The three-dimensional reconstruction of the dilated renal pelvicalyceal system by non-enhanced computed tomography

Introduction

Presently, non-enhanced computed tomography (CT) is a standard method of diagnostics of nephroureterolithiasis that provides preoperative visualization of the localization, density, and size of stone [1]. However, its significant drawback is the impossibility of a 3D-visualization of the pelvicalyceal system (PCS). A detailed study of the peculiarities of PCS structure (number and orientation of calyces, length and width of their necks, and an angle ratio of different parts) is required for the choice of optimal surgical tactics. Along with that, this data can be obtained by CT with intravenous contrast that allows for 3D-reconstruction before the surgery. However, this increases radiation exposure on the patient. Besides, a contrast agent is contraindicated in patients with Chronic Kidney Disease (CKD) and hypersensitivity to the drug [2]. The above-mentioned drawbacks require the development of 3D-reconstruction methods of PCS reconstruction based on contactless CT. There are scientific publications that describe methods of 3D reconstruction of renal structures by non-enhanced CT images but they target the visualization of its parenchyma [3, 4]. Only one article was dedicated to the reconstruction of PCS [5]. Besides, the isolation of the borders of the renal cavity system was made manually, which significantly elongated the time of preparation of the data and limited the application of this method into clinical practice.

Thus, the present study aimed to describe a method of semiautomated isolation of PCS in the images of non-enhanced CT with further 3D-reconstruction.

Materials and methods

A total of 5 patients with renal colic were recruited from April to May 2021. Ultrasound investigation was performed and pyelocalycoectasia was revealed. Patients signed a form of informed consent for a contrast CT with Omnipac 300.0 on a 64-slice CT scanner with 0.5 mm slice thickness Somatom Definition AS. The obtained images were studied in the program Medical Imaging Interaction Toolkit (MITK) that was used for 3D-reconstruction of PCS by the images of excretory phase and creation of a standard virtual model. Further, three random points were designated on each slice of the non-enhanced CT phase for the evaluation of the density differences around these points and determination of PCS boundaries without the need in their manual isolation (Fig. 1).

Further, automated mergence of the selected areas was performed. Due to a polygonality of the obtained 3D-constructions, the same algorithm was used for smoothing of the surface (Figs. 2-3).

The volume of the obtained virtual models and the volume of contrast reconstructed models were compared. Five urologists evaluated the informative value of such reconstructions for the investigation of the PCS anatomy of a certain patient. The obtained models were also analyzed by IT specialists for suitability for 3D-printing.

Statistical analysis. The statistical analysis was made in the IBM SPSS Statistics 22.0 software (SPSS Inc., Chicago, IL, USA). The evaluation of nominal data was made with a chi-square test. The difference was significant at p < 0.05.

Results

In all cases, the duration of PCS boundaries determination and its reconstruction with smoothing was less than 10 minutes. The mean surface area of contrast and non-enhanced 3D models was 3291 mm² and 2879 mm², respectively (p = 0.12). The comparison of contrast and non-enhanced 3D models as well as their evaluation and feasibility for preoperative planning scored 4.5 out of 5.0 by urologists. A discussion with engineers on the adequacy of the models for 3D-printing showed that all PCS parts were sufficiently visualized and did not require correction. Potentially weak parts of the printed-out models, in particular, minor calyx necks, did not require the formation of additional supports before printing. Thus, the models were completely suitable for potential 3D-printing.

Discussion

Non-enhanced CT provided a urologist with reliable information on the main parameters of ureteral stones that include size, site, and density. However, it is impossible to study in detail the renal cavity system, which requires additional diagnostic procedures for reliable planning of possible surgery. Up to now, 3D-reconstruction was possible only after contrast CT that has an excretory phase in 3-5 minutes. The high density of the contrast is comparable with the density of bone structures. And the filling of PCS provides a 3D reconstruction via an automated overlay of CT images. As a result, a urologist gets a possibility to evaluate all structural peculiarities of the renal cavity of a certain patient, in particular, the number and orientation of the minor calyces, length, and width of the necks, as well as the angle ratio between different parts of PCS. In turn, this information allows a specialist to choose proper tactics of the surgery for nephrolithiasis increasing the efficiency of the intervention and decreasing the rate of associated complications [6].

The application of a contrast agent has such disadvantages as an increase in radiative exposure and load on a patient’s organism and a dependence on the patient’s state of health, in particular, impairments of renal excretory function. This makes the development of 3D methods of reconstruction by non-enhanced images a relevant task.

Automated reconstruction is possible for structures with a relatively constant shape without branching, like heart chambers [7], aorta [8], intracerebral hemorrhage [9], and kidney parenchyma [10]. This approach is not feasible for the renal cavity because of its variable and branchy structure, small differences in the density of the surrounding structures (vessels, adipose tissue), and collapsed shape of the renal cavity in the norm. The last aspect prevents the development of available solutions for previous limiting factors. The authors’ point of view is shared only in one publication (Sung et al.) on the description of 3D-reconstruction of PCS by the images of non-enhanced CT [5]. Sung et al. proposed a protocol of intravenous infusion with diuretic load for artificial dilation of PCS before CT scanning. This approach showed to result in a significant volume of the renal cavity system reconstruction in comparison with the reconstruction of images made in the control group. The determination of PCS boundaries was performed manually by an IT specialist, which took more than 20 minutes and required sufficient experience. This limited the availability of this approach for routine application in medical institutions.

The present work included patients with renal colic that underwent CT scanning during the episode of obstruction that lead to a dilation sufficient for the visualization of the collecting system of a kidney. The determination of the boundaries was performed in a semi-automated mode. Several random points were designated within the boundaries of PCS in each non-enhanced CT image for the determination of its boundaries. After the mergence of the images, a 3D model was built. Further, the described algorithm automatically determined PCS boundaries significantly reducing the time of determination (to 5 minutes). This method did not require specialized experience and the procedure of segmenting could be performed by a urologist.

It should be mentioned that there were limitations in the present study. Firstly, the analysis included CT scans of five patients, which did not allow for the approbation of the described algorithm in the reconstruction of all variations of PCS. Secondly, the selected patients had nephrolithiasis complicated by the obstruction of the upper urinary tracts with dilation of PCS. Thus, this approach proved feasible in this cohort of patients. And thirdly, the reconstruction was performed in a semi-automated mode, which also required the involvement of the specialist in its realization. Finally, the described algorithm was not tested in patients with the presence of a stone in the PCS, which could negatively affect the precision of the automated determination of the boundaries. Further improvement of this algorithm will target a complete automated performance that will select the borders of PCS in all CT slices after the designation of one point.

Conclusion

A semi-automated reconstruction of the PCS by non-enhanced CT scans provides its possible visualization without enhancement in patients with an obstruction of the upper urinary tract.