Comparison of learning curve difference of laparoscopist according to controller type of flexible scope during laparoscopic liver resection

Abstract

Background

In laparoscopic liver resection (LLR), a skilled scopist is essential for maintaining a clear operative view. We evaluated the scopist’s expertise on surgical outcomes and whether the flexible scope controller type (conventional dual-lever versus joystick) influences the scopist’s learning curve.

Methods

We retrospectively reviewed LLRs performed by a single surgeon between December 2021 and March 2023. Patients with prior hepatectomy, Child–Pugh class B/C, concurrent extrahepatic abdominal surgery, or laparoscopic donor hepatectomy were excluded. Expert scopists (n = 2) had > 7 years of experience, whereas beginner scopists (n = 2) had < 6 months. Outcomes were compared by expertise and scope type. Learning curves were assessed using linear regression and CUSUM analysis of operative time over case sequence.

Results

Among 161 LLRs, 82 were performed with expert scopists and 79 with beginner scopists. Compared with the expert group, the beginner group had longer operative time (178.0 vs 139.5 min, p < 0.001), more intraoperative transfusions (8.9% vs 0%, p = 0.006), and a higher open conversion rate (6.3% vs 0%, p = 0.012). In 116 procedures using conventional scopes (experts, n = 57; beginners, n = 59), experts had shorter operative time (144.0 vs 178.0 min, p = 0.002) and fewer transfusions (0% vs 11.9%, p = 0.013). In 45 procedures using joystick scopes, operative time and transfusion rates did not differ significantly between expert and beginner scopists. In major hepatectomy, both groups showed decreasing operative time with joystick scopes. In minor hepatectomy, experts had shorter operative time with conventional scopes, whereas no significant difference was observed with joystick scopes.

Conclusion

This study demonstrated that the proficiency of laparoscopists can affect surgical outcomes. Moreover, newly learning laparoscopists may acquire proficiency more rapidly when using joystick-controlled laparoscopes compared to conventional ones. In procedures like minor hepatectomy, which are relatively simpler, it was confirmed that there is no significant difference in proficiency when compared to expert scopists.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s00464-026-12909-z.

Laparoscopic liver resection (LLR) has been increasingly adopted worldwide and is now supported by international consensus statements as an effective approach for selected indications, which provided that appropriate training and stepwise implementation are ensured [1, 2]. In high-volume experiences and meta-analyses, LLR has been associated with favorable short-term outcomes compared with open liver resection, and randomized evidence in colorectal liver metastasis also supports lower postoperative morbidity with laparoscopic approaches in selected settings [3, 4].

Despite these advances, safe LLR depends not only on the surgeon but also on stable visualization and effective camera navigation. Laparoscopic visualization is a distinct technical skill; inadequate camera navigation can impair surgical flow and may prolong operative time. Prior studies have validated dedicated camera navigation training models and demonstrated that camera-assistant performance can be improved by structured training or real-time navigation aids [5, 6]. Furthermore, visuospatial ability influences early learning and performance in laparoscopic camera navigation, suggesting that “scopist” proficiency can vary substantially across trainees [7] .

Alongside training, technology design may mitigate the learning burden. Modern three-dimensional (3D) systems have been reported to reduce task time and/or errors in many randomized trials summarized in systematic reviews [8]. The Olympus 3D flexible scope can be manipulated either by a conventional dual-lever controller or by a joystick controller, and such ergonomic/controller differences may influence how quickly a novice scopist acquires the skills required to provide a stable and intuitive view. However, evidence is limited regarding whether controller type modifies the performance gap between expert and novice scopists and whether it affects the scopist learning curve in LLR.

Therefore, this study aimed (1) to evaluate the impact of scopist experience (expert vs novice) on perioperative outcomes in LLR performed by a single surgeon and (2) to investigate whether the controller type of a flexible 3D laparoscope (conventional dual-lever vs joystick) is associated with differences in operative performance and learning curve patterns of scopists, including stratified analyses by extent of liver resection.

Materials and methods

Study design and cohort

This retrospective study evaluated LLRs performed by a single surgeon between December 2021 and March 2023 (Fig. 1). A total of 309 laparoscopic liver resections were screened; after excluding laparoscopic living donor hepatectomy (n = 78) and cases performed with scopists who had already been training for more than 1 year (n = 70), the final analytic cohort consisted of 161 LLRs, representing cases in which endoscopic visualization was provided by newly learning scopists (less than 6 months of experience) and expert scopists (more than 7 years of experience). Additional exclusion criteria included prior hepatectomy, Child–Pugh class B or C, and concurrent extrahepatic abdominal surgery.

Fig. 1.

Flow diagram shows the patient selection criteria for our study

Scopist definition and laparoscope systems

Expert scopists were defined as two scopists with more than 7 years of experience, whereas beginner scopists were two scopists with less than 6 months of experience at the time of cases included in this analysis. Olympus 3D flexible scopes were used in all cases and were operated with either a dual-lever (“conventional”) controller or a joystick-controlled interface (Fig. 2).

Fig. 2.

The Olympus Three-dimensional (3D) flexible scope with A dual-lever-controller type and B a joystick controller type

Surgical procedures and covariates

The extent of liver resection was classified as major or minor. Major resection was defined as resection of three or more Couinaud segments (e.g., right or left hemihepatectomy and trisectionectomy) and also included right anterior sectionectomy and right posterior sectionectomy. Minor liver resection was defined as resection of two or fewer Couinaud segments excluding the above major resections, as well as wedge resection, or left lateral sectionectomy. Baseline demographics and comorbidities (age, sex, BMI, hypertension, diabetes, previous operation history) were extracted.

Outcomes

To compare perioperative outcomes between the expert and beginner scopist groups, the following variables were evaluated: operation time (minutes), estimated blood loss (EBL), intraoperative transfusion, open conversion, postoperative hospital stay, and the occurrence of severe postoperative complications, defined as Clavien–Dindo classification grade IIIa or higher. In addition to the overall comparison by scopist experience, all perioperative outcomes were further analyzed after stratification according to the laparoscope controller type used in each case (conventional dual-lever vs joystick-controlled system), to assess whether scope type modified the association between scopist experience and perioperative outcomes.

To evaluate changes in operative efficiency over time and to explore the learning curve effect, we additionally assessed temporal trends in operation time. Operation time was compared between the expert and beginner groups across chronological time periods, and trend analyses were performed to determine whether the operation time gap between groups changed with increasing case experience.

Statistical analysis and learning curve evaluation

Numerical variables were compared with the two-sample t-test or the Wilcoxon rank-sum test and expressed as mean ± standard deviation (SD) or median with interquartile range (IQR). Categorical variables were compared with the chi-square test or Fisher’s exact test. The statistical significance for each variable was defined as a two-sided p value < 0.05.

The learning curve was evaluated by performing linear regression analyses of operation times according to chronological case order. Operation times were compared between the expert and beginner groups in the overall cohort and after stratification by laparoscope controller type and extent of liver resection. Because surgical learning curves may be nonlinear, CUSUM analysis was additionally performed. The target operation time was defined as the mean operation time of expert scopist cases, and subgroup-specific target values were recalculated using expert cases within each subgroup. CUSUM was calculated as the cumulative sum of the difference between each case’s operation time and the corresponding target operation time and was plotted separately for expert and beginner scopists according to chronological case sequence.

This study was approved by the Samsung Medical Center Institutional Review Board (IRB No. SMC2022-09–004). The requirement for informed consent was waived by the Institutional Review Board because of the retrospective nature of the study and the use of de-identified clinical data.

Results

Baseline characteristics and overall outcomes (Expert vs Beginner)

Table 1 compares the overall baseline characteristics and perioperative outcomes between the expert and beginner groups, regardless of the laparoscope type. Among 161 included LLRs, 82 were performed with expert scopists and 79 with beginner scopists. Baseline characteristics—including age, sex, BMI, diabetes mellitus, prior operation history, extent of liver resection, and ASA score—did not differ significantly between the two groups, except for hypertension.

Table 1.

Patient characteristics

			Expert (n = 82)	Beginner (n = 79)	p value
Age (years)			62.7 ± 10.0	60.4 ± 10.1	0.15
Sex (n, % male)			64 ( 78.0)	64 (81.0)	0.787
Height (cm)			166.3 ± 8.8	165.8 ± 7.3	0.723
Weight (kg)			67.2 [56.1, 73.2]	67.3 [59.9, 74.2]	0.312
BMI (kg/m2)			23.8 ± 3.2	24.6 ± 2.9	0.088
Hypertension (n, %)			22 ( 26.8)	35 (44.3)	0.031
Diabetes mellitus (n, %)			21 ( 25.6)	21 (26.6)	1
Previous OP history (n, %)			15 ( 18.3)	22 (27.8)	0.21
Liver resection extent (n, %)					0.128
Major resection			39 ( 47.6)	48 (60.8)
Minor resection			43 ( 52.4)	31 (39.2)
ASA (n, %)					0.229
1			0 ( 0.0)	2 ( 2.5)
2			78 ( 95.1)	70 (88.6)
3			4 ( 4.9)	6 ( 7.6)
4			0 ( 0.0)	1 ( 1.3)
Operation time (minutes)			139.5 [103.5, 184.0]	178.0 [140.0, 206.0]	< 0.001
Estimated blood loss (ml)			150.0 [100.0, 250.0]	200.0 [100.0, 300.0]	0.094
Transfusion (n, %)			0 ( 0.0)	7 ( 8.9)	0.006
Open conversion (n, %)			0 ( 0.0)	5 ( 6.3)	0.027
Hospital stay (day)			6.0 [5.0, 7.0]	6.0 [5.0, 8.0]	0.012
Severe complication (n, %)			2 ( 2.4)	8 (10.1)	0.054
Death (n, %)			0 (0.0)	0 (0.0)

Regarding perioperative outcomes, cases assisted by beginner scopists showed significantly worse operative efficiency and intraoperative events. Median operation time was significantly longer in the beginner group than in the expert group (178.0 [140.0–206.0] vs 139.5 [103.5–184.0] minutes, p < 0.001). Intraoperative transfusion occurred more often in the beginner group (8.9% vs 0.0%, p = 0.006), and open conversion was also more frequent in the beginner group (6.3% vs 0.0%, p = 0.027). Postoperatively, hospital stay was longer in the beginner group (6.0 [5.0–8.0] vs 6.0 [5.0–7.0] days, p = 0.012). The rate of severe complications tended to be higher in the beginner group (10.1% vs 2.4%), but this difference did not reach statistical significance (p = 0.054).

Perioperative outcomes by scopist experience stratified by laparoscope controller type

Among the 161 cases included in the analysis, 116 procedures were performed using the conventional controller, with 57 cases assisted by expert scopists and 59 cases assisted by beginner scopists. The remaining 45 procedures were performed using the joystick-controlled system, including 25 cases in the expert group and 20 cases in the beginner group (Table 2).

Table 2.

Comparison of perioperative outcomes between expert and beginner scopists according to laparoscope controller types

		Conventional (n = 116)			Joystick (n = 45)
		Expert (n = 57)	Beginner (n = 59)	p value	Expert (n = 25)	Beginner (n = 20)	p value
Operation time (minutes)		144.0 [100.0, 186.0]	178.0 [141.5, 203.5]	0.002	124.0 [106.0, 171.0]	174.0 [138.8, 211.0]	0.059
Estimated blood loss (ml)		150.0 [100.0, 200.0]	200.0 [100.0, 275.0]	0.092	200.0 [100.0, 280.0]	225.0 [107.5, 300.0]	0.468
Transfusion (n, %)		0 ( 0.0)	7 (11.9)	0.013	0 ( 0.0)	0 ( 0.0)	-
Open conversion (n, %)		0 ( 0.0)	4 ( 6.8)	0.119	0 ( 0.0)	1 ( 5.0)	0.444
Hospital stay (day)		6.0 [5.0, 7.0]	6.0 [5.0, 8.0]	0.084	5.0 [5.0, 6.0]	6.0 [5.8, 7.2]	0.040
Severe complication (n, %)		2 ( 3.5)	7 (11.9)	0.163	0 ( 0.0)	1 ( 5.0)	0.444

In the conventional controller subgroup, cases assisted by beginner scopists demonstrated significantly worse perioperative performance. Median operation time was significantly longer in the beginner group than in the expert group (178.0 [141.5–203.5] vs 144.0 [100.0–186.0] minutes, p = 0.002). In addition, intraoperative transfusion occurred only in the beginner group (11.9% vs 0.0%, p = 0.013). In contrast, in the joystick-controlled subgroup, the differences between expert and beginner scopists were attenuated. Operation time did not significantly differ between the two groups (174.0 [138.8–211.0] vs 124.0 [106.0–171.0] minutes, p = 0.059), and no transfusions occurred in either group (0.0% vs 0.0%).

Learning curve patterns according to the extent of liver resection

Operation times for individual cases were plotted over time using dot plots, separately for major and minor liver resections, and linear regression lines were presented to evaluate temporal trends in the expert and beginner groups. In addition, operation times were compared between groups across consecutive 4-month intervals to assess statistically significant differences within each time window (Fig. 3, 4).

Fig. 3.

Learning curve patterns of operation time in major liver resections

Fig. 4.

Learning curve patterns of operation time in minor liver resections

In the major liver resection subgroup (Fig. 3A–C), distinct temporal patterns in operation time were observed according to laparoscope controller type. In the overall analysis (Fig. 3A), operation time tended to decrease over time in the expert group, whereas the beginner group showed a relatively flat to slightly increasing trend, resulting in a progressively widening separation between the two groups. When stratified by controller type, this divergence was most pronounced in the conventional controller subgroup (Fig. 3B). The regression line for the beginner group demonstrated no appreciable reduction in operation time across the study period and instead showed an increasing tendency, while operation time in the expert group decreased, leading to an increasing gap between expert and beginner cases as time progressed. In contrast, in the joystick-controlled subgroup (Fig. 3C), both expert and beginner groups exhibited decreasing operation time across the study period, although the expert group maintained shorter operation times and the intergroup difference still tended to widen. Overall, these findings suggest that, for major resections, operative efficiency improved over time in both groups when the joystick-controlled system was used, whereas beginner-assisted cases did not show comparable improvement with the conventional controller type.

In the minor liver resection subgroup (Fig. 4A–C), overall operation times were shorter than those observed in major resections, and the difference in operation time between expert and beginner scopists was generally smaller across the study period. In the overall analysis regardless of controller type (Fig. 4A), the regression lines for the two groups were relatively close, indicating a limited separation in operative time between expert- and beginner-assisted cases. When stratified by controller type, however, distinct patterns emerged. In the conventional controller subgroup (Fig. 4B), operation time was consistently longer in cases assisted by beginner scopists, resulting in a more pronounced gap between the two groups. In contrast, in the joystick-controlled subgroup (Fig. 4C), the regression lines for expert and beginner scopists largely overlapped, suggesting minimal differences in operation time between groups when the joystick system was used.

CUSUM analysis of operation time

CUSUM analysis was performed to further characterize nonlinear learning curve patterns according to case sequence. In the overall cohort, the expert group showed only limited fluctuation around the target line and returned close to zero by the end of the case sequence, whereas the beginner group demonstrated a sustained upward CUSUM trend, indicating cumulative excess operation time compared with the expert-based target (Fig. 5A).

Fig. 5.

CUSUM analysis according to case sequence using expert-based target operation times. A overall cases, B conventional scope cases, C joystick scope cases, D major hepatectomy cases, E major hepatectomy with conventional scope, F major hepatectomy with joystick scope, G minor hepatectomy cases, H minor hepatectomy with conventional scope, and I minor hepatectomy with joystick scope

When stratified by controller type, the divergence between expert and beginner scopists was more pronounced in the conventional scope subgroup. In conventional scope cases, the beginner group showed a continuous upward CUSUM trajectory, while the expert group remained relatively stable around the target value. In contrast, in joystick scope cases, the difference between the two groups was attenuated. Although the beginner group still accumulated positive CUSUM values, the magnitude of increase was substantially smaller than that observed with the conventional scope (Fig. 5B, C).

In the major hepatectomy subgroup, CUSUM analysis again showed a clear separation between expert and beginner scopists. This pattern was most evident in major hepatectomy performed with the conventional scope, in which beginner-assisted cases demonstrated progressive accumulation of positive CUSUM values. By contrast, in major hepatectomy performed with the joystick scope, the beginner group showed only a modest positive CUSUM increase, and the separation from the expert group was less pronounced. (Fig. 5D, E, F).

In the minor hepatectomy subgroup, the beginner group also showed positive CUSUM accumulation overall, particularly in cases using the conventional scope. However, in minor hepatectomy performed with the joystick scope, both expert and beginner groups remained within a relatively narrow CUSUM range, suggesting minimal cumulative difference in operation time between groups. Overall, the CUSUM analysis supported the findings of the linear regression analysis and further demonstrated that joystick-controlled scope use was associated with attenuation of the cumulative operation time gap between expert and beginner scopists, particularly when compared with the conventional controller. (Fig. 5G, H, I).

Discussion

This study demonstrates that scopist proficiency is clinically meaningful in LLR. Cases assisted by beginner scopists had significantly longer operation times, higher intraoperative transfusion rates, and more frequent open conversions. These findings align with the broader LLR literature emphasizing that safe implementation requires structured training and team readiness, as highlighted by international consensus recommendations. [1, 2] Importantly, our data extend this concept by isolating the scopist as a key contributor to efficiency and selected safety outcomes, even when the operating surgeon is constant.

A plausible mechanism is that laparoscopic camera navigation is not a passive role but an independent technical skill that determines stability, horizon control, and timely exposure of critical planes. This interpretation is supported by studies showing that simulation-based camera navigation training improves technical skills and that novice camera assistants demonstrate measurable proficiency gain curves during structured image navigation tasks [9, 10]. Educational studies have shown that novice scopists frequently struggle with correct handling of angled scopes and consistent visualization, and that even brief, structured training can transform performance. [6, 11] Moreover, improved guidance systems (e.g., navigation overlays) and task-specific training tools measurably enhance image navigation accuracy and team efficiency, supporting the concept of a quantifiable scopist learning curve. [5, 12] Interindividual variability (including visuospatial ability) further supports why some beginners adapt quickly while others require more time and feedback. [7]

A key observation in our cohort is that the joystick-controlled flexible scope appears to mitigate the performance gap between expert and beginner scopists. Although operation time remained numerically longer with beginners in the joystick subgroup, the difference did not reach statistical significance, and transfusion events were not observed in either group. This suggests that joystick control may provide more intuitive camera steering and facilitate faster acquisition of stable exposure behaviors. While our study was not designed to measure camera motion metrics directly, existing evidence supports the idea that technology can accelerate visuomotor learning in laparoscopy. Randomized trial reviews have reported improved task speed/efficiency with 3D laparoscopy, particularly for novices, presumably by enhancing depth perception and reducing cognitive load during complex spatial maneuvers. [8] Recent randomized training data also show improved learning trajectories under 3D visualization. [13] In the LLR context—where the surgical field is dynamic and parenchymal transection requires frequent microadjustments of view—these advantages may translate into shorter “time to acceptable proficiency” for newly learning scopists.

The effect of controller type may depend on procedural complexity, and our findings suggest that the potential benefit of joystick control becomes more apparent in technically demanding cases. In major hepatectomy, operation time in the beginner group showed little improvement across the 16-month study period when the conventional controller was used and even demonstrated an increasing tendency, whereas a more noticeable reduction in operation time was observed when the joystick-controlled system was used. In the expert group, operation time decreased over time with both controller types; however, the decline appeared steeper with joystick control, supporting the notion that the joystick interface may be more intuitive and may facilitate earlier attainment of proficiency in camera manipulation. By contrast, in minor hepatectomy, overall operation times were shorter and the expert–beginner separation was relatively small, with minimal differences between groups when the joystick-controlled system was used. Taken together, these patterns indicate that the advantage of joystick control may be most relevant as procedural complexity increases, where stable visualization and efficient camera navigation are more likely to influence operative efficiency. This pattern is clinically plausible: in simpler operations, experts may primarily outperform beginners through refined anticipation and smoother coordination, while in complex operations the interface itself (controller type and 3D stability) may contribute proportionally more by preventing disorientation and reducing camera-related delays. Nonetheless, these interpretations require confirmation in studies that include objective camera motion analytics and standardized difficulty scoring.

The CUSUM analysis provided an additional perspective by visualizing cumulative deviation from expert-based target operation times. In the overall cohort and the conventional scope subgroup, beginner-assisted cases showed progressive accumulation of positive CUSUM values, whereas expert-assisted cases remained closer to the target line. This suggests that the performance gap was not limited to isolated prolonged cases but accumulated consistently across sequential cases. Notably, the cumulative separation between expert and beginner scopists was smaller in the joystick subgroup, particularly in major hepatectomy and in minor hepatectomy with joystick control. These findings support the interpretation that joystick-controlled scope manipulation may attenuate the cumulative learning burden of beginner scopists. Nevertheless, because CUSUM analysis is descriptive and sensitive to the selected target value, these results should be interpreted as exploratory and should be validated in larger studies incorporating objective camera navigation metrics.

Several limitations should be acknowledged. First, this was a retrospective, single-center analysis in which all procedures were performed by a single surgeon. Although this design enhances internal consistency by minimizing intersurgeon variability, it may limit the generalizability of our findings to other institutional settings and operative teams. Importantly, however, the joystick-controlled flexible scope had already been used in our center for more than 5 years before the study period, and the operating surgeon was already experienced with both the conventional dual-lever and joystick-controlled systems. Therefore, the observed learning curve patterns are unlikely to be explained primarily by surgeon adaptation to the joystick system during the study period. Rather, the main difference in prior exposure was among beginner scopists, who had less than 6 months of scopist experience and limited experience with flexible scope manipulation. Second, the observation period was relatively short, which restricts our ability to characterize the long-term learning curve of scopists and to determine whether performance differences persist, plateau, or converge over extended follow-up. Third, the sample size, particularly after stratification by controller type and extent of resection, was limited, reducing statistical power for subgroup analyses. Finally, the learning curve was assessed primarily using operation time as a surrogate measure. While operation time is clinically relevant, it may not fully capture technical aspects of scopist performance, such as camera path length, horizon stability, frequency of view loss, instrument–camera coordination, and communication dynamics within the operative team. Future studies incorporating objective camera motion metrics and longer follow-up across larger cohorts are warranted to validate and refine these observations.

In conclusion, scopist proficiency significantly influences operative efficiency and selected intraoperative outcomes in LLR. Joystick-controlled flexible laparoscopy may attenuate the scopist learning curve compared with conventional dual-lever control, suggesting that both training structure and camera system design are important levers for safer and more efficient dissemination of advanced laparoscopic liver surgery.

Supplementary Information

Below is the link to the electronic supplementary material.

Funding

Open Access funding enabled and organized by SungKyunKwan University. This study was supported by research funding from Olympus. This company was not involved in the study design, data collection, data analysis, data interpretation, manuscript preparation, or the decision to submit the manuscript for publication.

Declarations

Disclosures

This study was supported by research funding from Olympus. Olympus was not involved in the study design, data collection, data analysis, data interpretation, manuscript preparation, or the decision to submit the manuscript for publication. Drs. Sunghae Park, Namkee Oh, Jinsoo Rhu, Gyu-Seong Choi, and Jongman Kim have no conflicts of interest or financial ties to disclose.