To the best of our knowledge, no studies have investigated whether 3D kidney models allow for a more accurate assessment of the R.E.N.A.L. nephrometry score among urology residents. Our study had two main findings. First, the rate of overall R.E.N.A.L. nephrometry score consistency between residents and the expert improved with the use of the 3D kidney models. Second, the “E” score significantly improved in each category.
This study showed that using the 3D kidney models the R.E.N.A.L. nephrometry scores determined by residents were similar to those determined by the expert. Although the rate of R.E.N.A.L. nephrometry score consistency was reportedly low in physicians with lesser clinical experience [10], we observed that the use of the 3D kidney models resulted in an increase in the rate of consistency among residents. The effectiveness of 3D model for residents has been reported in various departments [11, 12]; however, this study found that it is also useful for renal cell carcinoma.
A previous study reported that when students used 3D kidney models, the rate of R.E.N.A.L. nephrometry score consistency between students and physicians improved [13]. Those cases were first assessed by CT, followed by a re-assessment using 3D kidney models; thus, assessing the same cases twice was a drawback. Our study differed from the aforementioned study because assessments of different cases were separately performed using CT only and CT and 3D models. As frequent assessments improve the accuracy of the R.E.N.A.L. nephrometry score for residents, we divided the cases into one group where the R.E.N.A.L. nephrometry score was first determined by CT only followed by CT and 3D kidney models, and another group where the R.E.N.A.L. nephrometry score was first determined by CT and 3D kidney models followed by CT only. This resulted in a study design in which a learning curve was taken into account; thus, we assumed that this would be helpful in reducing bias.
Based on the study results, we concluded that in professionals with few years of experience, the level of understanding of anatomy and ability to read CT images could be lower as compared with senior physicians; thus, 3D kidney models can be useful in improving the level of understanding of anatomy and can play important roles in education for residents.
Furthermore, our study showed that in the R.E.N.A.L. nephrometry score determined by residents, only the “E” score of each item significantly improved by using 3D kidney models. A higher rate of consistency in the C index than the R.E.N.A.L. nephrometry score and PADUA score between individuals was reported previously [10]. Measuring only the direct distance is easier and may reduce errors. In contrast, “E” requires space perception; thus, a 3D model, which is easy to understand in terms of spatial recognition, would be useful, particularly in less-experienced doctors.
Conversely, some studies reported a high rate of consistency of R.E.N.A.L. nephrometry scoring [14–17]. One of the reasons may be variation in the cases. In a study evaluating the rate of consistency between radiologists, though the rate of consistency was high (0.88), more than 50% of the cases were of high complexity [14].
In a study showing low rates of consistency, there were a few tumor cases with high complexity, and most cases had low or moderate complexity [4]. In our cases, most tumors were cT1a (95%) and only a few cases (6%). had high complexity. Accordingly, the difficulty in achieving consistency would vary depending on the case. The 3D kidney models were considered useful, at least for junior physicians, to determine the R.E.N.A.L. nephrometry score, particularly in cases in which the evaluation was difficult. Whether a similar result would be produced with 3D virtual reality images rather than 3D models is a question that remains unanswered. Because 3D models were created using STL data of 3DCT, a 3DVR could be similarly effective as the 3D model. An assessment of the R.E.N.A.L. nephrometry score conducted using 3DVR was shown to predict postoperative complications efficiently [18]; thus, assessment using 3DVR and 3D models are likely to become relevant in the near future.
In our study, 3D models were considered to improve junior resident’s anatomy understanding and improve their diagnostic ability. 3D kidney models have been reported to be effective for student education and patient education [13, 19]; however, it was also found to be effective for educating junior physicians. Due to the widespread use of robotic surgery, junior physicians have performed robot-assisted radical prostatectomy [20]. RAPN is also likely to be performed by junior physicians in the near future, and it is important to improve the ability to read CT images.
This study had some limitations. First, as most cases were cT1a high-risk cases, larger tumors could not be assessed. Further research is needed to verify in which cases the model is useful by increasing the number of renal tumors with high complexity.
Second, although it was better that there was CT and 3D model of the consistency rate of all the residents, because the evaluation of the expert was not validated by other experts, it is unknown whether the evaluation of the expert was accurate. The possibility remains that the expert could have made an incorrect assessment.