| Home | E-Submission | Sitemap | Contact Us |  
Korean Journal of Clinical Oncology > Article
Choi, Park, Jeon, Lee, and Baek: Measurement of human peritoneal surface area using artificial intelligence software in abdominal computed tomography



The calculation of the intraperitoneal organ surface area is important for understanding their anatomical structure and for conducting basic and clinical studies on diseases related to the peritoneum. To measure the intraperitoneal surface area in a living body by applying artificial intelligence (AI) techniques to the abdominal cavity using computed tomography and to prepare clinical indicators for application to the abdominal cavity.


Computed tomography images of ten adult males and females with a healthy body mass index and ten adults diagnosed with colon cancer were analyzed to determine the peritoneal and intraperitoneal surface areas of the organs. The peritoneal surface was segmented and three-dimensionally modeled using AI medical imaging software. In addition to manual work, three-dimensional editing, filtering, and connectivity checks were performed to improve work efficiency and accuracy. The colon and small intestine surface areas were calculated using the mean length and diameter. The abdominal cavity surface area was defined as the sum of the intraperitoneal area and the surface areas of each organ.


The mean peritoneal surface area of all participants was measured as 10,039±241 cm2 (males 10,224±171 cm2 and females 9,854±134 cm2). Males had a 3.7% larger peritoneal surface area than females, with a statistically significant difference (P<0.001).


The abdominal cavity surface area can be measured using AI techniques and is expected to be used as basic data for clinical applications.


The calculation of the intraperitoneal organ surface area is important for understanding their anatomical structure and for conducting basic and clinical studies on diseases related to the peritoneum. In the case of hyperthermic intraperitoneal chemotherapy (HIPEC) for the treatment of peritoneal carcinomatosis, the volume of HIPEC solution filling the abdominal cavity has not been objectively presented [1]. Furthermore, there is a lack of evidence regarding the effective dose of an anti-adhesive agent, which acts as a physical barrier at the anticipated adhesion sites to prevent adhesion formation between adjacent tissues for a specific duration.
To determine the effective dose of the intraperitoneally applied agent, it is necessary to measure the entire peritoneal surface area or the specific surface area to be treated. However, previous studies on the peritoneal area have been limited to results obtained from cadavers, and no studies have been conducted on noninvasive segmentation and surface area measurement of the abdominal cavity in living individuals. Utilizing computed tomography (CT) for measurements of the peritoneal area offers the advantage of in vivo assessment.
The objective of this study was to accurately measure the intraperitoneal surface area in living individuals using CT and artificial intelligence (AI) techniques applied to the abdominal cavity and to establish clinical indicators applicable to the abdominal cavity.



Adult males and females aged ≥ 19 years who had undergone abdominal CT were enrolled in the study. We enrolled 10 healthy adults and 10 adults diagnosed with colon cancer. The inclusion criteria for healthy person were as follows: a body mass index of ≥ 18.5 kg/m2 or ≤24.9 kg/m2 and is free from liver, kidney, nervous system, respiratory, endocrine, hematologic, urinary, or cardiovascular diseases. The inclusion criteria for the group diagnosed with colon cancer were as follows: histopathologically diagnosed colon cancer and is free from liver, kidney, nervous system, respiratory, endocrine, hematologic, urinary, or cardiovascular diseases. CT images of 10 healthy adults (five males and five females, mean age of 52.9 years) and 10 colon cancer patients (five males and five females, mean age of 64.8 years) were analyzed to determine the peritoneal and intraperitoneal surface areas of the organs. This study was approved by the Institutional Review Board (IRB No. GCIRB2019-037) of Gachon University Gil Medical Center. Written informed consent was exempted by the approval of Gachon University Gil Medical Center IRB.

Measurement of the peritoneal surface area

The total peritoneal surface area was defined as the sum of the visceral and parietal peritoneal surface areas. The visceral peritoneal surface includes organs such as the stomach, liver, gall bladder, spleen, mesentery, small bowel, colon, uterus, and intraperitoneal bladder. The parietal peritoneal surface area included the diaphragmatic walls, anterolateral and dorsal walls, and lateral pelvic walls. The peritoneal surface was segmented following the instructions provided by an abdominal radiologist. The small bowel loop surface area was obtained from a previous study [26]. To determine the peritoneal surfaces of the small and large bowel loops, a cylindrical bowel model was devised, as shown in Fig. 1. The surface area was calculated by using the dimensions (diameters and lengths) of the small and large intestines.

Segmentation of the abdominal cavity and organs using an AI algorithm

The AI medical imaging software MEDIP (MEDICALIP Co., Ltd.) and Mimics software (version 23.0, Materialise N.V.) were used to segment the abdominal cavity area at 5-mm intervals, allowing for the calculation of the surface area through three-dimensional (3D) modeling (Fig. 2). Alongside manual work, 3D editing, filtering, and connectivity checks were performed to enhance both work efficacy and accuracy.
Using the MEDIP software, segmentation of the visceral organs proceeded in the following order: draw cut, hole filling, erosion, dilation, region growing, and rolling ball (Table 1). To perform the segmentation of the visceral peritoneal organs, the regions of interest and exclusion were first specified using draw cuts in the transverse, coronal, and sagittal planes. The lambda value, which is the degree of expansion from the selected seed point, was adjusted from 20 to 50, based on the size of the dividing organ. Noise, which is an unwanted region, is generated when a draw cut is applied because the seed point is not designated for all slices in the region of interest. To remove noise, empty pixels inside the region of interest were filled using a hole-filling method, and then the entire surface was eroded by applying an erosion method. In this process, pixels on the surface of the noise area outside the region of interest are eroded, and a part connected to the region of interest is separated (Fig. 3A and B). In addition, when the region growth method was performed by selecting a seed point within the region of interest, the noise region separated in the previous step was removed. After repeating this process, the rolling ball technique was applied, followed by the dilation and erosion techniques to obtain the final result (Fig. 3C).
The area of the parietal peritoneum was measured in the following order: Threshold setting, split mask performance, and multiple slice editing using the Mimics software (Table 1). Because parietal peritoneal segmentation involves organs and air inside the intestine, the threshold was set to −1,024 to 300 HU, which is wider than the soft tissue range, and the work area was adjusted to the abdominal cavity range (Fig. 4A and B). After the threshold-based region segmentation was performed, only the region of interest was segmented in detail using a split mask (Fig. 4C and D). Segmentation was performed based on transverse sections, excluding the retroperitoneal organs. Split mask is performed by designating the inside of the peritoneum and the retroperitoneal cavity at 10-slice intervals. If a part other than the region of interest is included in the split mask result, it is modified through multiple slice edits. The modified results between the first and last slices that marked the modification were reflected. Contour-matching of the work area and area of interest was confirmed to achieve the final segmentation result (Fig. 4E and F).

Statistical analysis

Categorical variables were compared as counts and percentages, and associations were tested using the chi-squared test or Fisher exact test. Continuous variables were reported as mean±standard deviation, and differences between groups were analyzed using the Student t-test or the Mann-Whitney U test, as appropriate. The correlations were evaluated using Spearman correlation coefficients. A P-value of <0.05 was considered a statistically significant difference. IBM SPSS Statistics for Windows, version 25.0 (IBM Corp.) was used for the statistical analysis.


Small intestine modeling

Although there is a difference in measurement techniques and the size of the study groups, most studies that measure the length of the small intestine employ direct measurements using methods such as U-tape, from the ligament of Treitz to the ileocecal valve [25]. Table 2 shows the length of the small intestine from the previous studies. Additionally, Haworth et al. [6] reported the diameter of the small intestine in healthy individuals, including 77 adults with ages ranging from 19 to 77 years, to be 23.1±1.9 mm. Based on these reports, the mean length of the small intestine was determined to be 615.4 cm (623.0 cm for 494 males and 611.4 cm for 895 females, respectively), while the diameter was 2.3 cm. By substituting these values into the equation used to calculate the cylinder surface area, the mean small intestine surface area was determined to be 4,444 cm2 (4,499 cm2 of males and 4,416 cm2 of females, respectively).

Colon modeling

Phillips et al. [7] conducted a cadaver study to measure the length of the large intestine in 35 individuals (18 males and 17 females) with a mean age of 84 years by using direct measurements with tape. The mean length of the transverse colon and sigmoid colon was found to be 50.2 cm and 38.3 cm, respectively, and the mean diameter of the large intestine was measured at 5.5 cm. By substituting these values into the equation and summing up the results, the mean intraperitoneal colon surface area was calculated to be 1,528 cm2.

Total peritoneal surface area

The characteristics of the study subjects are presented in Table 3. The mean peritoneal surface area of all participants was measured as 10,039±241 cm2 (males 10,224±171 cm2 and females 9,854±134 cm2). Males had a 3.7% larger peritoneal surface area than females, with a statistically significant difference (P<0.001). Fig. 5 shows the modeling of peritoneal surface. The peritoneal surface consists of the liver, gallbladder, stomach, spleen, uterus (female) and parietal peritoneum, and the total surface area is obtained by adding them all. There was a significant correlation between the peritoneal surface and the height (r=0.768, P<0.001), the body weight (r=0.631, P<0.05), and the parietal peritoneum (r=0.887, P<0.001). However, there was no statistically significant correlation between the peritoneal surface and the surface of other peritoneal organs such as the liver, gall bladder, stomach, or spleen.


In the case of anti-adhesion agents, the efficacy of these agents in preventing adhesion has been demonstrated through several studies on efficacy and safety. However, there is a lack of studies determining the most effective doses for these agents according to the body surface area, as far as we know. Previous studies consistently used company-recommended doses, regardless of surgical site or scope [8,9]. Surgeons typically rely on empirical determination when selecting a dose among several commercially available options provided by the company.
Previous studies have primarily focused on measuring the peritoneal surface area postmortem or in animals, with limited studies conducted in vivo. However, in recent decades, there has been an increase in anthropometric studies that use CT 3D reconstruction methods and volume rendering techniques to measure the surface area or volume of organs within the abdominal cavity. Recent studies have shown that 3D volume rendering images can provide additional anatomical detail to cadaver and liver individuals [10,11]. However, the reconstruction process can cause random noise and degrade the quality of the image, in which case AI algorithms and software can help solve the problem. Our study suggests the possibility of using AI technique to calculate the peritoneal surface to reduce noise in CT reconstruction images. Measurement of the peritoneal surface is more important for obtaining data from live individuals than from cadavers to determine the dosage of drugs before the use of anti-adhesion agents or HIPEC during surgery.
In our study, the peritoneal surface area differed from previous studies; Rubin et al. [12] reported a mean value of 7,791 cm2, which was smaller than our findings. On the other hand, Albanese et al. [13] reported a larger mean peritoneal surface area of 14,323 cm2 in their study involving 10 female human cadavers without abdominal diseases. They directly measured the peritoneal surface area using cellophane film and found it to be 14,324±824 cm2, which is approximately 42% larger than our CT-base measurement of 10,039±241 cm2. This difference was presumed to be due to the inclusion of measurements of the pancreas, descending colon, and urinary bladder in the previous study. Additionally, CT cannot distinguish between the two layers of the greater omentum, lesser omentum, mesentery proper, and mesocolon. Therefore, a smaller surface area can be measured. Similarly, Esperanca and Collins [14] reported a larger mean peritoneal surface area of 10,379 cm2.
In our study, the mean body weights of the participants were 58.9 kg, which was heavier compared to the body weights reported in the studies by Esperanca and Collins (58.2 kg, six cadavers) [14] and Albanese et al. (49.5 kg, 10 cadavers, P<0.01) [13] but lighter than the body weights reported by Rubin et al. (76.8 kg, eight cadavers, P<0.05) [12]. Albanese et al. [13] showed a correlation between the measurement of peritoneal surface area and body weight (r=0.793, P<0.04). Similarly, in our study, there was a significant correlation between the peritoneal surface and the body weight (r=0.631, P<0.05). In addition, our study showed a significant association between peritoneal surface and height (r=0.768, P<0.001). However, there was a limit to the ability to derive correlations by conducting studies in a small number of patients with limited conditions. Because of the complexity of the abdominal cavity, it is important to conduct more studies in patients with various conditions.
In cadaver,s measured studies, peritoneal surface measurement was performed by attaching cellophane film to the peritoneal surface to measure the area [1214]. This may be the most accurate method, but there may be a disadvantage that it takes too much time and effort to measure a cadaver. In addition, there is a disadvantage that it is not possible to measure the peritoneum of a living patient. For measurements using live patients, the length of the intestine has been measured by surgeons in the operating room [24]. While this provides an accurate measurement because it is taken directly by the surgeon performing the surgery, it has some limitations because it is based on an actual live patient. Peritoneal surface measurement using CT can be used for both cadavers and live individuals, and analysis is possible within a short time. Although the accuracy of measurement may be lower than that of using cellophane films for cadavers, random noise arising from CT can be overcome with AI algorithms and software. In living patients, results close to the actual value can be achieved without surgical treatment. However, measuring the area of a moving organ such as a bowl has limitations in CT.
This study had some limitations. First, only a small number of participants were enrolled, and the ranges between the minimum and maximum height and body weight values were narrow. Thus, there are limitations to assessing the correlation between the measurement of the peritoneal surface and other parameters. Second, the small and large bowel loop surface areas were obtained based on previous literature using the intestine model. In the present study, it was very difficult to segment the entire bowel loop in CT using the AI technique without additional procedures such as bowel preparation or distension. To overcome this limitation, further studies are needed to plan bowel segmentation with an AI program after bowel preparation and continuous distension.
Despite these limitations, a non-cadaveric and non-eviscerated measurement tool of the peritoneal surface is important for calculating the peritoneal area of patients undergoing peritoneal dialysis and chemotherapy, such as HIPEC, during surgery. Thus, AI techniques using CT scans can be applied to a wider range of patients and performed repeatedly, and the demand for these techniques will continue to increase.
In conclusion, the abdominal cavity surface area can be measured using AI techniques and is expected to be used as basic clinical data in the future.


No potential conflict of interest relevant to this article was reported.




1. Di Giorgio A, Naticchioni E, Biacchi D, Sibio S, Accarpio F, Rocco M, et al. Cytoreductive surgery (peritonectomy procedures) combined with hyperthermic intraperitoneal chemotherapy (HIPEC) in the treatment of diffuse peritoneal carcinomatosis from ovarian cancer. Cancer 2008;113:315-25.
crossref pmid
2. Hosseinpour M, Behdad A. Evaluation of small bowel measurement in alive patients. Surg Radiol Anat 2008;30:653-5.
crossref pmid
3. Teitelbaum EN, Vaziri K, Zettervall S, Amdur RL, Orkin BA. Intraoperative small bowel length measurements and analysis of demographic predictors of increased length. Clin Anat 2013;26:827-32.
crossref pmid
4. Tacchino RM. Bowel length: measurement, predictors, and impact on bariatric and metabolic surgery. Surg Obes Relat Dis 2015;11:328-34.
5. Bekheit M, Ibrahim MY, Tobar W, Galal I, Elward AS. Correlation between the total small bowel length and anthropometric measures in living humans: cross-sectional study. Obes Surg 2020;30:681-6.
crossref pmid
6. Haworth EM, Hodson CJ, Joyce CR, Pringle EM, Solimano G, Young WF. Radiological measurement of small bowel calibre in normal subjects according to age. Clin Radiol 1967;18:417-21.
crossref pmid
7. Phillips M, Patel A, Meredith P, Will O, Brassett C. Segmental colonic length and mobility. Ann R Coll Surg Engl 2015;97:439-44.
crossref pmid pmc
8. Kim SG, Song KY, Lee HH, Kim EY, Lee JH, Jeon HM, et al. Efficacy of an antiadhesive agent for the prevention of intra-abdominal adhesions after radical gastrectomy: a prospective randomized, multicenter trial. Medicine (Baltimore) 2019;98:e15141.
pmid pmc
9. Kramer B, Andress J, Neis F, Hoffmann S, Brucker S, Kommoss S, et al. Adhesion prevention after endometriosis surgery: results of a randomized, controlled clinical trial with second-look laparoscopy. Langenbecks Arch Surg 2021;406:2133-43.
crossref pmid pmc
10. Ebert LC, Schweitzer W, Gascho D, Ruder TD, Flach PM, Thali MJ, et al. Forensic 3D visualization of CT data using cinematic volume rendering: a preliminary study. AJR Am J Roentgenol 2017;208:233-40.
crossref pmid
11. Johnson PT, Schneider R, Lugo-Fagundo C, Johnson MB, Fishman EK. MDCT angiography with 3D rendering: a novel cinematic rendering algorithm for enhanced anatomic detail. AJR Am J Roentgenol 2017;209:309-12.
crossref pmid
12. Rubin J, Clawson M, Planch A, Jones Q. Measurements of peritoneal surface area in man and rat. Am J Med Sci 1988;295:453-8.
crossref pmid
13. Albanese AM, Albanese EF, Mino JH, Gomez E, Gomez M, Zandomeni M, et al. Peritoneal surface area: measurements of 40 structures covered by peritoneum: correlation between total peritoneal surface area and the surface calculated by formulas. Surg Radiol Anat 2009;31:369-77.
14. Esperanca MJ, Collins DL. Peritoneal dialysis efficiency in relation to body weight. J Pediatr Surg 1966;1:162-9.

Fig. 1
Bowel modeling and surface area of the cylindrical structure.
Fig. 2
Measurement of surface area based on the triangular structure of the three-dimensional model.
Fig. 3
Example of hole filling and erosion during liver segmentation. (A) Segmentation results with a noise region. (B) The noise region is separated from the region of interest through the erosion process (red circle). (C) Final result of liver segmentation.
Fig. 4
Example of segmentation of the parietal peritoneum. (A) Coronal and (B) transverse plane of the threshold-based segmentation. (C) Coronal and (D) transverse plane of the results of split mask application. (E) Coronal and (F) transverse plane of the final results of segmentation.
Fig. 5
Peritoneal surface modeling. (A) Liver and gallbladder. (B) Stomach and spleen. (C) Uterus and parietal peritoneum.
Table 1
Peritoneal segmentation technique
Structure Order Algorithm Descriptions
Peritoneal organ 1 Draw cut It is a machine learning algorithm that separates foreground and background by configuring each pixel of the image as a node of a graph. It is a semi-automatic area division method that spreads from the seed point specified by the user.
2 Hole filling It is an algorithm with the ability to fill empty pixels in 2D or 3D images. For 2D images, it can be selectively applied in transverse, coronal, and sagittal planes.
3 Erosion The process of reducing the area of interest that has been worked on using a function that erodes the image boundary of the foreground and converts it into the background.
4 Dilation The task of enlarging the area of interest that has been worked on by expanding the boundary of the foreground image.
5 Region growing An image segmentation method that involves selecting one or multiple seed points and expanding them into pixel areas with similar properties.
6 Rolling ball This is an algorithm that smoothens the surface of the segmented image. It is a method that acts on all planes to eliminate stepping and smoothens the surface.

Parietal peritoneum 1 Threshold setting The process of setting an optimized working area by adjusting the upper and lower limits of the threshold.
2 Split mask It is an algorithm that can divide two or more regions of interest within one working mask. It performs semi-automatic regional segmentation based on user-specified regions.
3 Multiple slice edit It is an algorithm of a manual area segmentation function. It is a method used for obtaining results for entire slices by calculating and enhancing regions of interest selectively marked on some slices.
Table 2
Length of small intestine
Author (year) Sex No. of cases Length (cm)
Hosseinpour and Behdad (2008) [2] Total 100 460±79
Male 54 452±80
Female 46 468±80

Teitelbaum et al. (2013) [3] Total 240 506±105
Male 113 533±105
Female 127 482±99

Tacchino (2015) [4] Total 443 690±94
Male 101 729±85
Female 342 678±92

Bekheit et al. (2020) [5] Total 606 630±175
Male 226 662±186
Female 380 612±164

Values are presented as mean±standard deviation.

Table 3
Measurement of the peritoneal surface area of healthy person and colon cancer patients
Participants Age (yr) Height (cm) Weight (kg) BMI (kg/m2) Liver (cm2) Gallbladder (cm2) Stomach (cm2) Spleen (cm2) Uterus (cm2) Small bowel (cm2) Transverse colon (cm2) Sigmoid colon (cm2) Parietal peritoneum (cm2) Total (cm2)
Healthy person 866.9 661.4
 M1 31 175 76 24.8 995.4 68.8 138.8 239.3 - 4,499 2,793 10,263
 M2 57 170 69 23.9 949.6 49.6 124.9 381.4 - 2,659 10,192
 M3 67 165 67 24.6 878.2 91.6 86.0 304.1 - 2,795 10,182
 M4 64 179 79 24.7 893.4 54.8 155.9 274.3 - 3,237 10,643
 M5 36 173 69 23.1 1,084 33.2 124.9 236.5 - 2,625 10,131
 F1 51 158 53 21.2 807.1 57.7 70.0 272.6 177.8 4,416 2,743 10,072
 F2 53 160 59 23.0 1,056 52.6 115.9 289.0 140.7 2,271 9,869
 F3 58 156 59 24.2 947.2 66.4 88.6 244.0 111.9 2,651 10,053
 F4 52 163 61 23.0 767.6 36.8 127.8 246.7 104.7 2,627 9,855
 F5 60 166 68 24.7 740.0 33.2 129.3 161.3 133.0 2,673 9,813

Colon cancer 866.9 661.4
 M6 79 161 67 25.8 901.5 97.4 130.4 307.4 - 4,499 2,647 10,112
 M7 62 168 64 22.6 876.4 46.5 131.2 314.6 - 2,613 10,009
 M8 63 162 61 23.2 983.4 31.4 125.8 231.4 - 2,898 10,297
 M9 68 172 66 22.3 961.8 52.3 150.4 227.6 - 2,864 10,284
 M10 79 165 69 25.3 956.1 71.1 146.7 234.9 - 2,695 10,131
 F6 59 158 53 21.2 743.1 66.8 95.4 246.8 166.2 4,416 2,533 9,796
 F7 60 152 62 26.8 824.9 45.1 88.7 239.8 145.8 2,346 9,634
 F8 62 153 59 25.2 954.1 32.9 94.6 202.9 147.3 2,350 9,725
 F9 59 160 63 24.6 927.4 44.2 111.3 142.7 132.1 2,612 9,914
 F10 57 161 64 24.6 940.7 70.6 104.7 178.4 109.8 2,467 9,815

BMI, body mass index; M, male; F, female.

Editorial Office
101-3304 Brownstone Seoul, 464 Cheongpa-ro, Jung-gu, Seoul 04510, Korea
TEL : +82-2-393-2114   FAX : +82-2-393-1649   E-mail : office@kjco.org

Copyright© Korean Society of Surgical Oncology.                Developed in M2PI
About |  Browse Articles |  Current Issue |  For Authors and Reviewers