If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Correspondence to: Department of Radiology, Kanazawa University Graduate School of Medical Sciences, 13-1, Takaramachi, Kanazawa, Ishikawa 920-8641, Japan.
To evaluate the feasibility of renal artery-based segmentation of kidneys with renal cell carcinoma (RCC), based on three-dimensional (3D) software for the simulation of segmental artery clamping (SAC), and to correlate it with RENAL nephrometry score.
Methods
Fifty RCCs (< 4 cm) identified from a pathological database search between January 2015 and January 2018 were included retrospectively. On computed tomography (CT) images, the relevant kidney, tumor, and renal artery were annotated semi-automatically on the commercial workstation, and renal artery-based segmentation was performed using 3D Voronoi diagrams. Simulation of SAC was performed by a radiologist and urologist in consensus. The volume of the whole kidney and tumor and estimated rescued volume for possible SAC cases were calculated. The correlation between possible SAC and RENAL nephrometry score was investigated. The reproducibility of the calculation of each volume and the interrater reliability of SAC simulation were assessed.
Results
In the anatomical analysis, 44 patients had a single main renal artery and six had two main renal arteries, and of these, an early division pattern was observed in 11 cases. In the 3D simulation software, 22 out of 50 cases (44 %) were determined as possible SAC. The agreement of the SAC simulation was excellent (kappa = 0.96). RENAL nephrometry score was significantly different in the anterior/posterior and exophytic/endophytic components between possible and impossible SAC groups.
Conclusions
Renal artery-based segmentation of kidneys with RCC on CT images using 3D simulation software is feasible for effectively estimating the possibility of SAC with high reproducibility.
]. Elective partial nephrectomy is equivalent to radical nephrectomy in patients with clinical T1 RCC based on retrospective, comparative, and multi-institutional studies [
]. Regarding the procedure of partial nephrectomy, intermittent clamping of the main renal artery is a classical procedure used to prevent blood loss from the surgical site during tumor resection. However, warm ischemia of the kidney, caused by main renal artery clamping during partial nephrectomy could result in postoperative renal dysfunction of the affected kidney [
]. Segmental artery clamping (SAC), in which the selected renal arterial branch feeding the territory, including the target renal tumor, is clamped selectively, is one of the modified surgical approaches of partial nephrectomy for avoiding unnecessary warm ischemic injury of the normal parenchymal area [
]. The indication of SAC is controversial and generally based on the surgeon’s judgment preoperatively or on-site during the operation, and depends on the anatomical variation, tumor location, and the surgeon’s experience [
], which is essential to improve surgical planning that could avoid on-site determination of the surgical procedure.
Recent three-dimensional (3D) simulation software based on preoperative multi-phasic computed tomography (CT) can visualize not only the tumor but also the renal anatomy, including the renal parenchyma, renal veins, and urinary collecting system, and the relationship among these anatomical components [
]. Regarding 3D object segmentation by pin or line, Voronoi’s algorithm is a popular mathematical algorithm that can divide 3D space between predetermined points or lines based on the shortest distance to those points or lines [
]. When the objects and lines are replaced with organs and vessels, this method can be applied to medical images and enables the segmentation of organs based on arteries or other vessels by points/lines. It is also suitable for the estimation of the blood supplied area or responsible area from multiple vessels in the organ. This method has the potential to build the preoperative segmentation and simulation of the RCC operation; however, there is no clear report discussing the utility of the renal artery-based segmentation of the kidneys using Voronoi’s algorithm. In this study, we evaluated the utility of renal artery-based segmentation of kidneys with RCC using 3D simulation software based on preoperative arterial phase CT, investigated the possibility of SAC, and correlated this with the RENAL nephrometry score for the prediction of SAC.
2. Materials and methods
2.1 Patient selection
This single-center retrospective study was approved by the relevant institutional review board, and informed consent was waived. A total of 148 consecutive cases of RCCs, diagnosed by operation or biopsy, were selected from a pathological database search between January 2015 and January 2018. A total of 97 RCCs were diagnosed as pathological T1a. Of these 97 RCCs, 47 were excluded due to unilateral multiple RCC (five cases), dialysis related RCC (two cases), cystic RCC (four cases), multiple renal cysts (four cases), and absence of preoperative multi-phasic multidetector CT obtained by an appropriate protocol for the simulation in this study (32 cases). Finally, 50 RCCs were included in this study. Sex, age, laterality of RCC, method of diagnosis (radical nephrectomy, partial nephrectomy, or tumor biopsy), and tumor diameter were collected from medical records.
2.2 CT scan protocol
Contrast-enhanced multi-phasic CT images were obtained with one of three CT scanners in our hospital (Two Light Speed VCT64; GE Healthcare, Waukesha, WI, USA or one Somatom Definition Flash; Siemens, Erlangen, Germany). A 100–150 mL dose of contrast medium with 600 mg of iodine per kilogram was injected for 30–35 s after pre-contrast scanning. Arterial phase images were acquired 17 s after the detection of contrast medium by bolus tracking at the level of the thoracoabdominal aorta. On the arterial phase, 1.25 mm or 0.625 mm DICOM image data were finally collected and registered into the analysis on the workstation.
2.3 Kidney analysis process using 3D simulation software
Image analysis was performed using SYNAPSE® VINCENT (Fujifilm Medical Co., Ltd, Tokyo, Japan). A flow chart of the kidney segmentation process is shown in Fig. 1. This simulation software was implemented as a plug-in in the processing workstation (Dell Precision T5400; Dell Computer, Round Rock, TX, USA) running Windows 64-bit (Microsoft Corporation, Redmond, WA, USA). Acquired DICOM patient data were reconstructed via the following three steps.
Fig. 1Process of virtual nephrectomy using 3D simulation software. After loading the CT data, virtual nephrectomy images are created via the following three steps. In Step 1, the renal parenchyma of the affected kidney is extracted and the renal cortex is identified automatically. In Step 2, renal arteries and tumors are extracted, in which an additional setting of seeding points is necessary to extend the peripheral renal branches. In Step 3, after merging of the extraction data, the vascular perfusion territory of each selected target branch is calculated and visualized by transparent color with the tumor and arteries.
The renal parenchyma, including the RCC, was semi-automatically extracted from consecutive CT images. A rough 3D image of the kidney segmentation was created using a shape recognition algorithm. Tumors and small cysts were included together with the renal parenchyma. The renal contour was carefully verified, and the retrieval of segmentation was performed with touchup application if there was over- or under-segmentation. Cysts > 1 cm were manually removed in this step. After extraction of the renal parenchyma, the cortex was automatically identified and extracted.
2.3.2 Step 2: 3D reconstruction of the renal artery and tumor
By setting the start point and direction, the automatic algorithm selected consecutive voxel data with appropriate CT values and branching angles, which were developed by a modified threshold-based region-growing technique. The stem of the main renal artery near the aorta was set as a seeding point and the direction was set for the peripheral side; the arterial tree was then extracted automatically. The additional setting of seeding points is usually necessary to extend the peripheral thin branches for complete segmentation. The renal arteries were carefully verified, especially for erroneous selection of renal veins, and the correction of arterial reconstruction was performed. The tumor was designated mainly in the axial images with the region growing method from seeds semi-automatically. When extra renal arteries were identified, these arteries were also reconstructed. Finally, the extracted renal parenchyma, arteries, and tumors overlapped, and 3D images were created.
2.3.3 Step 3: virtual nephrectomy
The vascular perfusion territory was calculated using an algorithm based on Voronoi tessellation. By selecting the target branch, the territory belonging to any selected artery is bordered by a line that runs at an equal distance from the surrounding arteries. Each territory was visualized by a transparent color on the tumor and arteries. A virtual nephrectomy for the SAC was performed by extracting the perfusion territory of the target renal artery followed by its subtraction from the whole kidney. The total renal volume and tumor volume were calculated, and if the case was determined as possible SAC, and described subsequently, the estimated rescued renal volume was also calculated.
2.4 Evaluation of renal artery variation and RENAL nephrometry score
The anatomical variation of the renal artery was determined by two radiologists in consensus using CT images, aside from the kidney analysis using 3D simulation software. All cases were categorized into the following six types based on arterial anatomy (Fig. 2). Type 1 was defined as a single renal artery without an early division branch, Type 2 a single renal artery with an early division branch, and Type 3 as double renal arteries. The early division branch was defined as a replaced artery branching between the main stem and renal hilum. Moreover, the extra renal artery was defined as the accessory branch arising from the root of the main stem or aorta supplying the apical or lower segment. If an extra renal artery was present, the corresponding cases were categorized as Type 1a, 2a, and 3a. For the RENAL nephrometry score assessment, one urologist scored each RCC retrospectively using the relevant CT images.
Fig. 2Anatomical variation of the renal artery. Type 1: single renal artery without early division branch and extra renal artery. Type 1a: single renal artery with an extra renal artery and without an early division branch. Type 2: single renal artery with an early division branch and without an extra renal artery. Type 2a: single renal artery with an early division branch and an extra renal artery. Type 3: double renal arteries without an extra renal artery. Type 3a: double renal arteries with an extra renal artery.
Evaluation of SAC was performed independently from the evaluation of renal artery variation and the RENAL nephrometry score. Possible SAC was defined if the rescued artery existed in any one of the anterior, posterior, or extra renal artery, which did not supply the tumor with a 5 mm margin on the virtual nephrectomy. If there was no rescued artery in any of the branches, the case was judged as impossible SAC (Fig. 3 & 4). One urologist determined the possibility of SAC in each case using virtual nephrectomy images. In the cases judged as possible SAC, rescued renal volume and rescued renal ratio in the affected kidney were calculated using the simulation software. The distribution of possible SAC and rescue renal calculation in the anatomical six types were evaluated. To estimate the reproducibility of virtual nephrectomy creation, another radiologist performed the process of kidney analysis using 3D simulation software in identical cases.
Fig. 3A 61-year-old male with renal cell carcinoma determined as possible for segmental arterial clamping. The single renal artery with an extra renal artery supplying the apical area is observed. The tumor is located in the middle and posterior. The RENAL score is 7 (1 +2 +2 +p + 2). The tumor is estimated to be supplied from the anterior artery only, and selective clamping of the anterior artery is possible for transient ischemia of the tumor (white bar = clamping site).
Fig. 4A 50-year-old male with renal cell carcinoma determined as impossible for segmental arterial clamping. A single renal artery with an extra renal artery supplying the apical area is observed. The tumor is located in the upper pole and non-exophytic component. The RENAL score is 5 (1 +1 +1 +a+2). The tumor is estimated to be supplied from both the main renal artery and the extra-renal artery, which is judged as impossible segmental artery clamping.
Statistical analysis was performed using R software, version 4.0.3 (http://R-project.org). Comparison of the possibility of SAC between anatomical types was compared using the Chi-square test. Categorical variables of the RENAL nephrometry score were compared using Fisher’s exact test between possible and impossible SAC groups. The inter-procedure agreement of the evaluation of SAC between the two radiologists was assessed using the kappa statistic. Interrater reliability of the calculation of whole renal volume, tumor volume, and estimated rescued renal volume was evaluated using the intraclass correlation coefficient (ICC). Statistical significance was accepted when p < 0.05.
3. Results
The demographic and clinical features of the patients are shown in Table 1. All patients had single RCC and 70 % of RCC arose from the right kidney. The mean tumor size on the pathology was 2.3 cm. The anatomical characteristics of the renal artery are shown in Table 2. Forty-four patients had a single main renal artery (86.6 %) categorized into Type 1/1a/2/2a, and six had two main renal arteries, which were categorized as Type 3/3a. An early division pattern was observed in 11 cases, categorized as Type 2 or 2a. Thirty-five patients had extrarenal arteries categorized as Type 1a/2a/3a. Of these kidneys with extra renal arteries, 17 had apical branches, six had lower branches, and 12 had both apical and lower branches.
In the 3D simulation software analysis, 22 out of 50 cases (44 %) were judged as possible SAC (Table 3). In the subgroup analysis, 13 of 33 Type 1/1a cases (39 %), five of 11 Type 2/2a (45 %), and four of six Type 3/3a (67 %) were judged as possible SAC. Of the cases with an extra renal artery, 19 of 35 (54 %) in Type 1a/2a/3a were SAC possible, which was significantly different compared with the cases without an extra renal artery (Type 1/2/3) (54 % vs. 20 %, p < 0.05). The inter-procedure agreement of the SAC evaluation was excellent (kappa = 0.96). In one case of Type 2, the judgment was divided due to the difference in peripheral branch tracking. Interrater reliability of the calculation of renal volume, tumor volume, and estimated rescued renal volume was excellent (ICC [
] = 0.99, 0.98, and 0.90, respectively). The RENAL nephrometry score was significantly different in the anterior/posterior and the exophytic/endophytic components between possible and impossible SAC groups (p < 0.05 for both) (Table 4).
Table 3Possibility of SAC.
Anatomy type
n
SAC possible
Rescued renal volume (cm3, mean ± SD)
Rescued renal ratio (%, mean ± SD)
Overall
50
22 (44 %)
54 ± 35
36 ± 22
1
1
12
2 (17 %)
55 ± 17
32 ± 15
1a
21
11 (52 %)
52 ± 35
35 ± 20
2
2
2
1 (50 %)
119
68
2a
9
4 (44 %)
58 ± 44
39 ± 32
3
3
1
0 (0 %)
–
–
3a
5
4 (80 %)
37 ± 29
27 ± 19
SD, standard deviation; SAC, segmental artery clamping.
In this study, we assessed the utility of renal artery-based segmentation of kidneys with RCC on CT, using 3D simulation software to predict the possibility of SAC. We found that SAC for RCC was possible in 22 out of 50 cases with high reproducibility. Some preoperative planning or simulation methods for SAC have been proposed, but this is the first report, to the best of our knowledge, discussing the utility of renal artery-based segmentation of the kidney using Voronoi’s algorithm [
]. Whereas the Voronoi algorithm is well known on two-dimensional images, recent software can apply 3D geometrical computations onto 3D images. This 3D technique enables the creation of a segmentation plane on any organ by vessels or other landmarks on CT and magnetic resonance imaging, and this Voronoi diagram method has been used in the segmentation of the liver and heart as well [
]. We applied this algorithm to the kidney for segmentation of renal parenchyma based on renal artery perfusion using a commercial workstation.
In the 3D workstation used in this study, most segmentation processes comprise automated steps, including renal parenchymal extraction, renal tumor extraction, renal artery extraction, and virtual nephrectomy. Although, manual reteaches or creations of vessels are required in some parts, it takes 20–30 min per case to complete the analysis because of CT data retrieval, which is feasible for routine preoperative simulation.
This virtual nephrectomy simulation is expected to be useful for preoperative image creation and surgical planning, especially for SAC. Under these circumstances, the indication of SAC on partial nephrectomy has been mostly based on the surgeon’s subjective assessment using anatomical information acquired from CT images. The most important information for the operation is the anatomical position of the tumor in the kidney, arterial anatomy, and perfusion of each arterial branch; however, the assessment of renal perfusion of each branch and its relationship with the tumor is difficult and challenging. Some reports describe the assessment of the perfusion area of the referral artery using near-infrared fluorescence imaging in the operation [
]. However, this method is limited by the unavailability of preoperative planning. In contrast, the virtual nephrectomy simulation used in this study enables objective preoperative simulation.
Our simulation is also applicable for the simulation of vascular interventional procedures for the kidney. Arterial infusion of iodized oil is the procedure used to identify the tumor center for CT-guided percutaneous cryoablation [
]. This preoperative marking was performed before percutaneous cryoablation. Virtual nephrectomy imaging could also predict the distribution of embolized materials on each arterial branch.
The RENAL nephrometry score was developed to standardize the assessment of anatomical features of renal tumors and is used to assign the selection of performing radical nephrectomy or partial nephrectomy [
]. In our study, the total RENAL score did not significantly differ between the possible and impossible SAC groups but did show significant differences in the anterior/posterior and the exophytic/endophytic components between groups. However, the RENAL score is not sufficient to predict the possibility of SAC.
This study has several limitations. First, this analysis required thin-slice original images of at least 2.0 mm or less, and cases without thin slice data were not available. Second, cases of multiple cystic kidney disease or dialysis-related kidneys were excluded because of the annotation difficulty for the renal contour or renal artery. It is difficult to apply our simulation method for these cases, even if partial nephrectomy with SAC is rare among them. Third, we did not perform a correlation analysis between our simulation results and the operation results and did not correlate with ground truth data based on other cases because SAC was performed for only three cases in this study.
5. Conclusion
Renal artery-based segmentation of kidneys with RCC was assessed, and 44 % of cases were determined as possible for SAC using 3D simulation software. Renal artery-based segmentation of kidneys with RCC can be useful for effectively estimating the possibility of SAC with high reproducibility. Further prospective studies including correlation analysis between virtual nephrectomy simulation and intraoperative assessment of selective clamping with near-infrared fluorescence are needed to confirm the clinical utility of this method.
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
00070-3&id=gr1.jpg){kind=link}
00070-3&id=gr2.jpg){kind=link}
00070-3&id=gr3.jpg){kind=link}
00070-3&id=gr4.jpg){kind=link}