Scanning angle of transverse section in transrectal ultrasonography affects the accuracy of volume calculation of the prostate estimated as ellipsoid or cuboid.

Akira Kimura¹, Yuuji Kurooka², Kiyoshi Hirasawa², Tadaichi Kitamura¹, and Kazuki Kawabe<sup>2

1</sup>Department of Urology, Branch Hospital, ²Department of Urology, Faculty of Medicine, the University of Tokyo, Tokyo, Japan

Runnig title: Effect of the scanning angle on the sonometrics of the prostate

Abstract

The estimation of the prostatic volume by transrectal ultrasonography is now widely used to evaluate the effect of a conservative treatment and to improve the specificity of PSA. For the calculation of PSA density ( PSA/prostatic volume ratio ), accurate measurement of prostatic volume is necessary. The thrree dimensional models of the prostatic contour were constructed from transrectal ultrasonotomograms taken at 5-mm intervals in 20 cases. Using these models, the accuracy of ellipsoid volume calculation and prolate ellipse volume calculation was examined. Prolate ellipse volume calculation was more accurate than ellipsoid volume calculation, the error of which was about 20%. Moreover, ellipsoid volume calculation fluctuated markedly in accordance with the change of the scanning angle of transverse section. The error of prolate ellipse volume calculation became large when prostatic length was not measured perpendicular to the plane in which width and height were measured. This situation frequently occurs when the length from the internal urethral orifice to the prostatic apex is measured.

Key words: sonometrics, transrectal ultrasonography, ellipsoid volume calculation, prolate ellipse volume calculation, PSA density

Introduction

The estimation of the prostatic volume by transrectal ultrasonography is now widely used for the evaluation of a conservative treatment of prostatic cancer or hypertrophy. Though the most accurate method for sonometrics is multi-slice planimetric volume calculation, easier methods such as ellipsoid volume calculation or prolate ellipse volume calculation are more frequently used. This is partly because a chair-type scanner is required to take tomograms at 5-mm intervals. In many hospitals, however, transrectal ultrasonography is perfomed using a bi-plane hand scanner.
Since the chair-type scanner was exchanged with a hand scanner in our department, the estimation method was also changed to ellipsoid volume calculation. In the beginning, the error of this method was evaluated compared with multi-slice planimetry. The results showed that ellipsoid volume calculation underestimated the volume by 20%¹⁾.
Though the inaccuracy degree of 20% is acceptable for the evaluation of the conservative treatment of prostatic cancer or hypertrophy, it is not so for calculating PSA densityt²⁾.
As the ellisoid method uses only parameters of the widest transverse section, its value seems to depend on the angle between the scanning plane and the prostatic longitudinal axis. The prostatic contour in the widest transverse section varies depending on the insertion angle of the probe to the rectum.
Compared with the ellipsoid volume calculation, the prolate ellipse volume calculation seems to be more accurate, being calculated from both transverse and longitudinal sections. In this method, however, three diameters have to be measured so that they meet at right angles. Nevertheless it is not rare that length is substituted by urethral length. The urethra is not neccesarily perpendicular to the transverse section.
In this study, the accuracy of the ellipsoid volume calculation and the prolate ellipse volume calculation are compared with multi-slice planimetric volume calculation. Especially, the influence of the inclination of transverse plane on the accuracy was evaluated.

Materials and Methods

The three dimensional (3D ) models of the prostatic contour were constructed in twenty patients who underwent transrectal ultrasonotomography by Aloka SSD-60 or Tosiba SSL-51C with transrectal chair-type probe. Serial tomograms were taken at 5-mm intervals. Five cases had prostatic cancer (three having stage B disease, one stage C, one stage D ). Ten cases had prostatic hypertrophy. Five of them underwent operation and histological confirmation of the diagnosis was made. Another five underwent anti-androgen therapy, after which re-examination by ultrasonography was conducted. Three cases of hematospermia and two cases of superficial bladder tumor were categorized as normal groups.
Serial tomograms taken at 5-mm intervals were input into a personal computer via videocamera. From the prostatic contour of the transrectal ultrasonotomograms( Fig.1A ), 3D software reconstructs the 3D prostatic image³⁾( Fig.1B ). The transverse section of prostatic contour cut by a plane with arbitrarily chosen angle to prostatic longitudinal axis( Fig.1C ) could be drawn by the 3D software. In the widest transverse section, Height( H ), Width( W ), and Area( A ) were measured( Fig.1D ). Prostatic length were measured in two ways; Length( Lv ) vertical to transverse section( Fig.1D ), and Length( Lp ) parallel to prostatic urethra( Fig.1B ).
Because real prostatic volume was not obtained in the 20 cases( except those of adenomas in five hypertrophy patients ), true prostatic volume is defined as equal to that calculated by multi-slice planimetric volume calculation( V₀ ).
The formula for ellipsoid volume calculation is
V₁=8A²/3pW
which assumes the gland is ellipsoidal in shape, where Width is defined as a hypothetical axis of rotation of Area.
The formula for prolate ellipse volume calculation is
V=6HWL/p
in which the shape of the prostate is considered cuboidal.
The volume calculated using Lp as L is named as V₂, and the volume using Lv is named as V₃.
The rate of the error for each method was calculated by dividing the difference of the value from V₀ by V₀.
The angle between the transverse plane and the prostatic longitudinal axis was changed from 45^o to 135^o at 15^o intervals, and the rate of error of each volume calculation method was calculated for each angle. This calculation yielded seven data of the error rates for each volume estimation method and for each case. From the seven values, the influence of a scanning angle on the degree of errors was graphically plotted. From the seven data, the mean and range of errors depending on the angle were also calculated for each case. Finally, averages of the means and ranges for normal, hypertrophy, and cancer groups were calculated by averaging the data of five cases. Each volume calculation method was evaluated from a standpoint whether the errors were small and whether the degree of errors was independent of the angle between the transverse section and the prostatic axis.
In the five cases of hypertrophy who received anti-androgen therapy, the degree of errors of ellipsoid volume calculation before treatment was individually compared with that of after treatment.

Results

The rate of the errors of ellipsoid and prolate ellipse volume techniques in five normal cases was shown as a function of the inclination of transverse plane in fig.2.
The ellipsoid volume calculation had the tendency to underestimate the volume by 20%. Also, the degree of miscalculation fluctuated markedly depending on the inclination of the transverse plane. If the angle between the transverse plane and the prostatic axis became sharp so that the cross section approached the coronal section, the degree of underestimation became small.
Mean errors of ellipsoid volume calculation averaged for seven angles in each case were -30,-29,-25,-17, and -11%( average of five cases being -22.4% ). Ranges of errors in accordance with the change of the angle for each case were 53,39,39,35, and 32%( average of five cases being 39.6% ).
Although the prolate ellipse volume calculation was more accurate than the ellipsoid volume calculation, when Lp was used for calculation, the error became great as the angle became out of perpendicular.
Mean errors of prolate ellipse volume calculation using Lp for L averaged for seven angles in each case were -6,-5,-2,-2, and 3%( average of five being -2.4% ). Ranges of errors with the change of the angle for each case were 47,39,30,24, and 19%( average of five being 31.8% ).
The prolate ellipse volume calculation using Lv for L was the most reliable method. The degree of error was the least dependent on the angle between the plane and the axis.
Mean errors of prolate ellipse volume calculation using Lv for L averaged for seven angles in each case were -12,-9,-6,-5, and -1%( average of five being -6.6%) . Ranges of errors with the change of the angle for each case were 33,23,20,19, and 18%( average of five being 22.6% ).
By summing up the data of the five cases in each group, the average of the five cases of mean errors and ranges of errors, with the angle change for three volume calculation techniques in normal group, hypertrophy group, and cancer group, were calculated ( Table.1 ).
The tendencies, one of which being that the ellipsoid volume calculation was the least accurate and the other being that the prolate ellipse volume calculation using Lv is less likely to be influenced by the angle of the transverse section than that using Lp, were also demonstrated in hypertrophy and cancer group.
In order to evaluate the usefulness of the ellipsoid volume calculation in the followup of the conservative treatment, mean errors of ellipsoid volume calculation and the range of its change with the angle of the transverse section were examined before and after the anti-androgen therapy in five cases with hypertrophy. The results are listed in table2.
Though the degree of errors and its range were as large as those of the cases who underwent operation, the degree of errors before and after the treatment resembled each other in individual patients.

Discussion

Until recently, the sonometrics of the prostate have been used to evaluate the treament effect. If the usage of volume calculation is confined to the evaluation of the conservative treatment, the error rate of 20% may be acceptable. However, the volume calculation is now used for the differential diagnosis of cancer from hypertrophy. By dividing PSA by prostatic volume, false positive rate of PSA in hypertrophy can be diminished. In our department, the division of PSA by volume has been used since 1991 for the determination whether or not the cases in which PSA were positive but the first biopies were negative should receive re-biopsy ⁴⁾. This had been so before Benson mentioned the usefulness of PSA density in 1992²⁾.
However, several recent literatures proposed some doubts about the usefulness of PSA density^5,6). Though these authors agreed that PSA density reduced the overlap of cancer and hypertrophy, they stated that cutoff value of PSA density was difficult to determine with which superior sensitivity and specificity to single PSA could be obtained.
To compare the usefulness of PSA density with PSA, one has to calculate the volume with the same accuracy as PSA. If ellipsoidal volume calculation tends to underestimate the volume by 20%, PSA density calculated from the volume will be overestimated by 20%. If cutoff value of PSA density changes from 0.10 to 0.12, the sensitivity of PSA density in detecting cancer falls from 70% to 60%, while the specificity raises from 70% to 80%⁶⁾.
In this study, it was demonstrated that the prolate ellipse volume calculation is more accurate than ellipsoid volume calculation. Moreover, ellipsoid volume calculation fluctuated markedly in accordance with the change of the scanning angle of transverse section. For example, in a normal case shown in fig.2, the error rate of ellipsoid volume calculation varied from+25% to -24% as the inclination of the transverse plane changed from 45^o to 90^o. It was also demonstrated that to obtain accurate value by the prolate ellipse volume calculation, one has to take care to have three axes cross at right angles.
It should be mentioned that the results presented here were all emerged from the theoretical model constructed inside a computer. In the model, the computer correctly chooses the prostatic length perpendicular to the transverse plane. But, in real clinical examinaton it is not easy to measure Width, Height, and Length correctly so that they cross at right angles.
The bi-plane hand scanner currently used cannot visualize maximum transverse and longitudinal section at the same time. Though by a single keystroke, an operator can change the scanning plane from transverse to longitudinal after taking maximum transverse section, he/she has to insert the probe further into the rectum to obtain longitudinal section of the whole prostate, if he/she uses a sector-and-linear scanner-combined probe. With a mechanical sector scanner, the operator must pull back the probe to take the longitudinal section after taking the transverse section. Only by changing the scanning plane, the operator can see the bladder and the upper part of the prostate simultaneously.
By moving the probe to gain two sections, the relation between the transverse section and the longitudinal section, i.e. the relation of Height, Width, and Length, becomes obscure. We recommend to take an additional photograph just after changing the plane and before moving the probe. This facilitates understanding the relation between the two planes.
Though ellipsoid volume calculation has the errors of about 20% and fluctuates depending on the inclination of the transverse plane, it seems useful in the followup of the conservative treatment. The degree of errors were on the same level before and after the anti-androgen therapy in individual cases. Perhaps, as the whole prostate shrinks in the same manner, the shape of the prostate is not altered during the treatment. This is why the degree of the error when the shape is simulated by ellipsoid is not changed.
Though prolate ellipse volume is easily calculated by dividing the product of
H, W, and L by 2 ( 6/p is aproximately 2 ), it should be reminded that the measurement of three diameters itself may yield errors of about ten percent, which were neglected in this study.
In contrast, ellipsoid is more resistant to errors yielded during measurement. As was shown in the formula, V₁ uses Area twice in its calculation. As Area is more resistant to measurement error than W, H, and L, ellipsoid volume calculation should not be discarded. Instead, efforts to seek a new technique to improve accuracy of ellipsoid volume calculation or to develop a new formula using area of transverse or longitudinal section are awaited.

References

1. Kimura A, Hirasawa K, Aso Y. Accuracy of ellipsoid method for estimation of prostatic weight. Jpn J Med Ultrasonics 1991;18:620-625
2. Benson MC, Whang IS, Olsson CA, McMahon DJ, Cooner WH. The use of prostate specific antigen density to enhance the predictive value of intermediate levels of serum prostate specific antigen. J Urol 1992;147:817-821
3. Kimura A, Aso Y. Three-dimensional display of the prostate based on transrectal ultrasonogram. Jpn J Urol 1992;83:1490-1498
4. Okamura T, Hirasawa K, Kimura A, Aso Y. Usefulness of the combination of serum prostatic markers and the prostatic weight by sonometrics for the diagnosis of prostatic carcinoma. Jpn J Med Ultrasonics 1991;18 Supple 1:297-298
5. Brawer MK, Aramburu EAG, Chen GL, Preston SD, Ellis WJ. The inability of prostate specific antigen index to enhance the predictive value of prostate specific antigen in the diagnosis of prostatic carcinoma. J Urol 1993;150:369-373
6. Catalona WJ, Richie JP, Kernion JB, Ahmann FR, Patliff TL, Dalkin BL, Kavoussi LR, MacFarlane MT, Southwick PC. Comparison of prostate specific antigen concentration versus prostate specific antigen density in the early detection of prostate cancer: receiver operating characteristics curves. J Urol 1994;152:2031-2036