placed on a support platform at a fixed position above the detector, so that there is an air gap between the breast and the detector. The rate of magnification is the ratio of the source-to-image (x-ray tube to detector) distance to source-to-object (x-ray tube to breast lesion) distance. Increased magnification allows for increased effective resolution, thereby improving lesion detection and characterization (1). The most significant limitation of magnification is geometric blurring caused by finite focal spot size. Larger focal spots and higher magnification increases blurring, which causes a loss of resolution in the projected image. To maintain resolution, a smaller focal spot is used, typically 0.1 mm as opposed to the 0.3 mm focal spot for standard contact mammography. The use of smaller focal spots has been shown
Acad Radiol 2010; 17:791–794 From the Seattle Cancer Care Alliance, Department of Radiology, 825 Eastlake Avenue, Seattle, WA 98109 (F.L., C.D.L.); University of Washington Medical Center, Department of Radiology, Seattle, WA (K.M.K., B.K.S., C.D.L.) Received October 27, 2009; accepted February 17, 2010. Address correspondence to: F.L. e-mail: [email protected] ªAUR, 2010 doi:10.1016/j.acra.2010.02.007
to improve image quality in magnification mammography (2,3). Magnification mammography is a well-established technique for the evaluation of potential breast lesions identified on standard views. Given the degree of rotational freedom allowed by modern digital mammography units, the radiologist or radiologic technologist often has a choice with regards to the magnification view obtained and thus the lesiondetector distance. For example, a lesion in the lower breast can be further evaluated either in a magnified craniocaudal view or caudocranial view. With the former view, the lesion in question would be slightly further from the detector, with a slight corresponding increase in both magnification and geometric blur, whereas in the latter view, the lesion would be slightly closer to the detector, with less magnification and blur. Minimizing lesion-detector distance has been an established recommendation for optimizing mammographic imaging (4). However, to date, there has been no published data regarding the effect, if any, of lesion positioning on magnification mammography performance. We undertook this study to determine whether differences in detectorlesion distance resulted in appreciable effects on digital magnification mammography performance, as measured using the American College of Radiology (ACR) mammography phantom and a line pair test pattern. 791
LIU ET AL
Academic Radiology, Vol 17, No 6, June 2010
Figure 1. (a) Standard positioning: configuration with placement of American College of Radiology (ACR) insert further from the detector. (b) Inverted positioning: configuration with placement of ACR insert closer to the detector.
Figure 2. (a) Example of standard positioning of the American College of Radiology (ACR) insert on 1.5 magnification with positioning further from detector. (b) Example of inverted positioning of the inverted ACR insert on 1.5 magnification with positioning closer to detector. Note that for the purposes of blinded data gathering, this acquired image was later inverted and rotated after acquisition to match up to Figure 2a.
MATERIALS AND METHODS A 42-mm thick standard ACR mammography phantom (Gammex RMI, model 156) with a wax insert on one side containing five masses, five groups of calcifications, and six fibers was imaged with a digital mammography system (Senographe Essential, GE Healthcare, Waukesha, WI). Images were first obtained under standard positioning with the insert further from the detector (Fig 1a), as in quality assurance testing, then under inverted positioning so the insert was closer to the detector (Fig 1b). This procedure with then repeated with 1.5 magnification (Figs 2a, b) and 1.8 magnification using the corresponding standard support platforms. The entire process was then repeated with a line pair test pattern used for measuring resolution ranging between 5 and 20 line pairs/mm (Fluke Biomedical, Everett, WA) and 42 mm of acrylic to mimic the ACR phantom thickness (Figs 3a–3d). All images were obtained using automatic exposure control. The test pattern was angled at 45 to minimize the effect of aliasing. Focal spot sizes for the Essential system are 0.3 mm for large focal spot (contact mammography) and 0.1 mm for small focal spot (both 1.5 and 1.8 magnification modes). All exposures ranged from 29 to 31 kVp and 42 to 56 mAs. All images were independently evaluated by five board-certified radiologists specializing in breast imaging on a GE PACS monitor 792
with a resolution of 2048 2560 pixels. The radiologists were blinded before reviewing the images. Before blinded evaluation, images were inverted and rotated as necessary to maintain similar orientation throughout. All readers were allowed to adjust window level or magnification to their preference to optimize their evaluation.
RESULTS Results are summarized in Table 1. With the exception of radiologist #4, all radiologists identified four masses of the ACR insert in contact mammography mode, and five masses in 1.5 and 1.8 mammography. Radiologist #4 identified four masses in all modes. There was no difference in masses detected under standard or inverted positioning, regardless of magnification mode. Radiologists #1, #4, and #5 identified five fibers of the ACR insert in all modes, with no difference in fibers detected under standard or inverted positioning. Radiologist #2 detected six fibers at 1.8 in both positions and 1.5 under the standard position, but only five fibers at 1.5 under the inverted position. In contrast, radiologist #3 detected five fibers in all modes regardless of positioning, with the exception of 1.8 at standard positioning, under which six fibers were seen.
Academic Radiology, Vol 17, No 6, June 2010
EFFECTS OF LESION POSITIONING ON MAMMOGRAPHY
Figures 3. (a, c) Standard positioning of the line pair test pattern with 42 mm of acrylic under contact and 1.8 magnification mammography, respectively. No edge or image enhancement software was applied. (b, d) Inverted positioning of the line pair test pattern with 42 mm of acrylic under contact and 1.8 magnification mammography, respectively. Note that for the purposes of blinded data gathering, these acquired images were later inverted and rotated after acquisition to match up to Figures 3a and c.
All radiologists identified four total groups of calcifications regardless of magnification mode or positioning. Each radiologist identified increased line pair for spatial resolution detection with increasing magnification. Although values were slightly variable across radiologists, ranging from 6 to 8 line pairs in contact mode, 8 to 11 line pairs in 1.5 mode, and 10 to 13 line pairs in 1.8 mode, no radiologist detected a greater than one line pair difference between standard or inverted positioning. Of note, inversion of the ACR phantom did lead to slight minification. For example, measurement of length of the most prominent fiber on the ACR phantom decreased from 1.1 cm with the phantom placed upright (insert further away from the detector) to 1.0 cm with the phantom inverted (insert closer to detector).
DISCUSSION After a potential lesion is identified and localized on mammogram, further workup with magnification mammography is often the next step. For any given position of the breast during mammography, there is a pair of diametric views (ie, craniocaudal/caudocranial, mediolateral/lateromedial, mediolateral oblique/lateromedial oblique, and superoinferior oblique/ inferosuperior oblique). One of the pair will usually result in the lesion being closer to the detector, whereas the other results in the lesion being further away from the detector. We modeled the opposing view by inverting our phantoms, reasoning that this maneuver was equivalent to rotating the entire cathode-detector mammography unit 180 degrees.
This was confirmed on initial experiments (images not shown). In general, although magnification did not appear to substantially increase the number of objects seen on the ACR phantom, it did improve resolution as measured by the line pair test pattern, as expected. Overall, changing lesion-detector distance using standard versus inverted positioning did not appreciably increase the number of objects seen on the ACR phantom under all modes. There was more inter-observer variability regarding line pair resolution, likely from the increased sensitivity of the line pair test pattern to changes in resolution. Nonetheless, there was no greater than one line pair difference between standard versus inverted positioning. At 1.8 mode, no difference was noted in line pair resolution with a change in positioning. Given that the GE digital mammography system has a resolution of 100 microns, it would seem that at the maximum 1.8 magnification, lesion positioning would result in a visible effect on penumbra (Table 2). This difference may be visible on digital mammography systems with a 70- or 50micron resolution when using magnification. Because filmscreen systems have a resolution of up to 20 line pair resolution, which is approximately a 25-micron resolution, these differences might be also detectable with film. Surprisingly, this increase in geometric unsharpness may not be clinically significant, or may be offset by the gain in magnification. Altogether, these findings suggest that lesion positioning may not impact digital mammography performance. Of note, although a resolution of 100 microns would suggest a theoretical maximal detection of 5 lp/mm on standard contact digital mammography without magnification, angling of the line pair test pattern allows an increase in the 793
LIU ET AL
Academic Radiology, Vol 17, No 6, June 2010
TABLE 1. Radiologist Detection of the Objects (Masses, Fibers, Calcifications) within the American College of Radiology Mammography Phantom and Maximum Line Pair Resolution Radiologist
Total # of observed masses
Total # of observed fibers
Total # of observed calcifications
Spatial resolution (lp/mm)
Contact standard/inverted 1.5 standard/inverted 1.8 standard/inverted Contact standard/inverted 1.5 standard/inverted 1.8 standard/inverted Contact standard/inverted 1.5 standard/inverted 1.8 standard/inverted Contact standard/inverted 1.5 standard/inverted 1.8 standard/inverted
1
2
3
4
5
4/4 5/5 5/5 5/5 5/5 5/5 4/4 4/4 4/4 8/8 11/10 13/13
4/4 5/5 5/5 5/5 6/5 6/6 4/4 4/4 4/4 7/6 9/8 10/10
4/4 5/5 5/5 5/5 5/5 6/5 4/4 4/4 4/4 8/7 11/10 12/12
4/4 4/4 4/4 5/5 5/5 5/5 4/4 4/4 4/4 8/8 11/10 12/12
4/4 5/5 5/5 5/5 5/5 5/5 4/4 4/4 4/4 8/8 10/10 12/12
TABLE 2. Calculation of Penumbra Magnification Contact mode 1.5 1.8
Penumbra*
Focal Spot Size 0.3 mm 0.1 mm 0.1 mm
Object Closer to Detector (Inverted)
Object Farther from Detector (Standard)
11 microns 58 microns 92 microns
33 microns 76 microns 149 microns
*Calculated using the formula: Penumbra = focal spot size (distance from object to detector)/(distance from x-ray source to object).
sampling limit, thereby increasing detectable limiting resolution. For an angling of 45 , this theoretically increases the sampled limit from 5 lp/mm to approximately 7 lp/mm (by a factor of the square root of 2) on contact mammography, and from 9 lp/mm to 12.7 lp/mm on 1.8 magnification mode (Souchay H, PhD, Senior Engineer, GE Healthcare, personal communications, 2010). The fact that up to 8 and 13 lp/mm were detected on contact and 1.8 magnification mode, respectively, is likely the result of combined effects of inter-observer variability and aliasing. Scattered radiation degrades image contrast and, in mammography, increases with breast thickness and area. Using an antiscatter grid in contact mammography, for example, lowers scatter by allowing a greater proportion of primary photons to reach the detector. Magnification mammography uses the technique of spot compression to reduce motion and scatter by reducing breast thickness and area. The additional use of an air gap decreases scatter as the higher incident angles of scattered photons result in missing the detector. This, however, is offset by the increase in geometric unsharpness reducing resolution, as mentioned previously. Taken altogether, these factors (antiscatter grid, spot compression, air gap) in reducing scatter may play a larger role in digital mammography performance than lesion positioning.
794
We chose to use the ACR-approved mammography breast phantom for this study because it represents a standardized, widely accepted form of quality control. Given that actual breast tissue was not used, one of the limitations of this study is that background tissue density in the actual breast may obscure lesions and affect their characterization. Another limitation is our use of the ACR phantom to simulate the average width of the compressed breast. It may be possible that at some breast thickness greater than 42 mm, blur might reduce resolution more than the gain provided by magnification, or vice versa. Future studies could be performed with other available phantoms, which may be more sensitive to smaller changes in detection differences.
REFERENCES 1. Bushberg JT, Seibert JA, Leidholdt E, et al. The essential physics of medical imaging. 2nd ed. Philadelphia, PA: Lippincott Williams & Wilkins, 2002. 2. Koutalonis M, Delis H, Spyrou G, et al. Monte Carlo studies on the influence of focal spot size and intensity distribution on spatial resolution in magnification mammography. Phys Med Biol 2008; 53:1369–1384. 3. Boyce SJ, Samei E. Imaging properties of digital magnification radiography. Med Phys 2006; 33:984–996. 4. Eklund GW, Cardenosa G. The art of mammographic positioning. Radiol Clin North Am 1992; 30:21–53.
Effects of Lesion Positioning on Digital Magnification Mammography Performance Franklin Liu, MD, Kalpana M. Kanal, PhD, Brent K. Stewart, PhD, Constance D. Lehman, MD, PhD Rationale and Objectives: We undertook this study to determine whether differences in detector-lesion distance resulted in appreciable effects on digital magnification mammography performance as measured using the American College of Radiology (ACR) mammography phantom and a line pair test pattern. Materials and Methods: Images of the standard 42-mm thick standard ACR mammography phantom with a wax insert on one side containing simulated fibers, calcifications, and masses were obtained on a Senographe Essential digital mammography system with the phantom in upright and inverted positions. The process was repeated with a line pair test pattern for measuring resolution. All images were obtained in contact mode, and with 1.5 and 1.8 magnification, and evaluated on a GE PACS monitor. Results: Overall, changing lesion-detector distance using standard versus inverted positioning did not appreciably increase the number of objects seen on the ACR phantom under all modes. No greater than one line pair difference was seen in standard versus inverted positioning. At 1.8 magnification mode, no difference was detected in line pair resolution with a change in positioning. Conclusion: Differences in lesion-detector distance as modeled using both the ACR mammography phantom and a line pair test pattern did not make an appreciable difference in digital magnification mammography performance. Key Words: Breast imaging; mammography; lesion positioning. ªAUR, 2010
M
agnification mammography works through the principle of geometric magnification, wherein the breast is placed on a support platform at a fixed position above the detector, so that there is an air gap between the breast and the detector. The rate of magnification is the ratio of the source-to-image (x-ray tube to detector) distance to source-to-object (x-ray tube to breast lesion) distance. Increased magnification allows for increased effective resolution, thereby improving lesion detection and characterization (1). The most significant limitation of magnification is geometric blurring caused by finite focal spot size. Larger focal spots and higher magnification increases blurring, which causes a loss of resolution in the projected image. To maintain resolution, a smaller focal spot is used, typically 0.1 mm as opposed to the 0.3 mm focal spot for standard contact mammography. The use of smaller focal spots has been shown
Acad Radiol 2010; 17:791–794 From the Seattle Cancer Care Alliance, Department of Radiology, 825 Eastlake Avenue, Seattle, WA 98109 (F.L., C.D.L.); University of Washington Medical Center, Department of Radiology, Seattle, WA (K.M.K., B.K.S., C.D.L.) Received October 27, 2009; accepted February 17, 2010. Address correspondence to: F.L. e-mail: [email protected] ªAUR, 2010 doi:10.1016/j.acra.2010.02.007
to improve image quality in magnification mammography (2,3). Magnification mammography is a well-established technique for the evaluation of potential breast lesions identified on standard views. Given the degree of rotational freedom allowed by modern digital mammography units, the radiologist or radiologic technologist often has a choice with regards to the magnification view obtained and thus the lesiondetector distance. For example, a lesion in the lower breast can be further evaluated either in a magnified craniocaudal view or caudocranial view. With the former view, the lesion in question would be slightly further from the detector, with a slight corresponding increase in both magnification and geometric blur, whereas in the latter view, the lesion would be slightly closer to the detector, with less magnification and blur. Minimizing lesion-detector distance has been an established recommendation for optimizing mammographic imaging (4). However, to date, there has been no published data regarding the effect, if any, of lesion positioning on magnification mammography performance. We undertook this study to determine whether differences in detectorlesion distance resulted in appreciable effects on digital magnification mammography performance, as measured using the American College of Radiology (ACR) mammography phantom and a line pair test pattern. 791
LIU ET AL
Academic Radiology, Vol 17, No 6, June 2010
Figure 1. (a) Standard positioning: configuration with placement of American College of Radiology (ACR) insert further from the detector. (b) Inverted positioning: configuration with placement of ACR insert closer to the detector.
Figure 2. (a) Example of standard positioning of the American College of Radiology (ACR) insert on 1.5 magnification with positioning further from detector. (b) Example of inverted positioning of the inverted ACR insert on 1.5 magnification with positioning closer to detector. Note that for the purposes of blinded data gathering, this acquired image was later inverted and rotated after acquisition to match up to Figure 2a.
MATERIALS AND METHODS A 42-mm thick standard ACR mammography phantom (Gammex RMI, model 156) with a wax insert on one side containing five masses, five groups of calcifications, and six fibers was imaged with a digital mammography system (Senographe Essential, GE Healthcare, Waukesha, WI). Images were first obtained under standard positioning with the insert further from the detector (Fig 1a), as in quality assurance testing, then under inverted positioning so the insert was closer to the detector (Fig 1b). This procedure with then repeated with 1.5 magnification (Figs 2a, b) and 1.8 magnification using the corresponding standard support platforms. The entire process was then repeated with a line pair test pattern used for measuring resolution ranging between 5 and 20 line pairs/mm (Fluke Biomedical, Everett, WA) and 42 mm of acrylic to mimic the ACR phantom thickness (Figs 3a–3d). All images were obtained using automatic exposure control. The test pattern was angled at 45 to minimize the effect of aliasing. Focal spot sizes for the Essential system are 0.3 mm for large focal spot (contact mammography) and 0.1 mm for small focal spot (both 1.5 and 1.8 magnification modes). All exposures ranged from 29 to 31 kVp and 42 to 56 mAs. All images were independently evaluated by five board-certified radiologists specializing in breast imaging on a GE PACS monitor 792
with a resolution of 2048 2560 pixels. The radiologists were blinded before reviewing the images. Before blinded evaluation, images were inverted and rotated as necessary to maintain similar orientation throughout. All readers were allowed to adjust window level or magnification to their preference to optimize their evaluation.
RESULTS Results are summarized in Table 1. With the exception of radiologist #4, all radiologists identified four masses of the ACR insert in contact mammography mode, and five masses in 1.5 and 1.8 mammography. Radiologist #4 identified four masses in all modes. There was no difference in masses detected under standard or inverted positioning, regardless of magnification mode. Radiologists #1, #4, and #5 identified five fibers of the ACR insert in all modes, with no difference in fibers detected under standard or inverted positioning. Radiologist #2 detected six fibers at 1.8 in both positions and 1.5 under the standard position, but only five fibers at 1.5 under the inverted position. In contrast, radiologist #3 detected five fibers in all modes regardless of positioning, with the exception of 1.8 at standard positioning, under which six fibers were seen.
Academic Radiology, Vol 17, No 6, June 2010
EFFECTS OF LESION POSITIONING ON MAMMOGRAPHY
Figures 3. (a, c) Standard positioning of the line pair test pattern with 42 mm of acrylic under contact and 1.8 magnification mammography, respectively. No edge or image enhancement software was applied. (b, d) Inverted positioning of the line pair test pattern with 42 mm of acrylic under contact and 1.8 magnification mammography, respectively. Note that for the purposes of blinded data gathering, these acquired images were later inverted and rotated after acquisition to match up to Figures 3a and c.
All radiologists identified four total groups of calcifications regardless of magnification mode or positioning. Each radiologist identified increased line pair for spatial resolution detection with increasing magnification. Although values were slightly variable across radiologists, ranging from 6 to 8 line pairs in contact mode, 8 to 11 line pairs in 1.5 mode, and 10 to 13 line pairs in 1.8 mode, no radiologist detected a greater than one line pair difference between standard or inverted positioning. Of note, inversion of the ACR phantom did lead to slight minification. For example, measurement of length of the most prominent fiber on the ACR phantom decreased from 1.1 cm with the phantom placed upright (insert further away from the detector) to 1.0 cm with the phantom inverted (insert closer to detector).
DISCUSSION After a potential lesion is identified and localized on mammogram, further workup with magnification mammography is often the next step. For any given position of the breast during mammography, there is a pair of diametric views (ie, craniocaudal/caudocranial, mediolateral/lateromedial, mediolateral oblique/lateromedial oblique, and superoinferior oblique/ inferosuperior oblique). One of the pair will usually result in the lesion being closer to the detector, whereas the other results in the lesion being further away from the detector. We modeled the opposing view by inverting our phantoms, reasoning that this maneuver was equivalent to rotating the entire cathode-detector mammography unit 180 degrees.
This was confirmed on initial experiments (images not shown). In general, although magnification did not appear to substantially increase the number of objects seen on the ACR phantom, it did improve resolution as measured by the line pair test pattern, as expected. Overall, changing lesion-detector distance using standard versus inverted positioning did not appreciably increase the number of objects seen on the ACR phantom under all modes. There was more inter-observer variability regarding line pair resolution, likely from the increased sensitivity of the line pair test pattern to changes in resolution. Nonetheless, there was no greater than one line pair difference between standard versus inverted positioning. At 1.8 mode, no difference was noted in line pair resolution with a change in positioning. Given that the GE digital mammography system has a resolution of 100 microns, it would seem that at the maximum 1.8 magnification, lesion positioning would result in a visible effect on penumbra (Table 2). This difference may be visible on digital mammography systems with a 70- or 50micron resolution when using magnification. Because filmscreen systems have a resolution of up to 20 line pair resolution, which is approximately a 25-micron resolution, these differences might be also detectable with film. Surprisingly, this increase in geometric unsharpness may not be clinically significant, or may be offset by the gain in magnification. Altogether, these findings suggest that lesion positioning may not impact digital mammography performance. Of note, although a resolution of 100 microns would suggest a theoretical maximal detection of 5 lp/mm on standard contact digital mammography without magnification, angling of the line pair test pattern allows an increase in the 793
LIU ET AL
Academic Radiology, Vol 17, No 6, June 2010
TABLE 1. Radiologist Detection of the Objects (Masses, Fibers, Calcifications) within the American College of Radiology Mammography Phantom and Maximum Line Pair Resolution Radiologist
Total # of observed masses
Total # of observed fibers
Total # of observed calcifications
Spatial resolution (lp/mm)
Contact standard/inverted 1.5 standard/inverted 1.8 standard/inverted Contact standard/inverted 1.5 standard/inverted 1.8 standard/inverted Contact standard/inverted 1.5 standard/inverted 1.8 standard/inverted Contact standard/inverted 1.5 standard/inverted 1.8 standard/inverted
1
2
3
4
5
4/4 5/5 5/5 5/5 5/5 5/5 4/4 4/4 4/4 8/8 11/10 13/13
4/4 5/5 5/5 5/5 6/5 6/6 4/4 4/4 4/4 7/6 9/8 10/10
4/4 5/5 5/5 5/5 5/5 6/5 4/4 4/4 4/4 8/7 11/10 12/12
4/4 4/4 4/4 5/5 5/5 5/5 4/4 4/4 4/4 8/8 11/10 12/12
4/4 5/5 5/5 5/5 5/5 5/5 4/4 4/4 4/4 8/8 10/10 12/12
TABLE 2. Calculation of Penumbra Magnification Contact mode 1.5 1.8
Penumbra*
Focal Spot Size 0.3 mm 0.1 mm 0.1 mm
Object Closer to Detector (Inverted)
Object Farther from Detector (Standard)
11 microns 58 microns 92 microns
33 microns 76 microns 149 microns
*Calculated using the formula: Penumbra = focal spot size (distance from object to detector)/(distance from x-ray source to object).
sampling limit, thereby increasing detectable limiting resolution. For an angling of 45 , this theoretically increases the sampled limit from 5 lp/mm to approximately 7 lp/mm (by a factor of the square root of 2) on contact mammography, and from 9 lp/mm to 12.7 lp/mm on 1.8 magnification mode (Souchay H, PhD, Senior Engineer, GE Healthcare, personal communications, 2010). The fact that up to 8 and 13 lp/mm were detected on contact and 1.8 magnification mode, respectively, is likely the result of combined effects of inter-observer variability and aliasing. Scattered radiation degrades image contrast and, in mammography, increases with breast thickness and area. Using an antiscatter grid in contact mammography, for example, lowers scatter by allowing a greater proportion of primary photons to reach the detector. Magnification mammography uses the technique of spot compression to reduce motion and scatter by reducing breast thickness and area. The additional use of an air gap decreases scatter as the higher incident angles of scattered photons result in missing the detector. This, however, is offset by the increase in geometric unsharpness reducing resolution, as mentioned previously. Taken altogether, these factors (antiscatter grid, spot compression, air gap) in reducing scatter may play a larger role in digital mammography performance than lesion positioning.
794
We chose to use the ACR-approved mammography breast phantom for this study because it represents a standardized, widely accepted form of quality control. Given that actual breast tissue was not used, one of the limitations of this study is that background tissue density in the actual breast may obscure lesions and affect their characterization. Another limitation is our use of the ACR phantom to simulate the average width of the compressed breast. It may be possible that at some breast thickness greater than 42 mm, blur might reduce resolution more than the gain provided by magnification, or vice versa. Future studies could be performed with other available phantoms, which may be more sensitive to smaller changes in detection differences.
REFERENCES 1. Bushberg JT, Seibert JA, Leidholdt E, et al. The essential physics of medical imaging. 2nd ed. Philadelphia, PA: Lippincott Williams & Wilkins, 2002. 2. Koutalonis M, Delis H, Spyrou G, et al. Monte Carlo studies on the influence of focal spot size and intensity distribution on spatial resolution in magnification mammography. Phys Med Biol 2008; 53:1369–1384. 3. Boyce SJ, Samei E. Imaging properties of digital magnification radiography. Med Phys 2006; 33:984–996. 4. Eklund GW, Cardenosa G. The art of mammographic positioning. Radiol Clin North Am 1992; 30:21–53.