Abstract
Background/Aim: We investigated pelvic arterial deformation and shift due to intraoperative pneumoperitoneum and postural changes in an animal model. Materials and Methods: Computed tomography images of pigs were acquired in different body positions (supine, head down at 5° and 10°, right lateral recumbent at 5° and 15°) before and after insufflation. We used a free software (3D Slicer) for image analysis. After landmark registration using 10 markers inserted into the pelvis, pelvic arterial deformation and shift of seven arterial bifurcation points were evaluated. The distance moved was the target registration error (TRE) from the points registered in the supine position. Fiducial registration error (FRE) was measured using the 10 pelvic markers. Results: TRE average from postural changes ranged from 0.7 to 1.2 mm and was 1.4 mm due to pneumoperitoneum. TRE and FRE averages were 2.1 mm and 0.2 mm, respectively. Conclusion: The pelvis was useful for registering anatomical landmarks.
- Pelvic artery deformation
- navigation
- postural changes
- pneumoperitoneum
Rectal cancer is associated with a worse prognosis than colon cancer because of its high rate of local recurrence (1). Rectal cancer has a high rate of local recurrence, commonly presenting with lateral pelvic lymph node metastasis. The rates of lateral pelvic lymph node metastasis vary from 10% to 15% (2-4), and they increase with the depth of tumor invasion. A total of 16.5% of patients with T3 tumors and 37.2% of patients with T4 tumors had lateral pelvic lymph node metastasis (5). In Japan, the standard treatment procedure for advanced lower rectal cancer is a total mesorectal excision (TME) and a lateral pelvic lymph node dissection (LPLD) for pelvic lymph node metastasis. The procedure is generally performed with minimally invasive surgery, including laparoscopic and robotic surgery, because of the advantages of decreased postoperative pain, reduced hospitalization, and improved long-term outcome (6). However, the affected lymph nodes are located alongside the pelvic arteries and autonomic nerves; thus, LPLD carries the risk of blood loss and urinary and sexual dysfunction (7). Therefore, we aimed to develop a real-time navigation system to help surgeons perform this procedure more safely and efficiently by comparing the intraoperative images to the preoperative images, such as computed tomography (CT) or magnetic resonance images and displaying the pelvic arteries on the monitor.
Surgical navigation systems were first reported in neurosurgery (8). Surgical navigation systems have become widely prevalent in neurosurgery, as well as in orthopedic, otolaryngologic, and liver surgeries (9-11). For rectal cancer, Atallah et al. (12) have assessed transanal TME conducted with navigation, and Nijkamp et al. (13) have assessed open pelvic surgery with navigation. However, real-time navigation systems have not yet become widely available for abdominal laparoscopic colorectal surgery. Surgical manipulation such as pneumoperitoneum and postural changes have a considerable effect on the abdominal wall and intra-abdominal organs and introduce movement and deformation, which can affect the accuracy of the navigation systems. However, we thought that deformation of the pelvic arteries due to surgical manipulations would be relatively small, because the pelvic arteries are located against the pelvis.
Nimsky et al. (14) have reported an intraoperative brain shift during neurosurgery. Ohya et al. (15) reported carotid artery deformation caused by head and neck postural changes during otolaryngologic surgery. Vijayan et al. (16) and Zijlmans et al. (17) examined liver deformation due to pneumoperitoneum during liver surgery. These studies used pigs as an animal model. However, to our knowledge, no previous study has examined deformation of the pelvic arteries due to surgical manipulation. The aim of this study was to present a landmark registration method using the pelvis in an animal model, and to provide the first result of deformation of the pelvic arteries caused by pneumoperitoneum and postural changes.
Materials and Methods
Animals and anesthesia. Two pigs (Landrace), with a weight of 36 kg, were used in this study. The pigs were born at the Ibaraki farm (Ibaraki, Japan) approximately 3 months before the study, and housed for 10 days after a 7-day quarantine period. Both pigs were females, because it was easier to place a urethral catheter into the bladder. The urethra of male pigs is long and tortuous, and it is difficult to insert a catheter through it. The pigs fasted one day before the experiment.
The pigs were premedicated with atropine sulfate at a dose of 0.02 mg/kg, midazolam at 0.3 mg/kg, and medetomidine at 0.06 mg/kg 1 h before the start of the experiment. The pigs underwent tracheal intubation after deep anesthesia was induced by sevoflurane. The pigs were then attached to a ventilator with a respiration rate of 12 breaths per min and a tidal volume of approximately 400 ml. All pigs underwent placement of a urethral catheter. Sevoflurane and midazolam were used for the maintenance of general anesthesia. Pulse oximetry, electrocardiography, and non-invasive blood pressure were monitored during the experiment.
This study was approved by the President of Tokyo University after a review by the Institutional Animal Care and Use Committee (Permission number: KA15-3) and conducted according to the Tokyo University Animal Experimentation Regulations. This study complied with the ARRIVE guidelines. The study also adhered to the Guiding Principles in the Care and Use of Animals approved by the Council of the American Physiological Society
Surgical procedure and imaging. We made an incision in the abdominal wall above the umbilicus. A trocar was then placed through this site to create the pneumoperitoneum. Ten markers were inserted around the pelvis (both sides of the anterior superior iliac spine, the bilateral anterior and posterior surface of the ilium, both sides of the pubis, and the upper and lower sacrum) using a sliding needle under fluoroscopic guidance, for use as registration markers (Figure 1). The markers were 1-mm steel balls. Figure 2 shows an overview of the experimental set-up. During image acquisition, the pigs were strapped into a Styrofoam platform to keep them stationary.
We used a CT scanner (Somatom Sensation 16®; Siemens Medical Solutions, Erlangen, Germany) to obtain the CT images. On the examining table, the pigs were positioned in the supine position, with their heads facing down (5° and 10°), and to the right lateral recumbent position (RLR) (5° and 15°) by slanting the Styrofoam platform. The slant angle was measured at the same position on the Styrofoam platform using a digital angle gauge. The angle varied within±0.1° during image acquisition. After acquiring the CT images according to the postural changes, we created the pneumoperitoneum, and CT images were acquired according to the same postural changes (Figure 3). The target intra-abdominal pressure was set to 10 mmHg. The pressure varied within±1 mmHg during image acquisition. The ventilator flow was terminated at the end of expiration during the CT scanning to acquire the images without the movement caused by respiration. Lohexol, a non-ionic contrast medium, was injected to acquire the arterial phase image. CT image data were acquired as Digital Imaging and Communication in Medicine (DICOM) data. The image resolution was 512×512 pixels, and each pixel was 0.586×0.586 mm2. The CT slice thickness was 0.7 mm.
Imaging analysis. The 3D Slicer, a free and open source software, was used for visualization and registration of the landmarks (18). Figure 4 shows the workflow of the registration of the landmarks and evaluation that was conducted based on 3D Slicer. First, the CT images, in the DICOM format, were imported into 3D Slicer via the `DICOM` module.
Registration. Image registration was used for the alignment of two images of different types (supine, the postural change conditions, and the pneumoperitoneum) from the same animal. Using the ‘landmark registration’ module in 3D Slicer, 10 markers around the pelvis, which were clearly identified on the CT images, were selected and registered as the landmarks.
Assessment of the deformation of the pelvic arteries. After landmark registration, we calculated the fiducial registration error (FRE) for each marker. Coordinates (X, Y, Z) were obtained for the 10 markers used as landmarks in pre-and post-insufflation and for postural changes. The FRE was determined using the Euclidean distance formula:
FRE is the vector length between the marker positions that are used as the landmarks in pre-and post-insufflation, and in pre-and post-postural change scans; (Xpre, Ypre, Zpre) are the CT volume positions of the markers in the supine position before insufflation; and (Xpost, Ypost, Zpost) are the CT volume positions of the markers for postural changes or after insufflation. Both volumes are located in the same coordinated system in physical space.
We also assessed the target registration error (TRE) for each of the seven arterial bifurcations. Figure 5 shows seven bifurcations on the three-dimensional blood vessel image that was edited using a CT workstation (Ziostation, Ziosoft Inc., Tokyo, Japan). The coordinates (x, y, z) were obtained for each of the seven arterial bifurcations both before and after insufflation and for the postural changes. The TRE was determined using the Euclidean distance formula:
TRE is the vector length between the arterial bifurcation position before and after insufflation and before and after postural change scans; (xpre, ypre, zpre) are the CT volume positions of the arterial bifurcations in the supine position before insufflation; and (xpost, ypost, zpost) are the CT volume positions of the arterial bifurcations for postural changes or after insufflation. Both volumes are located in the same coordinated system in physical space.
Statistical analysis to examine the FRE associated with the 10 landmarks and TRE associated with the seven bifurcations was performed. All data are reported as mean±standard deviation.
Results
The average and standard deviation of the FRE in the supine position, the head down (HD) at 5°, the HD at 10°, the RLR at 5°, and the RLR at 15° were examined before and after insufflation. The average FRE was 0.2 mm. The FRE values ranged from 0.1 to 0.3 mm.
Tables I and II show the main findings of this study. Table I includes averages and standard deviations of the TRE in the supine, the HD at 5°, the HD at 10°, the RLR at 5°, and the RLR at 15° under pre- and post-insufflation. The TRE values before insufflation ranged from 0.7 to 1.2 mm, and those after insufflation ranged from 1.4 to 2.1 mm. The largest TRE measurement was in the RLR at the 15° position under insufflation.
Table II includes the TRE of each bifurcation in the HD and RLR positions, along with postural changes under insufflation. The range of the TRE of each bifurcation in the HD and RLR under pre-insufflation was approximately 1 mm. However, in postural changes under post-insufflation, the TRE of bifurcations ④ and ⑤ (external iliac arteries) was 2.7 mm and 2.6 mm, respectively.
Discussion
We presented the evaluations of pelvic artery deformations due to pneumoperitoneum and postural changes in an animal model. We used 10 markers inserted around the pelvis for registration. Regarding laparoscopic surgery, the skin and most of the intra-abdominal organs were largely affected by postural change and pneumoperitoneum. Hayashi et al. (19) reported a real-time navigation system for laparoscopic gastrectomy using skin-fixed markers. They reported an FRE of 14.0 mm. In addition, it is difficult to locate a permanent attachment of the skin-affixed markers at the same site, yet it is acceptable for patient comfort and cosmetic reasons, because CT is usually performed several weeks before surgery. This study revealed an FRE average of only 0.2 mm, which remained mostly unaffected by pneumoperitoneum and postural changes. Thus, our study suggests that the pelvis provides a useful registration as a landmark. The major problem associated with using the pelvis as a landmark is that it is difficult to insert markers around the pelvis in humans. Pandey et al. (20) reported a method that used 3D ultrasound with fast and automatic segmentation and included registration of the pelvis CT, which might be used clinically. Further, the study reported a TRE of 2.4 mm that was achieved with a mean runtime of 27.3 s. This procedure can be rapidly and accurately performed inter-operatively for pelvic registration. In addition, the ultrasound procedure is useful as a less invasive approach for patients.
Therefore, 3D Slicer has been used in various studies in the computer-aided surgery community, and the findings support the development of extension modules. For example, Ungi et al. (21) developed a free, open source surgical navigation software module ‘SlicerIGT’ to improve the success rate and time efficiency of facet joint injections.
For accuracy of the real-time navigation, it is important to demonstrate that the pelvic arteries as targets are slightly displaced by postural changes and pneumoperitoneum. Our results revealed that TRE values ranged from 0.7 to 1.2 mm due to postural changes and that the TRE value was 1.4 mm due to pneumoperitoneum. At maximum, the TRE value was 2.1 mm in the RLR at 15° after insufflation. As the angle of postural change increased, the TRE value also increased. However, in most clinical cases, the angles of the HD and RLR positions are no larger than 15°, because the lung is compressed, and respiratory function is depressed. With respect to each bifurcation, there was no difference in postural changes under pre-insufflation. However, the TRE value for the bilateral bifurcations in the external iliac arteries was 2.7 mm. These findings suggest that the external arteries are slightly affected by pneumoperitoneum. Previous reports found carotid artery deformation and brain shift due to postural change and liver deformation due to pneumoperitoneum (12-15). This study showed that pelvic arterial deformation was significantly reduced compared to the deformations reported in the other studies. We speculate that this is because the pelvic arteries are surrounded by a large bone and remain anchored to the retroperitoneum. Surgeons commented that real-time navigation systems would be useful to perform laparoscopic/robotic surgery if the TREs would be <5 mm. Pelvic arterial deformation due to postural change and pneumoperitoneum meets this requirement. For this reason, the pelvic arteries are a possible target for real-time navigation systems. As part of a new era of laparoscopic and robotic surgery, the images of the pelvic arteries could be projected onto an intraoperative screen during LPLD for rectal cancer.
However, we used an animal model. Human anatomy differs slightly from the pig in the size and configuration of the pelvis and the course of the pelvic arteries. The pelvis of a human that walks on two legs is more free-ranging and the pelvic floor muscles are thicker than those of a pig that walks on four legs. Therefore, we consider that in humans, the pelvic arterial deformations are smaller than those in pigs. In the future, human clinical studies will be needed.
Conclusion
In this study, we presented a new model to evaluate the deformation of the pelvic arteries due to postural change and pneumoperitoneum, and a method to register landmarks of the pelvis using 3D Slicer. This is the first study to demonstrate that the deformation of the pelvic arteries caused by pneumoperitoneum and postural changes was ≤2.1 mm. The pelvis was useful to register landmarks. In a real-time navigation system for LPLD, this finding contributes considerably to registration between intraoperative and preoperative images. Future work including human clinical studies is warranted to evaluate deformation of the pelvic arteries.
Acknowledgements
The Authors would like to thank Kazuhiro Endo and Syuuji Hishikawa at the Centre for Development of Advanced Medical Technology, Jichi Medical University, Japan for supporting this animal study.
Footnotes
Authors’ Contributions
Concept: Hidemichi Kiyomatsu; Design: Hidemichi Kiyomatsu; Supervision: Etsuko Kobayashi, Ichiro Sakuma, Souichiro Ishihara; Data collection and/or processing: Hidemichi Kiyomatsu, Lei Ma; Analysis and/or interpretation: Hidemichi Kiyomatsu; Literature review: Hidemichi Kiyomatsu, Etsuko Kobayashi, Ichiro Sakuma; Writing: Hidemichi Kiyomatsu; Critical review: Hidemichi Kiyomatsu, Junchen Wang, Tomomichi Kiyomatsu, Hiroyuki Tsukihara.
This article is freely accessible online.
Funding
JSPS KAKENHI Grant Number 26108008, JSPS-NSFC Joint Research Project JPJSBP 120197410, and JST COI Grant Number JPMJCEJPMJCE1304
Conflicts of Interest
The Authors have no conflicts of interest to declare regarding this study.
- Received October 30, 2020.
- Revision received November 23, 2020.
- Accepted November 24, 2020.
- Copyright© 2021, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved