ARTICLE IN PRESS

Original Investigation

Determination of Prostate Volume:

A Comparison of
Contemporary Methods
Adam Bezinque, DO, Andrew Moriarity, MD, Crystal Farrell, MD, Henry Peabody, BS,
Sabrina L. Noyes, BS, Brian R. Lane, MD, PhD

Rationale and Objectives: Prostate volume (PV) determination provides important clinical information. We compared PVs determined
by digital rectal examination (DRE), transrectal ultrasound (TRUS), magnetic resonance imaging (MRI) with or without three-
dimensional (3D) segmentation software, and surgical prostatectomy weight (SPW) and volume (SPV).
Materials and Methods: This retrospective review from 2010 to 2016 included patients who underwent radical prostatectomy ≤1 year
after multiparametric prostate MRI. PVs from DRE and TRUS were obtained from urology clinic notes. MRI-based PVs were calculated
using bullet and ellipsoid formulas, automated 3D segmentation software (MRI-A3D), manual segmentation by a radiologist (MRI-R3D),
and a third-year medical student (MRI-S3D). SPW and SPV were derived from pathology reports. Intraclass correlation coefficients com-
pared the relative accuracy of each volume measurement.
Results: Ninety-nine patients were analyzed. Median PVs were DRE 35 mL, TRUS 35 mL, MRI-bullet 49 mL, MRI-ellipsoid 39 mL, MRI-
A3D 37 mL, MRI-R3D 36 mL, MRI-S3D 36 mL, SPW 54 mL, SPV-bullet 47 mL, and SPV-ellipsoid 37 mL. SPW and bullet formulas had
consistently large PV, and formula-based PV had a wider spread than PV based on segmentation. Compared to MRI-R3D, the intraclass
correlation coefficient was 0.91 for MRI-S3D, 0.90 for MRI-ellipsoid, 0.73 for SPV-ellipsoid, 0.72 for MRI-bullet, 0.71 for TRUS, 0.70 for
SPW, 0.66 for SPV-bullet, 0.38 for MRI-A3D, and 0.33 for DRE.
Conclusions: With MRI-R3D measurement as the reference, the most reliable methods for PV estimation were MRI-S3D and MRI-
ellipsoid formula. Automated segmentations must be individually assessed for accuracy, as they are not always truly representative of
the prostate anatomy. Manual segmentation of the prostate does not require expert training.
Key Words: Prostate cancer; magnetic resonance imaging; prostate volume; TRUS.
© 2018 The Association of University Radiologists. Published by Elsevier Inc. All rights reserved.

INTRODUCTION adrenergic receptor blockers alone (1,2). PV may be a predictor

of BPH complications, such as urinary retention, hydrone-

P
rostate volume (PV) determination is an essential com-
phrosis, and renal injury, which may also influence therapy
ponent of evaluation and management of prostatic
(2). Treatment effectiveness can be evaluated with serial PV
disease, including benign prostatic hypertrophy (BPH)
estimations and can be stratified based on zonal anatomy as
and prostate cancer (PCa). With respect to BPH, PV esti-
well (2). Furthermore, PV frequently affects the choice of sur-
mation can provide objective data for the purposes of treatment
gical approach for BPH when considering between ablative
planning, monitoring response to therapy, and surgical tech-
procedures, transurethral resection, minimally invasive surgery,
nique. For example, treatment options for lower urinary tract
and open prostatectomy (2).
symptoms secondary to BPH depend on gland size, with larger
For patients with PCa, PV is helpful for risk stratification
glands (>40 mL) recommended for 5-alpha reductase inhibi-
during screening and risk assessment, especially when used in
tors and smaller glands (<30 mL) often treated with alpha-
conjunction with prostate-specific antigen (PSA). Although
PSA values correlate with risk of PCa, elevations in PSA can
Acad Radiol 2018; ■:■■–■■
also be due to BPH and inflammatory benign prostatic dis-
From the Michigan State University College of Osteopathic Medicine, East
Lansing, Michigan (A.B.); Spectrum Health, 145 Michigan Street NE, Grand
eases, such as prostatitis, limiting the specificity (3). Biopsies
Rapids, MI (A.M., C.F., H.P., S.L.N., B.R.L.); Advanced Radiology Services PC performed based on elevated PSA alone can result in biopsy
(A.M.); Grand Rapids Medical Education Partners, Grand Rapids, Michigan rates negative for malignancy in up to 76% of cases (4). In
(C.F.); Michigan State University College of Human Medicine, 145 Michigan
Street NE, Grand Rapids, MI (B.R.L.). Received February 5, 2018; revised March addition, overdiagnosis rates as high as 60% have been re-
5, 2018; accepted March 7, 2018. Address correspondence to: B.R.L. e-mail: ported, as patients may suffer from unnecessary morbidity from
brian.lane@spectrumhealth.org
treatment of clinically indolent, low-grade PCa (5). Multiple
© 2018 The Association of University Radiologists. Published by Elsevier Inc.
All rights reserved.
studies have shown that PSA derivatives that incorporate volume,
https://doi.org/10.1016/j.acra.2018.03.014 such as PSA density (6), outperform PSA in predicting overall

1
BEZINQUE ET AL
and clinically significant PCa (7,8), as larger prostates not only
have greater rates of smaller volume PCa but also less-
aggressive tumors across a variety of pathologic variables (9).
Another PSA derivative is the PV index, equal to the ratio of
transitional zone to peripheral zone volume, which is in-
versely correlated with PCa risk (10). Use of an approach to
risk stratification that includes PV helps to select the most ap-
propriate patients for active surveillance and may reduce
overtreatment of low-risk tumors (11).
In addition to risk stratification, PV can help guide biopsy
and treatment planning for PCa. Specifically, PV is useful in
determining how many needle cores are required for a suf-
ficient nontargeted prostate biopsy (12). Also, as in BPH, certain
treatment strategies for PCa depend on gland size. For in-
stance, because of decreasing effectiveness of brachytherapy
for large (>50 mL and >5.5 cm) glands, androgen depriva-
tion therapy is often used before treatment to downsize the
prostate and reduce pubic arch bone interference (13). There-
fore, PV is a valuable factor in predictive models for PCa and
can have a significant impact on patient management.
Methods for estimating PV are continually evolving, re-
flecting the growing interest and importance of improving
the accuracy of PV determination. The earliest technique,
PV estimation by digital rectal examination (DRE) (which
only assesses one surface of the prostate), is inherently sub-
jective, known to be unreliable, and often underestimates
gland volume, especially for prostates larger than 30 mL (14).
PV determined by transrectal ultrasound (TRUS) using the
ellipsoid formula is a significant improvement and currently
is the most commonly used approach because of its wide
availability and time and cost efficiency. However, it remains
user-dependent, sensitive to sonographic artifacts and defor-
mation of the gland by the probe (14), and assumes a regular
ellipsoid shape which most prostates do not have (2,15).
Planimetry, which involves the summation of areas in mul-
tiple slices, improves accuracy by accounting for various gland
shapes and complex boundaries, but is prohibitively time-
intensive and subject to inter- and intraobserver variability
(2,14,16–18).
Advancements in magnetic resonance imaging (MRI) have
allowed for more detailed PV analysis by providing higher
soft tissue contrast resolution and more complex computa-
tional abilities. The most widely performed methods to
determine PV with MRI are ellipsoid or bullet formulas based
on three-dimensional (3D) diameter measurements, and manual,
semiautomated, and fully automated segmentation using three-
dimensional software. PVs calculated by formula are again
subject to false assumptions of uniform gland configuration.
3D segmentation software is gaining popularity and allows for
faster analysis, but anatomic contours are not always accu-
rate, requiring manual intervention, and many algorithms are
still being developed and validated.
Given the various clinical implications of PV on the eval-
uation and management of prostate disease, accurate PV
determination is critical, but the optimal method is cur-
rently unknown. We set out to compare multiple PV methods,
2
Academic Radiology, Vol ■, No ■■, ■■ 2018
including DRE, TRUS, MRI with and without a 3D prostate
segmentation program, and prostatectomy specimens to de-
termine the optimal volume measurement technique.
MATERIALS AND METHODS
In this institutional review board–approved retrospective review,
patients who had a multiparametric prostate MRI (mpMRI)
between October 2010 and March 2016 within 1 year of radical
prostatectomy were identified using an institutional registry.
All mpMRI studies were performed on 3T GE Discovery
MR750 (GE Healthcare, Waukesha, WI, USA) scanners with
an eight-channel external phased array coil without endorectal
coil. The T2-weighted fast relaxation fast spin echo axial se-
quence used for all segmentations parameters suggested by the
software vendor included a 200-mm field of view, 320 × 256
matrix, 4500–6000 milliseconds time of repetition (TR), 120
milliseconds time of echo (TE), 3-mm slice thickness (gap
0 mm), number of excitations (NEX) of 4, echo train length
(ETL) of 15. These parameters were unchanged during the
study period.
Clinical, radiographic, and pathologic data were collected
for each patient, including age, serum PSA at the time of
mpMRI, clinical T stage, Gleason score of the index lesion
in whole mount specimens, and PVs for each of the follow-
ing methods: DRE, TRUS, MRI with 3D segmentation
software, and MRI SPW and SPV. Information was stored
in a Health Insurance Portability and Accountability Act
(HIPAA) compliant, password protected database.
PV data were compiled from multiple different sources. PVs
from DRE were recorded from the most recent clinic notes
before surgery. PVs from TRUS were automatically gener-
ated by the ultrasound machine using the ellipsoid formula
based on the transverse (L), anteroposterior (W), and
craniocaudal (H) dimensions as determined by the perform-
ing urologist. MRI-based PVs were calculated from radiologist
performed, maximum diameter, triplanar measurements of the
prostate gland on high-resolution, focused field of view T2-
weighted images, using both the bullet (L × W × H × 5π/24)
and ellipsoid (L × W × H × 0.52) formulas. MRI 3D pros-
tate segmentation software was used to generate automatic PVs
(MRI-A3D) (DynaCAD 3.3, Build 03015B, Invivo Corpo-
ration, Gainesville, FL). Manual adjustments to the contours
of the automatically generated MRI 3D segmentations were
performed independently by an experienced radiologist (MRI-
R3D), with fellowship training in prostate MRI and a trained
third-year medical student (MRI-S3D), each blinded to the
other’s results and to the PVs determined by TRUS, DRE,
and surgical specimens. SPW and SPV were obtained from
the pathology reports, the latter of which being calculated from
the recorded specimen dimensions (L, W, H) using both the
bullet and the ellipsoid formulas. Because of the variable in-
clusion of extraprostatic tissue in surgical specimens, MRI-
R3D was selected as our “gold standard” for comparison, similar
to multiple other studies (15,16) which have also demon-
strated better inter-reader correlation using MRI (19).
Academic Radiology, Vol ■, No ■■, ■■ 2018 METHODS TO DETERMINE PROSTATE VOLUME

A nonparametric Kruskal-Wallis test was performed to TABLE 1. Clinical, Radiographic, and Pathologic
compare the volume measurements and the Shrout-Fleiss inter- Characteristics
rater reliability coefficient was used to determine intraclass
N (%) or
correlation coefficients (ICCs) for the different types of volume
Variable Median (IQR)
measurements. The Spearman correlation coefficient was used
to test for correlation between volume measurements, PSA, Median age, y (IQR) 63 (59–68)
and age separately. A Wilcoxon rank-sum test was used to Median pretreatment PSA, ng/mL (IQR) 6.6 (4.3–11.4)
determine differences in volume measurements, patient age, Biopsy Gleason group*
1 (3 + 3) 29 (32.2%)
and PSA between the Gleason score groups (3 + 4 or lower
2 (3 + 4) 25 (27.8%)
vs 4 + 3 or higher) and pT stage groups (pT2 vs pT3/T4).
3 (4 + 3) 13 (14.4%)
All statistical analyses were generated using SAS Enterprise 4 (8) 14 (15.6%)
Guide software, Version 7.1 of the SAS System (SAS Insti- 5 (9–10) 9 (10.0%)
tute Inc., Cary, NC). Pathologic stage
Timed observations were also performed for 10 typical pT2 and below 63 (63.6%)
volume calculations to determine the amount of time spent pT3 and above 36 (36.4%)
performing 3D PV segmentation in the software by the staff Surgical Gleason group†
radiologist and our dedicated 3D technologists. The upper and 1 (3 + 3) 6 (6.1%)
lower quartile ranges for these measurements were recorded 2 (3 + 4) 43 (43.9%)
to estimate the amount of time spent for each individual. 3 (4 + 3) 28 (28.6%)
4 (8) 11 (11.2%)
5 (9–10) 10 (10.2%)
RESULTS Extraprostatic extension† 30 (30.6%)
Seminal vesicle invasion† 12 (12.2%)
During the study period, a total of 435 patients received a Lymph node involvement† 22 (22.4%)
mpMRI at our institution, the majority of which were per-
formed to evaluate patients with prior negative evaluations IQR, interquartile range; PSA, prostate-specific antigen.
* Nine patients were excluded due to no reported malignancy on
for prostate cancer or during active surveillance of patients
the pathology specimen.
with diagnosed prostate cancer. Of those, 99 patients had a †
One patient was excluded due to no reported malignancy in the
radical prostatectomy within 1 year of mpMRI and were in- pathology specimen.
cluded for analysis. Median age was 63 years, initial PSA was
6.6 ng/mL, pathologic Gleason scores were 3 + 4 or higher
in 92 patients (94%) and 22 patients (22%) had positive lymph 0.91 for MRI-S3D, 0.90 for MRI-ellipsoid, 0.73 for SPV-
nodes at surgery (Table 1). Clinical predictors of volume (by ellipsoid, 0.72 for MRI-bullet, 0.71 for TRUS, 0.70 for SPW,
MRI-R3D) included patient age (r = 0.29, P = .004) and PSA 0.66 for SPV-bullet, 0.38 for MRI-A3D, and 0.33 for DRE.
(r = 0.24, P = .02). Higher prostate-specific antigen density (PSAD) was sig-
Mean and median PV measurements for each technique nificantly associated with higher risk prostate cancer. Patients
are summarized in Table 2 and graphed with outliers in with non–organ-confined disease (stage pT3 or higher) had
Figure 1. PV estimates were consistent between MRI-R3D higher PSAD than those with organ-confined cancer (stage
and MRI-S3D, with slightly higher variability seen with MRI- pT2), using volumes determined by MRI-R3D (0.30 vs 0.15,
A3D. The average time needed for the R3D PV segmentation P = .0001), TRUS (0.27 vs 0.16, P = .002), and DRE (0.26
was 1:30 minutes (interquartile range 0:45–2:15) compared vs 0.15, P = .001). PSAD was also higher in those with pre-
to 9:45 minutes (interquartile range 7:20–14:55) for segmen- dominant Gleason pattern 4/5 disease (Gleason grade groups
tation by nonradiologists. Between the two estimation formulas 3–5) compared to predominant pattern 3 PCa (Gleason grade
for SPV, SPV-ellipsoid more closely approached the 3D MRI groups 1–2). PSAD was 0.20 vs 0.17 using volume by MRI-
estimates, whereas SPV-bullet resulted in larger estimates on R3D (P = .014), 0.21 vs 0.16 by TRUS (P = .088), and 0.20
average, which were similar to SPW measurements. SPW was vs 0.16 by DRE (P = .033), respectively.
consistently higher than the volumetric methods, attribut-
able to the inclusion of extraprostatic tissue. DRE measurements
DISCUSSION
were smaller, on average, whereas TRUS measurements were
similar to the manual 3D MRI measurements. Traditionally, the most accurate method for PV estimation
Agreement between the different volume measurement tech- was considered to be planimetry by TRUS (2,14,16–18). In
niques was measured using the Shrout-Fleiss inter-rater reliability the past, the time and cost burden involved with manual
correlation coefficient for a random subset of raters (Table 3). segmentation has made this approach prohibitive in routine
Compared to MRI-R3D volume measurements, the ICC was clinical practice. With the development of mpMRI and
0.53 when all measurements were considered as one large autosegmentation software, PV can now be calculated in a
group, indicating significant heterogeneity of measure- matter of seconds. MRI-A3D is becoming routinely per-
ments. For individual comparison to MRI-R3D, the ICC was formed using commercially available software systems, but

3
BEZINQUE ET AL Academic Radiology, Vol ■, No ■■, ■■ 2018

TABLE 2. Prostate Volume Calculations (mL)

Variable N Mean Standard Deviation Median Lower Quartile Upper Quartile Quartile Range
DRE 87 38.0 12.8 35.0 35.0 40.0 5.0
TRUS 85 40.4 18.8 35.0 28.0 46.0 18.0
MRI-A3D 99 46.3 35.2 37.0 29.0 51.0 22.0
MRI-R3D 99 40.8 20.1 36.0 26.0 48.0 22.0
MRI-S3D 99 39.8 18.2 36.0 27.0 47.0 20.0
MRI-ellipse 99 45.1 23.6 39.0 29.0 53.0 24.0
MRI-bullet 99 56.9 29.7 49.0 37.0 66.0 29.0
SPV-ellipse 99 38.9 23.2 37.4 26.2 47.0 20.8
SPV-bullet 99 49.0 29.2 47.1 33.0 59.2 26.2
SPW 89 57.8 21.2 54.0 42.0 65.0 23.0

DRE, digital rectal examination; MRI-A3D, magnetic resonance imaging with automated 3D segmentation software; MRI-R3D, magnetic
resonance imaging with manual segmentation by a radiologist; MRI-S3D, magnetic resonance imaging with manual segmentation by a third-
year medical student; TRUS, transrectal ultrasound; SPV, surgical prostatectomy volume; SPW, surgical prostatectomy weight.

Figure 1. Prostate volume measure-

ments and outliers. Box and whisper plot
shows median, 25th and 75th percentiles
as boundaries to each box, and 5th and
95th percentiles as the ends of the error
bars. Outliers are indicated with individu-
al symbols.

accurate PV calculation can be a challenging task for algo-
rithms to perform correctly depending on many technical factors
(20). At least one study has shown a noncommercial automated
segmentation algorithm can produce accurate estimates of gland
size (21). Turkbey et al. used a multifeature active shape model
via MRI to estimate PV, revealing it to be quite accurate with
prostatectomy specimens for comparison, with the assurance
that automated segmentation software will save time (21). Our
results suggest that current commercially available 3D PV au-
tomated segmentation software does not perform as well as
reconfiguration by a staff radiologist, or even a trained
nonradiologist. In general, however, such software is a prom-
ising tool, providing fast, reasonably accurate anatomical
boundaries which are user-friendly to manipulate for error.
Compared to MRI-R3D, the most reliable methods for
PV estimation were MRI-S3D and MRI-ellipsoid formula.
We used a medical student as a surrogate for a 3D laborato-
ry technologist to assess if the measurements could be performed
by a trained, nonradiologist. The correlation between MRI-
S3D and MRI-R3D values was strong (ICC 0.91), indicating
that a trained, nonradiologist can satisfactorily perform this
task. Nonradiologists (medical student and technologists) took
longer on average to segment the prostate gland; however,
the reduced labor cost and radiologist interpretation time may
still result in overall savings. This could be further examined
with a time-driven activity-based costing study. With addi-
tional training and experience, it is expected that the amount
of time needed to perform segmentation by nonradiologists

4
Academic Radiology, Vol ■, No ■■, ■■ 2018
TABLE 3. Prostate Volume Measurement Inter-rater
Reliability Correlations
Volume Calculation Method Shrout-Fleiss
Compared to MRI-R3D Reliability
MRI-S3D 0.91
MRI-ellipse 0.90
SPV-ellipse 0.73
MRI-bullet 0.72
TRUS 0.71
SPW 0.70
SPV-bullet 0.66
MRI-A3D 0.38
DRE 0.33
DRE, digital rectal examination; MRI-A3D, magnetic resonance
imaging with automated 3D segmentation software; MRI-S3D, mag-
netic resonance imaging with manual segmentation by a third-year
medical student; TRUS, transrectal ultrasound; SPV, surgical pros-
tatectomy volume; SPW, surgical prostatectomy weight.
would improve and compare favorably to that of a trained
radiologist.
The ellipsoid formula had the next strongest correlation re-
liability score (ICC 0.90) compared to MRI-R3D measurement
and our data support prior work demonstrating this formula
to be a useful substitute if 3D techniques are not available (22).
Our findings are consistent with published literature showing
that TRUS and MRI using the ellipsoid formula provide ac-
curate estimates of PV for most patients, with MRI being
slightly more accurate (23). The next most reliable methods
were SPV-ellipse (ICC 0.73), MRI-bullet (ICC 0.72), and
TRUS (ICC 0.71). Lee and Chung recommended that TRUS
is preferred over MRI, as it is less expensive and almost as
accurate (24). Our data support the opinion that PV esti-
mated by in-office TRUS can be viewed as reasonably accurate
when mpMRI is not performed.
The least reliable method of determining PV in compar-
ison to MRI-R3D was DRE (ICC 0.33), underscoring the
imprecision of physical examination. The MRI-A3D was the
next least reliable method (ICC 0.38), as the automatically
generated prostate contours were not always consistent with
the true anatomic boundaries of the gland. The lower ICC
values reflect an overall higher variability of PV measure-
ments across patients and overall lower confidence in estimating
the true PV with these methods. Some types of automated
3D segmentation have been regarded as time-effective tools
with accurate representation of prostate volume (21), but we
stress the need to always check for segmentation errors when
using such software. Other studies have reported similar find-
ings of autosegmentation and its anatomic limitations (16).
Prostatectomy specimens weighed for volume also contain
periprostatic fat, seminal vesicles, and vas deferens. These speci-
mens are also noted to be dehydrated and institution-
dependent correction factors have been published in the
literature accounting for shrinkage of the prostate in the lab-
oratory (25). Our specimen weights were consistently high,
METHODS TO DETERMINE PROSTATE VOLUME
attributable to the inclusion of extraprostatic tissue. Fixation
volume loss further complicates measurement, and there-
fore, we consider SPW a less reliable indicator of in situ gland
volume. Prior studies of PV have described similar measurement
difficulty and inconsistencies when dealing with extraprostatic
tissue (9,26).
This retrospective study is not without limitations. First,
we used MRI-R3D as our reference standard instead of sur-
gical specimens because the limitations and nonstandardization
of pathology protocol for measuring tissue was not prospec-
tively standardized. Ideally, the prostatectomy specimens would
have been measured without periprostatic tissue and using water
displacement to more accurately assess volume; however, this
is time intensive and has not been standard practice at our
institution. Second, we had one staff radiologist perform the
reference measurements, and are therefore, unable to address
interobserver variation. Although, prior studies have shown
low variability among interobservers for MRI of PV thus mini-
mizing our concern for using MRI-R3D as our gold standard
(19,23). Similarly, the DRE and TRUS measurements were
determined by a single urologist, precluding interobserver anal-
ysis of these methods as well. Although all prostatectomies
were within 1 year of the mpMRI, it is possible that pros-
tate size may have increased or decreased in the interval,
depending on tumor growth or adjunctive therapy. The sample
size was small but adequate to produce results that reflect similar
findings in previous studies. Finally, we did not include PV
using CT segmentation techniques; few patients underwent
CT evaluation before surgery; therefore, no conclusions about
this method can be drawn. Future studies using standardized
surgical specimens to compare the accuracies of PV estima-
tion using segmentation software for various imaging modalities
may be beneficial.
CONCLUSIONS
Accurate PV estimates are essential because new methods of
PCa risk stratification by serum markers depend on PV for
density calculations, in addition to multiple other clinical ap-
plications (16,23). MRI-R3D segmentation was the most robust
method for PV assessment in this study. With MRI-R3D as
the reference, other methods that reliably estimated PV include
MRI-S3D measurement, MRI-ellipsoid formula, TRUS, and
SPV-ellipsoid formula. PV is commonly recorded using TRUS
for in-office measurements and our data agree that this remains
a reliable tool. This study indicates that a trained, nonradiologist
can review and adjust 3D boundaries of prostate MRI with
A3D to ensure anatomic coverage and provide a more ac-
curate PV estimate. Improved confidence in PV estimates
should benefit numerous aspects of clinical care for prostatic
disease.
ACKNOWLEDGMENTS
The corresponding author would like to thank the Betz Family
Endowment for Cancer Research for their support (RG0813-
5
BEZINQUE ET AL
1036). Funding was provided in part by the Spectrum Health
Foundation.
REFERENCES
1. McConnell JD, Bruskewitz R, Walsh P, et al. The effect of finasteride
on the risk of acute urinary retention and the need for surgical treat-
ment among men with benign prostatic hyperplasia. Finasteride Long-
Term Efficacy and Safety Study Group. N Engl J Med 1998; 338:557–
563.
2. Garvey B, Turkbey B, Truong H, et al. Clinical value of prostate segmen-
tation and volume determination on MRI in benign prostatic hyperplasia.
Diagn Interv Radiol 2014; 20:229–233.
3. Oesterling JE. Prostate specific antigen: a critical assessment of the most
useful tumor marker for adenocarcinoma of the prostate. J Urol 1991;
145:907–923.
4. Schroder FH, Hugosson J, Roobol MJ, et al. Screening and prostate-
cancer mortality in a randomized European study. N Engl J Med 2009;
360:1320–1328.
5. Welch HG, Black WC. Overdiagnosis in cancer. J Natl Cancer Inst 2010;
102:605–613.
6. Roobol MJ, Schroder FH, Hugosson J, et al. Importance of prostate volume
in the European Randomised Study of Screening for Prostate Cancer
(ERSPC) risk calculators: results from the prostate biopsy collaborative
group. World J Urol 2012; 30:149–155.
7. Bul M, Zhu X, Valdagni R, et al. Active surveillance for low-risk prostate
cancer worldwide: the PRIAS study. Eur Urol 2013; 63:597–603.
8. Roobol MJ, van Vugt HA, Loeb S, et al. Prediction of prostate cancer
risk: the role of prostate volume and digital rectal examination in the ERSPC
risk calculators. Eur Urol 2012; 61:577–583.
9. Pierorazio PM, Kinnaman MD, Wosnitzer MS, et al. Prostate volume and
pathologic prostate cancer outcomes after radical prostatectomy. Urology
2007; 70:696–701.
10. Porcaro AB, Novella G, Molinari A, et al. Prostate volume index and chronic
inflammation of the prostate type IV with respect to the risk of prostate
cancer. Urol Int 2015; 94:270–285.
11. Cher ML, Dhir A, Auffenberg GB, et al. Appropriateness criteria for active
surveillance of prostate cancer. J Urol 2017; 197:67–74.
12. Eskicorapci SY, Guliyev F, Akdogan B, et al. Individualization of the biopsy
protocol according to the prostate gland volume for prostate cancer de-
tection. J Urol 2005; 173:1536–1540.
Academic Radiology, Vol ■, No ■■, ■■ 2018
13. Kucway R, Vicini F, Huang R, et al. Prostate volume reduction with an-
drogen deprivation therapy before interstitial brachytherapy. J Urol 2002;
167:2443–2447.
14. Roehrborn CG, Girman CJ, Rhodes T, et al. Correlation between pros-
tate size estimated by digital rectal examination and measured by
transrectal ultrasound. Urology 1997; 49:548–557.
15. Ghose S, Oliver A, Marti R, et al. Statistical shape and texture model of
quadrature phase information for prostate segmentation. Int J Comput
Assist Radiol Surg. 2012; 7:43–55.
16. Toth R, Bloch BN, Genega EM, et al. Accurate prostate volume estima-
tion using multifeature active shape models on T2-weighted MRI. Acad
Radiol 2011; 18:745–754.
17. Eri LM, Thomassen H, Brennhovd B, et al. Accuracy and repeatability
of prostate volume measurements by transrectal ultrasound. Prostate
Cancer Prostatic Dis 2002; 5:273–278.
18. Littrup PJ, Kane RA, Williams CR, et al. Determination of prostate volume
with transrectal US for cancer screening. Part I. Comparison with prostate-
specific antigen assays. Radiology 1991; 178:537–542.
19. Smith WL, Lewis C, Bauman G, et al. Prostate volume contouring: a 3D
analysis of segmentation using 3DTRUS, CT, and MR. Int J Radiat Oncol
Biol Phys 2007; 67:1238–1247.
20. Ghose S, Oliver A, Marti R, et al. A survey of prostate segmentation
methodologies in ultrasound, magnetic resonance and computed to-
mography images. Comput Methods Programs Biomed 2012; 108:262–
287.
21. Turkbey B, Fotin SV, Huang RJ, et al. Fully automated prostate seg-
mentation on MRI: comparison with manual segmentation methods and
specimen volumes. AJR Am J Roentgenol 2013; 201:W720–W729.
22. Mazaheri Y, Goldman DA, Di Paolo PL, et al. Comparison of prostate
volume measured by endorectal coil MRI to prostate specimen volume
and mass after radical prostatectomy. Acad Radiol 2015; 22:556–562.
23. Paterson NR, Lavallee LT, Nguyen LN, et al. Prostate volume estima-
tions using magnetic resonance imaging and transrectal ultrasound
compared to radical prostatectomy specimens. Can Urol Assoc J 2016;
10:264–268.
24. Lee JS, Chung BH. Transrectal ultrasound versus magnetic resonance
imaging in the estimation of prostate volume as compared with radical
prostatectomy specimens. Urol Int 2007; 78:323–327.
25. Schned AR, Wheeler KJ, Hodorowski CA, et al. Tissue-shrinkage cor-
rection factor in the calculation of prostate cancer volume. Am J Surg
Pathol 1996; 20:1501–1506.
26. Hsieh TF, Chang CH, Chen WC, et al. Correlation of Gleason scores
between needle-core biopsy and radical prostatectomy specimens in pa-
tients with prostate cancer. J Chin Med Assoc 2005; 68:167–171.