British Journal of Radiology (2003) 76, 104-108
© 2003 British Institute of Radiology
doi: 10.1259/bjr/75640835
Comparative analyses of the dynamic properties of the rectum studied by cryo-sections of human cadavers and pelvic CT scans of patients
E Dale, PhD1,2,
T P Hellebust, MSc1,
Ø S Bruland, MD, PhD3 and
D R Olsen, PhD1,2
1 Centre for Training and Research in Radiotherapy, Departments of 2 Medical Physics and 3 Radiotherapy, The Norwegian Radium Hospital, University of Oslo, Oslo, Norway
Correspondence: E Dale, Department of Medical Physics, The Norwegian Radium Hospital, N-0310 Oslo, Norway
 |
Abstract
|
|---|
Optimization of radiotherapy treatment plans based on dose–volume histograms relies on accurate organ delineation. Hollow organs, such as the rectum, are difficult and time-consuming to delineate owing to unclear visualization of the border between wall tissue and filling. Automated hollow organ delineation would be a valuable tool, but its development depends upon improved understanding of the dynamics of the rectum in response to filling. Two reasonable assumptions proposed in the literature are that (1) the rectal wall tissue along a constant length of the rectal cylinder is preserved over time and (2) the rectal wall tissue is distributed homogeneously along the cylinder. Therefore, variations in wall thickness can be explained by variable rectal filling. To investigate these assumptions, transversal cross-sectional areas enclosed by the outer contour (Aout) and inner contour (Ain) of the rectum were recorded from digital photographs of cadaver cryo-sections from the U.S. National Library of Medicine's Visible Human Project. In addition, Aout and Ain were recorded from 19 CT scans of 5 of our own patients. The transversal cross-sectional area of the wall of the rectum, Awall=Aout-Ain, was calculated. The data derived both from cryo-sections and repetitive CT scans of patients, revealed that there was a significant correlation between Awall and Aout, in contradiction to assumption (1) stated above (male cryo-sections: p<0.001, female cryo-sections: p=0.03, repetitive CT scans p<0.001). Moreover, the mean Awall calculated from one CT scan differed significantly from the mean Awall from other CT scans and was correlated with the mean Aout, i.e. rectal filling (p<0.001). This finding was confirmed by careful analysis of another study (p=0.001) and opposes assumption (2). Hence, the amount of wall tissue within a constant length of rectum is not preserved over time, but increases with increased filling. This implies that the longitudinal length of the rectum decreases in response to distension of the organ.
|
Introduction
|
|---|
In current radiotherapy practice, the use of dose–volume histograms (DVH) derived from dose calculations and 3D treatment planning in optimization of treatment plans, is essential. A DVH describes the radiation dose distribution within an organ at risk (OAR) or a target volume and may be used to calculate the normal tissue complication probability (NTCP) or the tumour control probability (TCP). The validity of DVHs relies on accurate organ delineation on the CT images in the treatment planning system [1–6]. Furthermore, the materials filling hollow organs are irrelevant from a clinical point of view and should be excluded from the DVH. Hence, different concepts have been introduced, such as the dose–wall histogram (DWH) [1] and the dose–surface histogram (DSH) [2]. To calculate a DSH, only the outer contours of an organ have to be delineated, whereas the DWH requires both the outer and inner contours. Theoretically, the DWH is a more accurate representation of the dose distribution than the DSH, as variable wall thickness of the organ is accounted for. However, the fact that the organ wall may be quite thin and close to the CT image resolution, with often inferior contrast between the inner layer of a hollow organ and the filling, makes it difficult and time-consuming to delineate the inner contour. The outer contour, which demarcates the boundary between the outer longitudinal rectal muscular layer and adjacent fatty tissue, is delineated with rather high precision [7].
One organ of particular interest in pelvic irradiation is the rectum [8, 9]. Recently, automated delineation of the rectal inner contours based on manually drawn outer contours and assumptions regarding the properties of the rectal wall in response to filling, was reported by Meijer et al [7]. They proposed a cylindrical shell model for rectum with an elastic wall undergoing changes in shape and size as a response to variable filling. The cylinder comprised small transversal slices, each assumed to have the same constant wall volume regardless of luminal filling. Hence, knowing the location of only the outer contour enabled the authors to calculate the position of the inner contour assuming a constant slice (annulus) volume [7].
In the present study our aim was to challenge the assumption that rectal wall tissue is distributed homogeneously along the rectal cylinder axis. We carried out a comparative study using photographs and corresponding CT scans of 1 mm transversal cryo-sections of a male and female cadaver (U.S. National Library of Medicine's Visible Human Project). In addition, repetitive CT examinations of cervix cancer patients obtained in a previous study on brachytherapy treatment reproducibility, were investigated [10].
|
Methods and materials
|
|---|
Photographs of cadaver cryosections
Digital photographs of cadaver slices and corresponding CT images from the U.S. National Library of Medicine's Visible Human Project were imported into GNU Image Manipulation Program v1.2 (GIMP; freeware by P Mattis and S Kimball). Outer and inner contours of rectum and bladder were carefully delineated every 3 mm by a medical physicist (ED) in collaboration with an experienced oncologist (ØSB). The most distal image on which the rectum was delineated was approximately 20 mm proximal to the distal part of the anal sphincter musculature, whereas the most proximal image was at the sigmoid flexure. The number of image pixels within each contour was obtained from the GIMP software. Cadaver image resolution was 0.33 mm pixel-1, CT image resolution 1.0 mm pixel-1, and the area within each contour could thus be calculated:
Aout=Transversal cross-sectional area enclosed by the outer contour of the organ, i.e. including wall tissue and lumen (filling).
Ain=Transversal cross-sectional area enclosed by the inner contour of the organ, i.e. the transversal area of the lumen.
Awall=Aout-Ain, the transversal cross-sectional area of the organ wall.
Repetitive CT examinations
Repetitive CT examinations from a previous clinical study of cervix cancer patients [10] were used to investigate the variation of Awall at different time points. These CT examinations were performed without contrast with 3–5 mm slice thickness and 0.8–0.9 mm pixel-1 image resolution. Minimum time between CT scans was 1 day and maximum time was 24 days (Table 1
). Aout, Ain and Awall were obtained as described above, from organ contours delineated on the PLATO dose-planning system (Nucletron BV, Veenendaal, The Netherlands). CT images with insufficient contrast between the organ wall and lumen were not used.
Transversal cross-sectional area of the rectal wall
Values of Awall as a function of caudo-cranial position (CCP) were plotted to investigate the assumptions of Meijer et al: "(1) The basic shape of the rectum is a hollow cylinder ... (2) Rectal filling only stretches the rectum wall transversely ... Consequently, the rectal filling does not affect the length of the rectum ... we assume that the total amount of cross-sected rectum wall tissue in every intersection perpendicular to the central axis is constant throughout the entire rectum." [7]. If these assumptions are correct, Awall and CCP are independent and should describe a horizontal linear relationship in a scatter plot. Moreover, Awall should not be affected by local filling measured by Aout.
Statistics
Correlation analysis was performed with Pearson's (linear) correlation coefficient, R2. A test on difference between two R2 values was based on Fisher's finding that the statistic 
x ln[(1+R)/(1-R)] is approximately normally distributed with mean x ln[(1+R)/(1-R)] and variance 1/(n-3) [11]. Comparison between mean values was performed using the t-test (two-tailed) or the paired t-test when appropriate. A p-value smaller than 0.05 was considered statistically significant. The standard deviation (SD) divided by the mean value was denoted SD% (also known as coefficient of variation). Calculations were performed with Microsoft Excel 97.
|
Results
|
|---|
Correlation between transversal areas from cadaver cryo-sections and corresponding areas from CT images of cadavers
CT has traditionally been recognized as a reliable imaging method, and measurements of Aout and Ain from cryo-sections as compared with CT images revealed high correlation coefficients (Pearson R2 0.94 and 0.93) for the images of the female, although R2 were smaller for the images of the male (Table 2). As expected, the correlation between Awall from cryo-sections and CT scans were lower because of the smaller variation in Awall. Inclusion of only Aout and Ain from the proximal 2/3 of the rectum in the correlation analysis revealed higher coefficients as compared with corresponding coefficients for the distal 2/3 of rectum. For instance, a test on the difference between the correlation coefficients 0.72 (proximal 2/3) and 0.50 (distal 2/3) showed a statistically significant difference (p<0.001). Therefore, there is a larger uncertainty associated with delineation of the distal part of the rectum on CT images.
View this table:
[in this window]
[in a new window]
|
Table 2. Correlation analysis between transversal areas from Visible Human cryo-sections and corresponding CT images
|
|
Variation of Awall: photographs of cadaver cryo-sections
The main purpose of this study was to investigate whether the transversal cross-sectional area of the rectum wall, Awall is independent of local filling, Aout, and/or CCP. Awall and Aout were obtained by delineating the outer and inner contours on the Visible Human image collection of cryo-sections (Figure 1). The key finding is that Awall and Aout is significantly correlated within a specific image set, i.e. the local wall volume of a single tissue slice of constant thickness increases with increasing rectal filling (Table 3 and Figure 2). Moreover, there is also an association between Awall and lower CCP, which means that there is more wall tissue in the distal region of the rectum (Table 3). Multivariate analysis including both Aout and CCP, revealed that CCP remained a significant variable predicting Awall, although less important than Aout (data not shown).
View larger version (29K):
[in this window]
[in a new window]
|
Figure 1. transversal cross-sectional area of the wall (Aout) and Transversal cross-sectional areas enclosed by the outer contour (Awall) from both male and female cryo-sections as a function of caudo-cranial position. There was a substantial difference between the male and female rectal filling. The position of the prostate was included in the figure.
|
|
View this table:
[in this window]
[in a new window]
|
Table 3. Descriptive data of cadaver cryo-sections and repetitive CT data of patients including correlation analyses. Descriptive measures calculated from transversal cross-sectional areas enclosed by the outer contour (Aout) and transversal cross-sectional area of the wall (Awall) data. Data from both Visible Human cryo-sections and CT examinations on our own patients, are included. In the rightmost columns the p-values from tests on correlation between Aout, Awall and caudo-cranial position (CCP) and correlation between Aout and Awall are shown. (+) indicates a positive correlation and (-) indicates a negative correlation. To perform this calculation for the repetitive CT data, the area measurements have been normalized according to the mean area of each CT scan.
|
|
View larger version (16K):
[in this window]
[in a new window]
|
Figure 2. Transversal cross-sectional area of the wall (Awall) vs transversal cross-sectional areas enclosed by the outer contour (Aout) determined from male cryo-sections. The two filled data points are derived from the two cryo-sections shown in Figure 3.
|
|
As stated previously, for traditional CT images, the outer contour is delineated with high precision, while the delineation of the inner contour is associated with uncertainty. However, on the cryo-sections, the inner contour is also reliably delineated. For the male cryo-sections the mean Awall along the caudo-cranial length was 3.9 cm2 (Table 3
). In Figure 3, two cryo-sections representing two extreme data points from the plot of Awall vs Aout (Figure 2), are shown. On purpose, the inner contours have been manually delineated to fix Awall equal to the mean value Awall of 3.9 cm2 of this specific image set (Table 3). This cannot be achieved unless the inner contour is erroneously located inside the lumen (collapsed lumen; Figure 3a) or inside the wall tissue (filled lumen; Figure 3b), clearly demonstrating that Awall increases with Aout.
View larger version (86K):
[in this window]
[in a new window]
|
Figure 3. Delineation of the outer and inner contours (yellow) of (a) an empty and (b) a filled rectum on photos of the male cryo-sections. The caudo-cranial distance between the images is 5.9 cm. For both images, the operator has delineated the inner contours erroneously on purpose, to fix the transversal wall area, Awall, equal to the mean Awall of the whole image set of 3.9 cm2 (Table 3).
|
|
Variation of Awall: repetitive CT examinations
Obviously, the cryo-sections cannot contribute to information on the variation of Awall at different time points. However, repetitive CT scans performed on seven patients at our institution revealed that the mean Awall within one CT scan differed significantly from the mean Awall measured from another CT scan of the same patient (Figure 4). Moreover, careful evaluation of the study of Meijer et al [7] provided more data supporting this finding (Figure 4).
View larger version (20K):
[in this window]
[in a new window]
|
Figure 4. Mean transversal cross-sectional area of the wall (Awall) vs mean transversal cross-sectional areas enclosed by the outer contour (Aout) determined from repetetive CT scans performed at our institution (•) and obtained from Figure 4 in the study by Meijer et al [7] ( ).
|
|
Analysis of the relationships between Aout, Awall and CCP within a patient CT examination gave the same trends as found from the photographs of cryo-sections (Table 3).
|
Discussion
|
|---|
In general, there was good agreement between transversal areas measured from cryo-sections as compared with those measured from corresponding CT images of the same individual. For the female, only 6–7% of the variation in areas delineated on CT images could not be explained by area variation in the cryo-sections (Table 2). However, the match between male cryo-sections and CT images was lower, but there was a better agreement in the proximal than the distal part of the rectum. In the most distal region towards the anus, the lumen may be quite collapsed with a little faecal material with similar CT numbers as the extensive surrounding musculature (levator ani muscle and the external and internal sphincter), making it a difficult task to delineate the internal contour on CT images.
Within a cryo-section image set, a correlation between Awall and Aout was found. In addition, both large Awall and Aout were independently associated with the distal part of the rectum, i.e. low CCP. Awall increases because the rectal wall gets thicker as more of the levator ani muscle and the external and internal sphincter musculature are included on images of the distal part of the rectum. The increase in Aout towards the anus may partly be explained by the corresponding increase in Awall, but also by the displacement of faecal content distally both by rythmical, intestinal smooth muscle contractions and by gravity. The association between Aout and CCP was generally stronger than the association between Awall and CCP. On the other hand, the correlation between Awall and Aout was much stronger than their correlation with CCP. The relationship between Awall and Aout was also notified by Meijer et al but was explained as a delineation artefact. They suggested that the operator tends to add a safety margin when delineating the poorly visualized inner contour of a heavily filled rectum with a thin wall. Our data indicate that the variation in Awall cannot be explained by a delineation artefact alone (Figure 3).
A key finding of the present study is the significant variation in the mean transversal wall area (Awall) of the rectum from CT scan to CT scan of the same patient. Moreover, the mean Awall increased with increasing mean area within the outer rectal circumference (Aout), i.e. rectal filling. This result violates Meijer et al's assumption of preserved rectal wall tissue volume within a CT slice of constant thickness. The correlation between the mean Awall and mean Aout was also observed by careful examination of their Figure 4 [7]. These findings support the statement that the volume of the rectal wall tissue is not preserved over time. However, such a view is based on the idea that the length of the rectal cylinder is constant over time and that rectum contours should be delineated on the same number of CT images at different CT examinations (wall volume=mean Awall x length).
The correlation between Awall and Aout found in this and another study [7], may be explained by the rectal length not being constant but rather varying in response to filling. When delineating the rectum for, for example, radiotherapy planning purposes, there are no anatomical clues to the location of the most proximal part of the rectum, except the quite uncertain position of the sigmoid flexure. If the rectal length is kept fairly constant between different CT scans, a seemingly variable total wall volume, which increases with filling, is seen. We found inter-CT variations in the mean Awall of typically 20%, indicating that a 12 cm long rectum varies ±2.4 cm between maximum and minimum filling. This represents a rough estimate and may be artificially enlarged by other systematic effects such as a delination artefact outlined above. Another possibility is errors due to the curving nature of the rectum, which is not accounted for in the present study. However, a correction factor [7] has to be equal for both areas (Awall and Aout) so this is an unlikely explanation. We believe that the correlation between the Awall and Aout is too strong to be explained solely by systematic errors in our methods (Figure 4). The final verification may be an experiment in which a high density marker (surgical clip) is placed on the proximal end of the rectum. The rectal length may thus be monitored during variable rectal filling by several CT scans. If our results are confirmed, the number of CT slices on which the rectum is delineated should vary according to the filling status to keep the volume of wall tissue constant at different CT scans. Moreover, an automatic delineation technique of the inner rectum contour based on the preservation of wall tissue in a constant-thickness CT slice would not be correct.
|
Acknowledgments
|
|---|
We thank our LINUX expert Torbjørn Sund, MSc, for invaluable assistance. We are also grateful to Professor Per Holck, Department of Anatomy, Faculty of Medicine, University of Oslo, for fruitful discussions on rectal anatomy.
|
Footnotes
|
|---|
This work was supported by the Norwegian Cancer Society. 
Received for publication March 21, 2002.
Revision received October 9, 2002.
Accepted for publication November 4, 2002.
|
References
|
|---|
- Lebesque JV, Bruce AM, Kroes APG, Touw A, Shouman T, Van Herk M. Variation in volumes, dose-volume histograms, and estimated normal tissue complication probabilities of rectum and bladder during conformal radiotherapy of T3 prostate cancer. Int J Radiat Oncol Biol Phys 1995;33:1109–19.[Medline]
- Lu Y, Li S, Spelbring D, Song P, Vijayakumar S, Pelizzari C, Chen GTY. Dose-surface histograms as treatment planning tool for prostate conformal therapy. Med Phys 1995;22:279–84.[Medline]
- Lu Y, Song PY, Li S, Spelbring DR, Vijayakumar S, Haraf DJ, Chen GTY. A method of analyzing rectal surface area irradiated and rectal complication in prostate conformal radiotherapy. Int J Radiat Oncol Biol Phys 1995;33:1121–5.[Medline]
- Li S, Boyer A, Lu Y, Chen GTY. Analysis of the dose-surface histogram and dose-wall histogram for the rectum and bladder. Med Phys 1997;24:1107–16.[Medline]
- Ting JY, Wu X, Fiedler JA, Yang C, Watzich ML, Markoe A. Dose-volume histograms for bladder and rectum. Int J Radiat Oncol Biol Phys 1997;38:1105–11.[Medline]
- MacKay RI, Hendry JH, Moore CJ, Williams PC, Read G. Predicting late rectal complications following prostate conformal radiotherapy using biologically effective doses and normalized dose-surface histograms. Br J Radiol 1997;70:517–26.[Abstract]
- Meijer GJ, Van den Brink M, Hoogeman MS, Meinders J, Lebesque JV. Dose wall histograms and normalized dose surface histograms for the rectum. A new method to analyze the dose distribution over the rectum in conformal radiotherapy. Int J Radiat Oncol Biol Phys 1999;45:1073–80.[Medline]
- Pedersen D, Bentzen SM, Overgaard J. Early and late radiotherapeutic morbidity in 442 consecutive patients with locally advanced carcinoma of the uterine cervix. Int J Radiat Oncol Biol Phys 1994;29:941–52.[Medline]
- Schulteiss TE, Hanks GE, Hunt MA, Lee WR. Incidence of and factors related to late complications in conformal and conventional radiation treatment of cancer of the prostate. Int J Radiat Oncol Biol Phys 1995;32:643–9.[CrossRef][Medline]
- Hellebust TP, Dale E, Skjønsberg A, Olsen DR. Inter-fraction variations in rectum and bladder volumes and dose distributions during high dose rate brachytherapy treatment of the uterine cervix investigated by repetitive CT-examinations. Radiother Oncol 2001;60:273–80.[CrossRef][Medline]
- Larsen RJ, Marx LM. An introduction to mathematical statistics and its applications. Upper Saddle River, NJ: Prentice-Hall, 1986.
This article has been cited by other articles:
|
|
|
|
 
E Dale, T P Hellebust, O S Bruland, and D R Olsen
Comparative analyses of the dynamic properties of the bladder wall studied by repetitive pelvic CT scans of patients and cryo-sections of cadavers
Br. J. Radiol.,
June 1, 2005;
78(930):
528 - 532.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
