220
TECHNICAL NOTES
An Automated Ultrasound Transducer Beam Profiling System 1 Mitchell M. Goodsitt, M.S., Richard A. Banjavic, Ph.D., James A. Zagzebski, Ph.D., and Ernest L. Madsen, Ph.D. An automatic, three-dimensional, ultrasound beam profiling system was developed and incorporated into a radiotherapy planning minicomputer system with few modifications and at minimal cost. This profiler was invaluable in assessing the characteristics of beams emitted by ultrasound transducers. INDEX TERMS:
Computers. Ultrasound, instrumentation
Radiology 132:220-222, July 1979
A transducer usually comes equipped with a data sheet containing its characteristics, including real time r.f. wave form, frequency spectrum, and pulse-echo beam profiles. The latter are often linear beam profiles at different depths in water, usually obtained by plotting the lateral transducer position versus the relative amplitude of the signal reflected from a small steel or nylon rod at a given depth as the transducer scans across the rod. These linear profiles yield only two-dimensional information about three-dimensional ultrasound beams. The imaging capabilities of an ultrasound transducer are highly dependent upon the three-dimensional nature of the beams emitted and received. Any asymmetries in the transducer beam, either as a result of a flaw in the construction of the transducer or of transmission through tissues will affect the images produced. The beam profiles provided by the manufacturer are deficient in that properties in only one of many directions at a given depth are depicted, and the effects of transmission through tissues are not considered. To obtain a better understanding of the spatial pulse-echo response of transducers, three-dimensional ultrasound beam profilers have been built (1-4). Such devices typically facilitate the positioning of a target at known points within the ultrasound beam emitted from a stationary transducer. Data consisting of transducer-detected peak echo amplitudes, associated with different target positions in a plane parallel to the transducer face, can be used to plot transducer "isoresponse contours". These contours are analogous to the isodose contours in radiotherapy planning, each representing points in the plane parallel to the transducer face from which equal amplitude echoes are obtained. A beam profile, or more specifically, a planar beam profile is made up of a collection of these contours. The first three-dimensional profiler employed in our studies was a modified stereotaxic device which was manually operated. With this profiler, it was discovered that ultrasound beam profiles obtained with attenuating media such as castor oil between the transducer face and target showed greater beam divergence or spreading than those obtained in nonattenuating media such as water. It was also found that isoresponse contours obtained after propagation through tissue were more irregular and the physical dimensions of specific contours greater than those obtained in water (3). The latter finding would seem to imply that resolution in tissue would be worse than in water, and this was observed. These isoresponse contour changes are currently being investigated for their dependence on parameters such as tissue types and models, as well as transducer diameter, frequency, and
JUly 1979
focusing properties. To facilitate in collection of the data, an automated three-dimensional beam profiling system has been developed. METHODS AND MATERIALS The ultrasound beam profiling system was constructed and incorporated as a peripheral device to a PDP 8/1 minicomputer system, which is routinely used at the University of Wisconsin Hospital for radiation therapy planning. Many of the devices employed in the PDP 8/1 system are also utilized in the profiling system. These include a DEC tape drive, a Teletype terminal, a Tektronix 611 monitor, a Houston Complot plotter, and an Artronix isodensitometer. A Biomation 610 B transient recorder used to digitize A-mode signals in previous projects is also a part of the profiling system. The profiler is shown schematically in Figure 1. Its components include: a steel stand (A); an aluminum frame assembly (B); two rails (C); a sliding aluminum cart (0); two gears and racks for two rack and pinion drives (E and F); and a stainless steel spherical target either of 4.8 or 12.7 mm diameter (G) (4). Together, the two rack and pinion drives provide for scanning of the target in a single plane. One stepping motor is rigidly attached to the sliding cart. As can be seen in Figure 1, the rack and pinion drive system associated with this stepping motor controls the x (up and down) position of the target attached to the rack. The other stepping motor, that of rack and pinion drive E, is rigidly attached to the aluminum frame assembly. This rack and pinion drive system controls the y (side to side) position of the sliding cart and therefore the target. The only necessary change in clrcuitry required to interface the profilers x-ydrives to the computer is the rerouting of the isodensitometer's computer-controlled x-y drive signals to a switch which connects these signals either to their normal destination, the stepping motors of the isodensltorneter. or their new destination, the stepping motors of the profiler. This scanning in a single lateral plane accounts for two degrees of freedom of the profiler; the third is that associated with the z axis in Figure 1. The aluminum frame assembly can be located at different positions in the z direction, thereby permitting the profiler to scan planes at different axial distances from the transducer face. A block diagram of the entire profiling system is shown in Figure 2. A lucite block holds the transducer in a fixed position in the irradiation chamber. The transducer is pulsed using a broadband clinical pulser-receiver unit. A-mode signals reflected from the spherical target are directed from the pulse-echo unit to the Biomation 610 B transient recorder where they are digitized to 6 bits at a rate of 10 MHz. The digitized A-mode signals are then input into the computer. The in-house developed trigger delay circuit enables the use of the 10 MHz maximum digitization rate of the transient recorder for target distances of up to 38 cm. When the transient recorder samples at a rate of 10 MHz, the peak echo signal reflected from the spherical target typically consists of 15 points. This number of data points has resulted in accurate and reproducible measurements. Beam profiles in the form of isoresponse data points are plotted on the computer-interfaced Houston Instruments plotter. An interactive computer program controls the acquisition and display of beam profiles. To obtain the pulse-echo beam profile of a transducer in water at an axial distance of 10 cm, for example. the operator first positions the spherical target 10 cm from the approximate physical center of the transducer face.
221
TECHNICAL NOTES
Vol. 132
The computer program then directs that the sensitivity control in the pulser-receiver system (a calibrated attenuator in this study) be adjusted, as well as the transient recorder sensitivity and the trigger delay until an A-mode signal obtained from the target is adequately digitized and displayed on the computer monitor. The position of the echo signal of interest is localized by using a window. The computer then determines the peak signal amplitude from within this window. At this point, the operator inputs the physical size of the rectangular array that the profiler must scan to determine the maximum reflection amplitude in the profile plane. The spherical reflector position at which this maximum occurs is defined as the center of all subsequent scans. For all scans, the maximum array size is limited to 45 X 45 data points; the minimum physical distance between each data point is 0.45 mm. The computer next instructs the operator to increase the system's sensitivity by a fixed amount. Once this task is completed, the operator inputs scanning dimensions of choice, and the computer controls the collection of cross-sectional profile data. Horizontal and vertical interpolations are performed to determine the x and y positions of the original maximum amplitude value within the new profile data array, and the computer plots these positions to scale on the interfaced plotter. These points comprise an isoresponse contour corresponding to the difference between the original sensitivity setting on the receiver and the new sensitivity setting. For example, an increase in system sensitivity of 10 dB would result in an isoresponse contour corresponding to echo signals 10 dB below the maximum signal obtained in that plane. When the operator again increases the sensitivity setting and sets the scanning dimensions, another isoresponse contour is determined and plotted. For an original maximum amplitude with a transient recorder value of either 13 or 14, accurate and precise contours have been produced ranging to approximately 40 dB below the maximum pulse-echo response in a given plane. RESULTS AND DISCUSSION Examples of some of the beam profiles that have been obtained are seen in Figures 3 and 4. Figure 3 shows the isoresponse profiles in a water/alcohol mixture in the focal region at a 10 cm depth, obtained by two different manufacturers' 3.5 MHz, 19 mm diameter, long internal focused transducers. Although the transducers had identical specifications, the -20, -30, and -40 dB contours in profile (A) were much smaller than their counterparts in (B), indicating that the transducer associated with the former had a much finer focus. A planar profile is composed of an almost infinite number of linear profiles. This is very important in asymmetric profiles such as the one in Figure 3, a, in which a linear beam profile would fail to provide complete information about many of the asymmetries in the ultrasound beam emitted by the transducer. The effects of transmission through tissue on beam profiles can be seen in comparing the profile in Figure 4, a with that in Figure 4, b. The former was obtained in a water/glycerine mixture; the latter, in the same mixture at the same axial distance with the same 2.25 MHz, 13 mm diameter, nonfocused transducer, but with an 8 cm thick excised pig liver placed between the transducer and the spherical reflector. Note that the shapes of the isoresponse contours in Figure 4, b are much more distorted than those in Figure 4, a, while the physical dimensions'
x
'A
Technical Notes
)
~~
50.7cm
Z
I·
.IV
45cm
Fig. 1. Sketch of the profiler shows several of its dimensions. The target (G) is scanned in x-y planes which are fixed z-distances from the transducer face. (All lettered components are described in the text.)
of each contour in the latter are greater than those in the former. CONCLUSION Preliminary studies have shown the profiling system to be invaluable in assessing characteristics of the beams emitted by ultrasound transducers. The profiles can be used to obtain data for comparing the isoresponse contours and focusing properties of different transducers, and for correlating the cross-sectional beam profiles with the B-scan images produced. ACKNOWLEDGMENTS: The authors would like to thank Orlando Canto for preparing the illustrations, and Su Race, Kathy McSherry, and Nancy Clark for typing the manuscript in its various stages.
TRIGGER
A-MODE SIGNAL
PULSE -
ECHO ULTRASOUND UNIT
ISODENSITOMETER (TRANSLATOR)
Fig. 2. General block diagram of the automatic beam profiling system shows paths of interaction between the various components.
222
TECHNICAL NOTES
...
~
-.
:
July 1979
..
. . . ." :: .-.:" •••f.,,_, .-~..... .:. : .. : :+ r .... '.: :.:
.. :+ ! ::.., ~ : : :~ ~ :.::~:;: .~ ~: ~ :~ t\. ••. :+
!
+.
~
B
A
I I
-....•• ••••....-..+....... '" . . . ......" 1···· . '.
...
•••• •••••• • 1
B
I I I I I I I I I I I
Z =10 em H20/ALCOHOL
Fig. 3. Computer plots show planar profiles obtained under the same conditions for two different manufacturers' 3.5 MHz, 19 mm diameter, long internal focused transducers. The contours of -6, -10, -20, -30, and -40 dB are seen in order from the center of the plots outward (graticule dimension = 1 mm).
Fig. 4. Two planar profiles obtained at the same axial distance with a 2.25 MHz, 13 mm diameter, nonfocused transducer . a. The -6, -10, -20, -30, and -40 dB contours for a spherical target scanned in a plane 10 em from the transducer face in an Irradiation chamber filled with a water/glycerine mixture (speed of sound ~ 1,570 m/sec). b. The -6, -10, -20, and -30 dB contours with 8 em of freshly excised pig liver wedged between the transducer face and a 6 mm thick sheet of lucite containing a 19.7 X 13.8 cm polyethylene scanning window (speed of sound in liver ~ 1,570 rn/sec),
REFERENCES 1. Gordon 0: Comparison of ultrasonic pulse echo apparatus used in medic ine. Ultrasonics 2:199-202, 1964 2. Mountford RA, Halliwell M: Physical sources of registration errors in pulse-echo ultrasonic systems. Part II beam deformation, deviation and divergence. Med Bioi Eng 11:33-38, 1973 3. Banjavic RA, lagzebski JA, Madsen EL, et al: Ultrason ic beam sensitivity profile changes in mammalian tissue. Ultrasound Med Bioi 4:515-518, 1978 4. Brendel K, Filipczynski LS, Gerstner R, et al: Methods of measuring the performance of ultrasonic pulse-echo diagnost ic equipment. Ultrasound Med Bioi 2:343-350, 1977
, From the Medical Physics Sect ion, Departments of Radiology (M.M.G., R.A.B., ELM., J.A.Z.1 and Human Oncology (J.A.Z.), University of Wisconsin, Madison, Wisc. 53706 . Received Aug. 18, 1978; accepted and revision requested Dec . 22; revision received Jan. 25, 1979. Supported in part by the National Cancer Institute, Wisconsin'Clinical Center Grant 5-PO1-CA-19278-02 and by the University of Wisconsin Graduate School, Research Proj. No. 181304. as
A Modification of the Craniocaudal View in Mammograph y 1 Lawrence W. Bassett, M.D., and Shirley Axelrod, C.R.T. A modified version of the craniocaudal projection, designed to better demonstrate lesions located near the chest wall, is presented. INDEX TERM:
Mammograpl'ly . tec hnique
Radiology 132:222-224, Jul y 1979
The standard craniocaudaf mammographic projection will occasionally fail to demonstrate a lesion found in the posterior aspect.of the breast in the mediolateral view. It is important to locate these structures in the former view for the purpose of biopsy. Some radiologists have accomplished this by positioning the breast so that the film is centered on the outer or inner portion, thereby obtaining an exaggerated-outer or exaggeratedinner view . If neither view shows the lesion in the outer or inner breast, it is assumed to be in the midline. The modified craniocaudal view presented here proved to be more effective in 10-
Fig. 1.
Modified craniocaudal projection.
TECHNICAL NOTES
An Automated Ultrasound Transducer Beam Profiling System 1 Mitchell M. Goodsitt, M.S., Richard A. Banjavic, Ph.D., James A. Zagzebski, Ph.D., and Ernest L. Madsen, Ph.D. An automatic, three-dimensional, ultrasound beam profiling system was developed and incorporated into a radiotherapy planning minicomputer system with few modifications and at minimal cost. This profiler was invaluable in assessing the characteristics of beams emitted by ultrasound transducers. INDEX TERMS:
Computers. Ultrasound, instrumentation
Radiology 132:220-222, July 1979
A transducer usually comes equipped with a data sheet containing its characteristics, including real time r.f. wave form, frequency spectrum, and pulse-echo beam profiles. The latter are often linear beam profiles at different depths in water, usually obtained by plotting the lateral transducer position versus the relative amplitude of the signal reflected from a small steel or nylon rod at a given depth as the transducer scans across the rod. These linear profiles yield only two-dimensional information about three-dimensional ultrasound beams. The imaging capabilities of an ultrasound transducer are highly dependent upon the three-dimensional nature of the beams emitted and received. Any asymmetries in the transducer beam, either as a result of a flaw in the construction of the transducer or of transmission through tissues will affect the images produced. The beam profiles provided by the manufacturer are deficient in that properties in only one of many directions at a given depth are depicted, and the effects of transmission through tissues are not considered. To obtain a better understanding of the spatial pulse-echo response of transducers, three-dimensional ultrasound beam profilers have been built (1-4). Such devices typically facilitate the positioning of a target at known points within the ultrasound beam emitted from a stationary transducer. Data consisting of transducer-detected peak echo amplitudes, associated with different target positions in a plane parallel to the transducer face, can be used to plot transducer "isoresponse contours". These contours are analogous to the isodose contours in radiotherapy planning, each representing points in the plane parallel to the transducer face from which equal amplitude echoes are obtained. A beam profile, or more specifically, a planar beam profile is made up of a collection of these contours. The first three-dimensional profiler employed in our studies was a modified stereotaxic device which was manually operated. With this profiler, it was discovered that ultrasound beam profiles obtained with attenuating media such as castor oil between the transducer face and target showed greater beam divergence or spreading than those obtained in nonattenuating media such as water. It was also found that isoresponse contours obtained after propagation through tissue were more irregular and the physical dimensions of specific contours greater than those obtained in water (3). The latter finding would seem to imply that resolution in tissue would be worse than in water, and this was observed. These isoresponse contour changes are currently being investigated for their dependence on parameters such as tissue types and models, as well as transducer diameter, frequency, and
JUly 1979
focusing properties. To facilitate in collection of the data, an automated three-dimensional beam profiling system has been developed. METHODS AND MATERIALS The ultrasound beam profiling system was constructed and incorporated as a peripheral device to a PDP 8/1 minicomputer system, which is routinely used at the University of Wisconsin Hospital for radiation therapy planning. Many of the devices employed in the PDP 8/1 system are also utilized in the profiling system. These include a DEC tape drive, a Teletype terminal, a Tektronix 611 monitor, a Houston Complot plotter, and an Artronix isodensitometer. A Biomation 610 B transient recorder used to digitize A-mode signals in previous projects is also a part of the profiling system. The profiler is shown schematically in Figure 1. Its components include: a steel stand (A); an aluminum frame assembly (B); two rails (C); a sliding aluminum cart (0); two gears and racks for two rack and pinion drives (E and F); and a stainless steel spherical target either of 4.8 or 12.7 mm diameter (G) (4). Together, the two rack and pinion drives provide for scanning of the target in a single plane. One stepping motor is rigidly attached to the sliding cart. As can be seen in Figure 1, the rack and pinion drive system associated with this stepping motor controls the x (up and down) position of the target attached to the rack. The other stepping motor, that of rack and pinion drive E, is rigidly attached to the aluminum frame assembly. This rack and pinion drive system controls the y (side to side) position of the sliding cart and therefore the target. The only necessary change in clrcuitry required to interface the profilers x-ydrives to the computer is the rerouting of the isodensitometer's computer-controlled x-y drive signals to a switch which connects these signals either to their normal destination, the stepping motors of the isodensltorneter. or their new destination, the stepping motors of the profiler. This scanning in a single lateral plane accounts for two degrees of freedom of the profiler; the third is that associated with the z axis in Figure 1. The aluminum frame assembly can be located at different positions in the z direction, thereby permitting the profiler to scan planes at different axial distances from the transducer face. A block diagram of the entire profiling system is shown in Figure 2. A lucite block holds the transducer in a fixed position in the irradiation chamber. The transducer is pulsed using a broadband clinical pulser-receiver unit. A-mode signals reflected from the spherical target are directed from the pulse-echo unit to the Biomation 610 B transient recorder where they are digitized to 6 bits at a rate of 10 MHz. The digitized A-mode signals are then input into the computer. The in-house developed trigger delay circuit enables the use of the 10 MHz maximum digitization rate of the transient recorder for target distances of up to 38 cm. When the transient recorder samples at a rate of 10 MHz, the peak echo signal reflected from the spherical target typically consists of 15 points. This number of data points has resulted in accurate and reproducible measurements. Beam profiles in the form of isoresponse data points are plotted on the computer-interfaced Houston Instruments plotter. An interactive computer program controls the acquisition and display of beam profiles. To obtain the pulse-echo beam profile of a transducer in water at an axial distance of 10 cm, for example. the operator first positions the spherical target 10 cm from the approximate physical center of the transducer face.
221
TECHNICAL NOTES
Vol. 132
The computer program then directs that the sensitivity control in the pulser-receiver system (a calibrated attenuator in this study) be adjusted, as well as the transient recorder sensitivity and the trigger delay until an A-mode signal obtained from the target is adequately digitized and displayed on the computer monitor. The position of the echo signal of interest is localized by using a window. The computer then determines the peak signal amplitude from within this window. At this point, the operator inputs the physical size of the rectangular array that the profiler must scan to determine the maximum reflection amplitude in the profile plane. The spherical reflector position at which this maximum occurs is defined as the center of all subsequent scans. For all scans, the maximum array size is limited to 45 X 45 data points; the minimum physical distance between each data point is 0.45 mm. The computer next instructs the operator to increase the system's sensitivity by a fixed amount. Once this task is completed, the operator inputs scanning dimensions of choice, and the computer controls the collection of cross-sectional profile data. Horizontal and vertical interpolations are performed to determine the x and y positions of the original maximum amplitude value within the new profile data array, and the computer plots these positions to scale on the interfaced plotter. These points comprise an isoresponse contour corresponding to the difference between the original sensitivity setting on the receiver and the new sensitivity setting. For example, an increase in system sensitivity of 10 dB would result in an isoresponse contour corresponding to echo signals 10 dB below the maximum signal obtained in that plane. When the operator again increases the sensitivity setting and sets the scanning dimensions, another isoresponse contour is determined and plotted. For an original maximum amplitude with a transient recorder value of either 13 or 14, accurate and precise contours have been produced ranging to approximately 40 dB below the maximum pulse-echo response in a given plane. RESULTS AND DISCUSSION Examples of some of the beam profiles that have been obtained are seen in Figures 3 and 4. Figure 3 shows the isoresponse profiles in a water/alcohol mixture in the focal region at a 10 cm depth, obtained by two different manufacturers' 3.5 MHz, 19 mm diameter, long internal focused transducers. Although the transducers had identical specifications, the -20, -30, and -40 dB contours in profile (A) were much smaller than their counterparts in (B), indicating that the transducer associated with the former had a much finer focus. A planar profile is composed of an almost infinite number of linear profiles. This is very important in asymmetric profiles such as the one in Figure 3, a, in which a linear beam profile would fail to provide complete information about many of the asymmetries in the ultrasound beam emitted by the transducer. The effects of transmission through tissue on beam profiles can be seen in comparing the profile in Figure 4, a with that in Figure 4, b. The former was obtained in a water/glycerine mixture; the latter, in the same mixture at the same axial distance with the same 2.25 MHz, 13 mm diameter, nonfocused transducer, but with an 8 cm thick excised pig liver placed between the transducer and the spherical reflector. Note that the shapes of the isoresponse contours in Figure 4, b are much more distorted than those in Figure 4, a, while the physical dimensions'
x
'A
Technical Notes
)
~~
50.7cm
Z
I·
.IV
45cm
Fig. 1. Sketch of the profiler shows several of its dimensions. The target (G) is scanned in x-y planes which are fixed z-distances from the transducer face. (All lettered components are described in the text.)
of each contour in the latter are greater than those in the former. CONCLUSION Preliminary studies have shown the profiling system to be invaluable in assessing characteristics of the beams emitted by ultrasound transducers. The profiles can be used to obtain data for comparing the isoresponse contours and focusing properties of different transducers, and for correlating the cross-sectional beam profiles with the B-scan images produced. ACKNOWLEDGMENTS: The authors would like to thank Orlando Canto for preparing the illustrations, and Su Race, Kathy McSherry, and Nancy Clark for typing the manuscript in its various stages.
TRIGGER
A-MODE SIGNAL
PULSE -
ECHO ULTRASOUND UNIT
ISODENSITOMETER (TRANSLATOR)
Fig. 2. General block diagram of the automatic beam profiling system shows paths of interaction between the various components.
222
TECHNICAL NOTES
...
~
-.
:
July 1979
..
. . . ." :: .-.:" •••f.,,_, .-~..... .:. : .. : :+ r .... '.: :.:
.. :+ ! ::.., ~ : : :~ ~ :.::~:;: .~ ~: ~ :~ t\. ••. :+
!
+.
~
B
A
I I
-....•• ••••....-..+....... '" . . . ......" 1···· . '.
...
•••• •••••• • 1
B
I I I I I I I I I I I
Z =10 em H20/ALCOHOL
Fig. 3. Computer plots show planar profiles obtained under the same conditions for two different manufacturers' 3.5 MHz, 19 mm diameter, long internal focused transducers. The contours of -6, -10, -20, -30, and -40 dB are seen in order from the center of the plots outward (graticule dimension = 1 mm).
Fig. 4. Two planar profiles obtained at the same axial distance with a 2.25 MHz, 13 mm diameter, nonfocused transducer . a. The -6, -10, -20, -30, and -40 dB contours for a spherical target scanned in a plane 10 em from the transducer face in an Irradiation chamber filled with a water/glycerine mixture (speed of sound ~ 1,570 m/sec). b. The -6, -10, -20, and -30 dB contours with 8 em of freshly excised pig liver wedged between the transducer face and a 6 mm thick sheet of lucite containing a 19.7 X 13.8 cm polyethylene scanning window (speed of sound in liver ~ 1,570 rn/sec),
REFERENCES 1. Gordon 0: Comparison of ultrasonic pulse echo apparatus used in medic ine. Ultrasonics 2:199-202, 1964 2. Mountford RA, Halliwell M: Physical sources of registration errors in pulse-echo ultrasonic systems. Part II beam deformation, deviation and divergence. Med Bioi Eng 11:33-38, 1973 3. Banjavic RA, lagzebski JA, Madsen EL, et al: Ultrason ic beam sensitivity profile changes in mammalian tissue. Ultrasound Med Bioi 4:515-518, 1978 4. Brendel K, Filipczynski LS, Gerstner R, et al: Methods of measuring the performance of ultrasonic pulse-echo diagnost ic equipment. Ultrasound Med Bioi 2:343-350, 1977
, From the Medical Physics Sect ion, Departments of Radiology (M.M.G., R.A.B., ELM., J.A.Z.1 and Human Oncology (J.A.Z.), University of Wisconsin, Madison, Wisc. 53706 . Received Aug. 18, 1978; accepted and revision requested Dec . 22; revision received Jan. 25, 1979. Supported in part by the National Cancer Institute, Wisconsin'Clinical Center Grant 5-PO1-CA-19278-02 and by the University of Wisconsin Graduate School, Research Proj. No. 181304. as
A Modification of the Craniocaudal View in Mammograph y 1 Lawrence W. Bassett, M.D., and Shirley Axelrod, C.R.T. A modified version of the craniocaudal projection, designed to better demonstrate lesions located near the chest wall, is presented. INDEX TERM:
Mammograpl'ly . tec hnique
Radiology 132:222-224, Jul y 1979
The standard craniocaudaf mammographic projection will occasionally fail to demonstrate a lesion found in the posterior aspect.of the breast in the mediolateral view. It is important to locate these structures in the former view for the purpose of biopsy. Some radiologists have accomplished this by positioning the breast so that the film is centered on the outer or inner portion, thereby obtaining an exaggerated-outer or exaggeratedinner view . If neither view shows the lesion in the outer or inner breast, it is assumed to be in the midline. The modified craniocaudal view presented here proved to be more effective in 10-
Fig. 1.
Modified craniocaudal projection.
