Surgical Training Outcomes Using a Mixed Reality Combination System

doi:10.21203/rs.3.rs-4978732/v1

Download PDF

Research Article

Surgical Training Outcomes Using a Mixed Reality Combination System

https://doi.org/10.21203/rs.3.rs-4978732/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

A significant number of medical errors are surgical, when patients are in the operating room, and are also a prominent cause of death. Prior work introduced a Mixed Reality Combination System (MRCS) that integrates Augmented Reality (AR) technology, an inertial measurement unit (IMU) sensor, and 3D-printed, collagen-based specimens to enable realistic and versatile synthetic environments for surgical training, with the goal of alleviating this issue. The MRCS creates an iterative learning scenario that allows a user to pick up a skill set in a limited time. This paper demonstrates the advantages of the MRCS technology through human subjects studies that assess its effectiveness. Results from a controlled study show that the MRCS study participants are 25% better at task execution, reach a net zero difference in expected task outcomes in 75% of the cases, and demonstrate self-confidence in their surgical skills compared to the non-MRCS study participants. A second qualitative study with medical residents demonstrates MRCS’s proposed benefit as a training tool in a medical school curriculum. This work has the potential to benefit future surgical training and surgical planning.

Surgical procedure errors account for over 40% (Pham et al. 2012; Landers 2015; Makary and Daniel 2016; Anderson and Abrahamson 2017) of known medical errors when administering health care to patients in the United States, and over 50% (Balogun et al. 2019) in the developing world. These medical errors account for approximately $17.1B of medical costs in North America alone (Encinosa and Hellinger 2008; Van Den Bos et al. 2011) and are responsible for over 250,000 deaths annually (Anderson and Abrahamson 2017). There is a range of reasons for these errors. However, there is a strong correlation between inadequate skills training and limited clinical exposure (Bari et al. 2016), especially in specific surgical scenarios that only present themselves to surgeons in clinical practice. These advanced scenarios can be anywhere from few to no instances during their time in the clinical field, leading to errors during the surgical operation (Makary and Daniel 2016).

Efforts to combat errors made during surgical operations involve using surgical operation aids like shared decision-making during surgical procedures (Thunnissen et al. 2021). However, these are difficult to implement across surgeons of varying skills. To attain these skills before clinical field exposure, training is done using cadavers in medical school training programs (Sheikh et al. 2016). This training model is based on the Halstedian model of "see one, do one, teach one" (Yiasemidou et al. 2018; Selcuk et al. 2019; Robinson et al. 2020) and, as a form of advanced surgical skills curriculum early in surgeon residency (Bowyer et al. 2013; Kuhls et al. 2013), is employed to simulate these medical scenarios and improve surgical techniques. However, these cadavers not only lack the realism (M.W. et al. 2018; Waerlop et al. 2021) of the body anatomy and its functions due to organ shrinkage and tissue degradation (COLEMAN and KOGAN 1998; Robinson et al. 2020) but also are complimentary at best when not accompanied by other innovative learning methods and technologies like computerized tomography (CT) that produces cross-sectional images through computerized x-ray imaging (Sheikh et al. 2016; Ghosh 2017). Hence, in specific scenarios, when medical school graduates do encounter these or new medical scenarios as practicing surgeons, they have limited experience and are then dependent on long-term exposure during clinical practice (HAMMOND et al. 2003; Kuhls et al. 2013; Balta et al. 2015; Gabrysz-Forget et al. 2020). In addition, the current nature of surgical training requires a significant time investment to gain the necessary proficiency (Blasier 2009; Kennel et al. 2018; Agius et al. 2018; Weininger et al. 2020; Kowalczyk and Majewski 2021), Clinical experience can help alleviate deficiencies in surgical technique effectiveness for new practicing surgeons as they seek to shorten the time required to gain proficiency (Mackenzie et al. 2013; Phillips et al. 2016). However, the clinical exposure time necessary for this gain can work against the lifespan of a practicing surgeon (Gottlieb-Vedi et al. 2019; Markar et al. 2020; Satkunasivam et al. 2020; Lee et al. 2021) despite efforts to refine the medical curriculum to focus on performance goals like skill retention to improve this proficiency gain (Scott 2008; Fried et al. 2012).

This gap motivates using enabling technologies to provide a cheap and versatile method for user training and iteration practice. In prior work, we introduced a Mixed Reality Combination System (MRCS) Platform (Fig. 1) (Kabuye et al. 2023). The MRCS combines Augmented Reality (AR), 3D Bioprinting, precise location sensing, and computational simulation to provide a system that enables a new type of surgical training that is realistic, iterative, and departs from the needs of cadavers. The MRCS platform enables a realistic and real-time synthetic environment in which the user can both see and accurately feel the dynamism of tissue within a surgical site, providing a potential strategy for ensuring that surgeons grasp medical knowledge and technical expertise that they can then execute during a surgical procedure while minimizing error.

This paper explores the effectiveness of using the MRCS system for user training and demonstrating user proficiency gain in task execution in users. This is accomplished through two human subjects studies designed to test skill acquisition in terms of proficiency gain amongst the users when subjected to a training module by the MRCS. One study quantitatively tests effectiveness against a control; the other qualitative study uncovers medical residents' reactions and feedback.

The use of enabling technologies like Virtual Reality (VR) in operating rooms for training, due to their iterative scope as a benefit, has also exposed a lack of haptic feedback and high implementation costs(Van Der Meijden and Schijven 2009; Våpenstad et al. 2013; Kim et al. 2019; Bing et al. 2019). Essentially, the user can see the environment but not feel it. Instead, the MRCS uses a combination of Augmented Reality (AR) and novel and more realistic haptic feedback. Augmented Reality systems that input virtual imagery into physical spaces have been introduced in medical training(Lam et al. 2014; Hettig et al. 2018; Sánchez-Margallo et al. 2021) and surgical operations(Kanevsky et al. 2019; Sakai et al. 2020) and for navigation and accurate dissection by identifying anatomical features (Chen et al. 2015; Pratt et al. 2018). Yet all of these work at an abstraction from physical interaction(Chen et al. 2015; Pratt et al. 2018). MRCS also uses AR via the Microsoft HoloLens in its current implementation as a visualization aid during surgical interventions using its onboard sensors for depth perception needed for the hologram patient registry (Sylos Labini et al. 2019).

To overcome physical abstraction and enable physical realism, the MRCS uses 3D-printed hydrogels to represent human anatomy constructs. (Lee et al. 2019; Noor et al. 2019; Mirdamadi et al. 2020), resulting in biospecimens whose bio-compatibility and highly tunable mechanical properties make it easier for them to act as anatomical substitutes (Chuah et al. 2017; Zheng and Du 2021). 3D Bio-printing using collagen has been used to provide structures and behaviors similar to the tissue (Armstrong et al. 2016; Rastogi and Kandasubramanian 2019; Sahoo and Biswal 2021; Stapenhorst et al. 2021). A 3D-printed biospecimen can be customized to reflect the surgical pathology for which surgical training is challenging to navigate for a practicing clinician. In the MRCS platform, the virtual environment from an AR technology is merged with the 3D printed collagen models, resulting in a Mixed Reality system. Within MRCS, collagen is chosen for the Bioprinting of the specimen because it can mimic human tissue properties (Lee et al. 2019).

The MRCS platform merges these two capabilities, i.e., virtual worlds and 3D Bioprinting, to create an AR visual environment with realistic haptic interaction, representing the feel of deformation of actual tissue. The last major component is introducing a flexible IMU integrated into a scalpel to track the scalpel’s location if it cuts the collagen sample (Kabuye et al. 2023). Further, dynamic finite element models represent how the tissue deforms under that cut within the AR visual model. The entire MRC platform is seen in Fig. 1. In a previous paper (Kabuye et al. 2023), we introduced this novel system and demonstrated its ability to track and respond in real-time to user interactions. This paper provides a pair of human subjects studies to assess its benefit by comparing its use by novices and more advanced medical residents.

A Mixed Reality Combination System (MRCS) platform is introduced to provide surgical skills training to users. Two human studies are conducted: a quantitative human analysis to capture user metrics and performance and a qualitative study to capture a user's experience, behavior, and learnings with the MRCS Platform as a training tool. The first human subjects study is run to understand the benefit of using the MRCS as a training platform among users with limited to no knowledge of medical anatomy, determining the user proficiency gain and user effectiveness of the MRCS system. The second study informs its use within a surgical training environment. Both studies also have a series of questionnaires for the participants, pre- and post-task execution, to determine and ascertain their level of expertise and track their user perception of any effect of interacting with the MRCS platform to assess its effectiveness. At the start, participants are trained on how to use the system. In the first study, as a control, only half of the study participants were exposed to the MRCS training and task planning. All study participants are instructed to execute the identical task, and their task execution metrics are tracked and evaluated.

2.1. Human Subjects Study Participants Setup

Human Subjects Study Approval and Participant Consent

This study is approved by the Institutional Review Board at Carnegie Mellon University and is conducted according to the relevant guidelines and regulations. Informed consent is obtained from the human study participants.

Human Subjects Study I – Non-Medical Graduate Engineering Students (Quantitative)

A group of 40 study participants is recruited from the Carnegie Mellon University College of Engineering. All the study participants are graduate students with a bachelor's degree in engineering. As part of a pre-participation questionnaire, the study participants are required not to be prone to seizures from strobe lights or have any negative connotation towards the medical profession. These study participants are broken up into two distinct groups. The first group is exposed to the MRCS Platform training, and the second group is not exposed to the MRCS training. In total, 40 students are randomly assigned to either use the MRCS platform (MRCS condition) or not use the MRCS platform (non-MRCS condition). Each participant receives a total of $70 for their full participation.

Human Subjects Study II – Medical Residents (Qualitative)

Seven study participants who are active medical residents at the Allegheny Health Network Wexford Hospital are recruited. The residents are undergoing medical training in the field of Pediatric Orthopedics. Similarly to the first quantitative human subjects study with graduate engineering students, these residents are taken through similar checkpoints via the pre-questionnaire. They are required not to be prone to seizures from strobe lights. All medical residents in the human study are required to participate in the execution of the task using the MRCS. During engagement and use with the MRCS, their feedback is recorded via audio and video. During the recording, there are no visible or identifiable marks of the participants as recorded to maintain an anonymous study. In addition, no performance metrics, in terms of incision depth, task completion time, etc., from these medical residents are captured. An anonymous proctor analysis of their performance in the task execution is obtained. The proctor is a highly trained surgeon in the field of surgical training. Unlike the first study, there is no control in the second study; instead, it provides qualitative feedback on using the platform in training practice. Each study participant receives a total of $70 for their full participation.

As a safety precaution, each participant from either study is informed that they can stop their participation at any time throughout its execution if they feel uncomfortable. After getting verbal agreement from the participants, consent forms indicating their participation are read and signed.

2.2. Human Study Surgical Training and Task Execution Design

The human subjects study is designed to take a maximum of 2 hours for the MRCS condition participants and 1.5 hours for the non-MRCS condition participants, with the difference of 30 minutes in execution time for the MRCS condition participants set aside for tutorials on the usage of the MRCS platform, specifically, the Microsoft HoloLens 2 system. During the task tutorial and walkthrough section, participants with non-MRCS conditions must not be aware of the MRCS. Each participant executes the study separately from the others and follows a timeline based on their group designation (Table 1).

Surgeon experience positively correlates with surgical outcomes, especially in scenarios requiring an advanced skill set. Proficiency Gain in surgery skills is an excellent metric tracked for predicting this success (Scott 2008; Mackenzie et al. 2013). In the current study, selecting a scenario requiring an advanced technique for success is critical. In this regard, a scenario is artificially created that involves the presence of a dense mass, such as a tumor at a surgical site for a ventral hernia incision in an advanced-age patient that is considered high-risk (Ammar 2009; Choi and Lee 2018; Asencio et al. 2021). A ventral hernia is a protrusion of the intestine or other tissue through a weakness or gap in the abdominal wall (Misiakos et al. 2015). The initial surgical technique step is identifying the target site, followed by a 5mm to 10mm keyhole incision at the primary location for access afterward. However, a dense abnormal mass at the surgical site means the tissue material has lost its elasticity and form factor. This means that the tissue is much more susceptible to tears along the axis of a vertical incision cut. If the target depth is exceeded, the patient being operated on can suffer extreme pain and, in some instances, death from additional complications. This scenario ensures that the task planning and task execution of the user is critical to the positive outcome of the surgical technique. For the study, the participant is trained to approach the site, make a vertical incision at this target site, and not exceed a 2.5mm depth cut. The task execution of making the incision is repeated over a set number of times to demonstrate user proficiency in skill acquisition (Fig. 2).

2.2.1. User Proficiency

To evaluate user proficiency gain, the user's performance in the task, precisely the difference between the user's surgical incision depth and the target incision depth, is measured and tracked over the subsequent iterations using 9DOF IMU attached to the surgical tool (Kabuye et al. 2022). This difference for each iteration is plotted to determine when the user difference approaches zero and maintains a minimum zero value on their curve. At this point, the user is determined to have acquired a skillset and can be considered proficient (Mackenzie et al. 2013; Novoa and Varela 2020). The deviation from the target value is compared across both groups to determine which group has achieved proficiency gain. In the created task scenario, the task execution is carried out for 40 iterations, as similar surgical techniques estimate a proficiency gain curve change point of 31 for lower volume surgeons (van Workum et al. 2018) in addition to reaching independent performance for similarly tested tasks between 20 and 40 exposures (Mackenzie et al. 2013).

2.2.2. User Effectiveness

To evaluate the user effectiveness derived from using the MRCS system, the study is designed per Rasmussen's model of human behavior in Laparoscopy training(Wentink et al. 2003) and the NASA-TLX rating of task difficulty and physical and mental efforts (Hart and Staveland 1988).The Rasmussen model is based on three behaviors tested during training to establish user proficiency: skill-based behavior, rule-based behavior, and knowledge-based behavior. The behaviors increase in complexity from skill to rule and then knowledge.

The skill-based behavior to be tested is a vertical incision by a study participant into a target zone at a set depth. The rule-based behavior is based on a procedural step that ensures that the target tissue is oriented in the right direction for surgical operation. This is tracked for the MRCS condition participants when they place the AR imagery correctly to align with the 3D bio print specimen for a surgical incision. The non-MRCS condition participants are tracked in terms of how they identify the placement of the 3D bio print specimen on the test table. Knowledge-based behavior is the most complex and is determined when a user faces a task without previous rules from previous encounters, namely identifying an anatomy whose shape is outside the norm. In this case, the knowledge-based behavior is to identify the location of the representative human anatomy, the 3D-printed biospecimen, where the vertical incision is to be made.

The NASA Task Load Index (Hart and Staveland 1988; Hart 2006; Hoonakker et al. 2011) examines a subject's perception of workload rating before and after task performance. The user survey is customized for this rating to include questions that speak to the six workload ratings of Mental, Physical, Time, Performance, Effort, Frustration, and Competence. In addition, MRCS-condition participants are asked how helpful the MRCS is in their task execution.

To capture the user's perception of effectiveness, a pre- and post-study survey for study participation is completed for both groups. The survey captures metrics such as age, gender, surgical expertise, and skill self-evaluation using a 7-point Likert scale. The average incision depth and the self-perceived and measurable user performance between both groups are compared to determine the MRCS's effect.

2.2.3. Human Subjects Study Task Training Setup

Before task execution, both sets of participants fill out a pre-qualification survey on their basic knowledge of surgical tasks, human anatomy, and any medical history of seizures. Only study participants who meet satisfactory criteria proceed to the actual study.

The human subjects study is done in seven sessions covering tutorials, questionnaires, and the human study task execution, as shown in the timeline in Table 1.

The first session is a pre-questionnaire, asking for demographics such as age, gender, and university department information. It also asks about the study participant's video game exposure habits and perceived skill rating (Table 2). This question eliminates any potential users for whom hand-eye coordination might have already been built into their skillset due to increased exposure to video games (2005; Giannotti et al. 2013; Datta et al. 2020; Gupta et al. 2021).

Table 1

Human study session timeline
Session #	Surgical Task	Study Participant	Skill Under Test	Time	#Data Points Collected
1	Pre-Questionnaire
2	Study Setup	All	Surgical Tool Training	0.25 hours	N/A
3	Human Anatomy Identification	All	Training & Planning	0.25 hours	Attribute Data(Y/N)
4	MRCS Platform tutorial	MRCS-Only	Mixed Reality Training	0.5 hours	N/A
5	Task Execution	All	Surgical Incision Depth	0.5 hours	20
	Optional − 15 min break
6	Task Execution	All	Surgical Incision Depth	0.5 hours	20
7	Post Questionnaire

Table 2

Pre-Task survey on participants' perceived level of skill – pre-questionnaire
QN. #	Question (Designation)	Rating
1	How confident are you currently in your ability to make an effective surgical plan before its execution? (Competency; Pre-Study)	o Very Incompetent, o Incompetent, o Somewhat Incompetent, o Neither Incompetent nor Competent, o Somewhat Competent, o Competent, o Very Competent
2	How confident are you in executing an effective surgical plan? (Pre-Task Execution)	1–10

The second session, the study setup, covers the use of surgical tools by both groups of study participants to execute the task. After being briefed from the first study walkthrough, participants are shown and instructed on handling a surgical tool as a safety precaution. They are also offered a measurement scale demonstrating what target depth and verbal confirmation are sought after demonstration as part of this training. Both groups during this session are shown what constitutes a single vertical incision (Fig. 3).

Both groups of study participants are then shown how to hold (Fig. 3) and use a scalpel fitted with a 9 DOF IMU to track their real-time information, and the MRCS condition participants, specifically, are shown how to relay this location to the Microsoft HoloLens component of the MRCS (Kabuye et al. 2023), so they obtain updated imagery of the new guided position and match their hand motion to the path for complete task execution of a single vertical incision. This step is then to be repeated 40 times. The viewpoint of the progression of the vertical cut during the task execution is captured via video on the MRCS for additional proctor review.

In the third session, the study participants are provided a walkthrough of the study's disease condition of the ventral hernia and the incision's target site (Misiakos et al. 2015). This target site, located at the middle right center of the human abdomen, is referred to as the "Primary Position" and is shown to both participants for identification and clarification. Their acknowledgment of the "primary position" and location is captured.

The fourth session is primarily for the MRCS condition participants, and the non-MRCS condition participants skip this session and move on to session five. In the fourth session, the MRCS condition participants are trained to use the MRCS through exposure to interactive images (Fig. 4). All the anatomy models used for this study are built using the Unity platform, and the corresponding dynamic properties are added (Kabuye et al. 2023). During this exposure, the participant is taught the basics of mixed reality interaction, such as image manipulation, registration, and spatial awareness. The AR imagery for image manipulation and hand registration is a human brain with highlighted bounded corners. The MRCS study participant must use the bounded corners to master hand-eye coordination for dexterous tissue manipulation and use imagery to get exposure to human anatomy. For spatial awareness, the MRCS study participant is required to recalibrate the eye tracking for the environment per the instructions. Once done, the MRCS study participants have full accessibility and spatial awareness of the imagery they interact with during the task execution.

After the hand manipulation tutorial, the MRCS condition participants identify the "primary position" for surgical access on the projected human anatomy imagery. In this portion of session four, they are also tasked with identifying the "primary position" for surgical access in multiple positions of the human anatomy, from upright to back and finally to a reverse supine position. Their identification is noted throughout all three parts of the human anatomy they are tasked with identifying (Fig. 5).

The following two sessions, five and six, are the actual task execution sessions for both groups of study participants. They are instructed to execute the surgical task in the assumed primary position on the surgical target site. In this task execution, the surgical site target is a 3D bio-printed specimen (Lee et al. 2019; Mirdamadi et al. 2020). The study participants are to make 40 vertical incisions with an optional break given to them at the halfway point.

Human Study I Task Execution

The study participants begin the task execution and are video recorded and timed throughout the task completion. The video recording is only made of their hands during the task execution, and no identifying markers are present.

For the MRCS condition participants, the test begins with instructions from the MRCS Component, Microsoft HoloLens, on how to proceed. The study participants indicate on the imagery of the body anatomy the primary position, at which point an animation informs them of their primary position identification (Fig. 6).

After identifying the "Primary Position," the MRCS condition participant manipulates the virtual imagery of the human anatomy to the supine position and resizes its size and position to the test platform to ensure that the "primary position" is placed correctly as instructed during the tutorial as shown in Fig. 7.

Throughout the task execution, the MRCS condition participant is guided by the position of their movements in the location. As they interact with the virtual imagery, the updated status of their location triggers an animation at the surgical site corresponding to the exact incision depth they are making. When they stop, the animation of the separated tissue retracts to the new position of their surgical blade, with their interactions acting as a guide. In scenarios during task execution, when the target depth is exceeded, the virtual anatomy becomes transparent, signaling them to stop (Fig. 8).

Unlike the MRCS condition participant, the viewpoint of the non-MRCS condition participant is only the 3D bioprint specimen, as shown in Fig. 9.

The last session, 7, has all the study participants filling out a post-questionnaire survey. In the post-questionnaire, they provide self-evaluation ratings from 1–10. The questions are listed in Table 3.

Table 3: Post-task execution survey for human subjects study participants

Qn. #	Question (Designation)	Rating
1	How confident are you currently in your ability to make an effective surgical plan before its execution? (Effective HSP)	1- 10
2	How confident are you currently in your ability to make an effective surgical plan before its execution? (Competency; post-study)	Very Incompetent, Incompetent, Somewhat Incompetent, Neither Incompetent nor Competent, Somewhat Competent, Competent, Very Competent
3	Was the instruction from the MRCS/Instructor Clear throughout the session	(Yes/No)
4	Was the MRCS comfortable during the whole study? (MRCS Study Participants Only)	(Yes/No)
5	Do you feel satisfied with the outcome of your surgical cut? (post-test satisfaction)	(Yes/No)
6	Did the iterative and successive sessions improve your skill set?	(Yes/No)
7	In your opinion, how would you rate your current skill level for surgical planning? (Post-Task Planning)	1- 10
8	In your opinion, how would you rate your current skill level for surgical task execution? (Post-Task Execution)	1- 10
9	Is there anything that can improve the setup?	(Yes/No)

Finally, using the NASA Task Load Index (TLX), study participants completed a questionnaire to close out the human subjects study, as shown in Table 4. The NASA Task Load Index (TLX) obtains the first six questions. Each question corresponds to one scale, and participants mark between 0 and 100 on the scale, with 0 corresponding to Very Low and 100 corresponding to Very High as the answers. In the fourth of the sixth original NASA TLX question, 0 corresponds to Perfect, and 100 corresponds to Failure. This change in scoring is communicated to all study participants before taking the assessment. These results from the NASA TLX study are to be compared across both groups of study participants.

Table 4: NASA TLX post-task execution questionnaire

Qn. #	Question (Designation)	Rating
1	How mentally demanding is the task? (Mental)	0 (Very Low)- 10 (Very High)
2	How physically demanding is the task? (Physical)	0 (Very Low)- 10 (Very High)
3	How hurried or rushed is the pace of the task? (Temporal)	0 (Very Low)- 10 (Very High)
4	How successful were you in accomplishing what you were asked to do? (Performance)	0 (Perfect)- 10 (Failure)
5	How hard did you have to work to accomplish your level of performance? (Effort)	0 (Very Low)- 10 (Very High)
6	How insecure, discouraged, irritated, stressed, and annoyed were you? (Frustration)	0 (Very Low)- 10 (Very High)
7	How helpful was the MRCS/Instructor to you in helping to execute the task? (Medium of Helpfulness)	Very Unhelpful Unhelpful Somewhat helpful Neither unhelpful nor helpful Somewhat helpful Helpful Very Helpful

A summary of the study outlines and goals is presented below in Table 5.

Table 5

Human subject studies format
Parameters	Study 1	Study 2(Medical Resident)
Type	Quantitative	Qualitative
Participants	20 for each group 40 participants total	7
Group Breakdown	Group 1a: non-MRCS Group 1b: MRCS	MRCS
Clinical Skill	None	Obtaining/Obtained Medical Degree
Primary Objective	Proficiency Gain User Effectiveness	MRCS User Effectiveness
Metric	Vertical Incision Depth Minimum Zero curve	Self-Assessment Surveys NASA-TLX
Analysis	2 Sample t-Test NASA -TLX Self-Assessment Surveys	Proctor Video Analysis

3.1. Human Subjects Study I – Engineering Graduate Students (Quantitative)

The pre-task survey demonstrates that for a p-value of 0.951, both sets of participants (MRCS and non-MRCS condition participants) have similar perceived skills, with an average rating of approximately 3.80 out of 10, and similar standard deviation, when asked to evaluate their skillset before the Human Study task execution, as shown in Table 6. This ensures that both sets of study participants start at the same level of pre-task execution, as they have similar skill set ratings.

Table 6

Descriptive statistics: pre-task-execution
Type	N	Mean	StDev
MRCS	20	3.80	2.55
Non-MRCS	20	3.75	2.59

In addition, the study participants provided an additional confidence rating with surgical skill planning and task execution, with over 90% identifying as incompetent with the task at hand, or any for that matter, as shown in Fig. 10. Only 10% demonstrated a somewhat competency with surgical skill planning and task execution. This helps confirm that they have no medical training. No study participant indicated they were either Competent and/or Very Competent.

The study participants' average incision depths were calculated based on the two study populations. Despite the target depth being 2.5mm for both sets of study participants, both groups could not achieve it as an average of their attempts. Despite this, the MRCS condition participants demonstrate an average incision depth of 3.304 mm, which is lower than the average incision depth of 4.349 mm for the non-MRCS condition participants, demonstrating the MRCS system's effectiveness in improving the performance of study participants (Table 7). The two averages are statistically significant, as their p-value is less than 0.05. In addition, the standard deviation for the MRCS study participants, at 0.474, is lower than that for the non-MRCS study participants, at 0.528. This indicates that the MRCS study participants' incision depth data points are much less spread around their mean and more accurate than those of the non-MRCS study participants.

Table 7

Performance metrics for human subject study participants
	Non-MRCS	MRCS
Sample Size	20	20
Target Depth (mm)	2.5	2.5
Average Incision Depth (mm) Effectiveness)	4.349	3.304
St Dev (Incision Depth)	0.528	0.474
SE Mean	0.084	0.075
Minimum Zero Difference (Proficiency Gain)	N/A	N/A¹

iMinimum Zero Point is not reached.

Proficiency gain, the point at which a near zero-sum difference between the target and actual values is achieved and maintained, could not be seen in either group before the end of the 40-attempt task execution. This is likely because the participants lack prior surgical experience, and the task iterations are limited to 40 attempts.

The data is then analyzed as an average of the difference at each iteration for all study participants within their study participant groups. The data is plotted with the target value of 2.5mm as a negative as the study participants cut a vertical depth into a simulated tissue. it is also seen from Fig. 11 that the MRCS condition participants started much lower in the difference between their task execution incision depths at approximately 1.7 mm from the target value of 2.5 mm as compared to the 3.2 mm difference for the non-MRCS condition participants (Fig. 11). The slope in Fig. 11 shows that the vertical incision depth for both groups is trending towards the desired target zero difference. In addition, the slope for the non-MRCS group is steeper at 0.0168 than the MRCS group at 0.0117, indicating that the non-MRCS condition participants learn faster despite starting much higher in task execution.

The intercept for the trendline for the MRCS condition group is also lower at -3.5432 mm compared to the non-MRCS condition group at -4.6926 mm (Table 8), showing that they start at an incision depth whose difference is much closer to the target value of 2.5 mm.

Table 8

Trendline for MRCS and non-MRCS condition participants
	Non-MRCS	MRCS
Slope	0.0168	0.0117
Intercept	-4.6926	-3.5432

The study participants who used the MRCS were far more likely to stay within their group average, as evidenced by collating their data points around the group mean as the iteration count increased. Non-MRCS condition participants performed much worse over time, with their group mean of 4.46 mm higher than the MRCS condition participant group mean of 3.38 mm. It can also be seen that the non-MRCS condition participants failed to reach that grouping within the two standard errors. The collation of the incision depth points within two standard errors of the group means for the MRCS condition participants simply translates to the ability of the MRCS platform to keep these study participants within their task execution range much more consistent as compared to the non-MRCS condition participants whose performance is unable to be within the two standard errors as they continued to perform the task.

To better understand the study participants' self-perception of performance, their attribute responses are analyzed to determine their confidence and task execution satisfaction. They are also asked to self-rate their confidence and task execution performance (Table 9). Despite participants displaying low confidence (Fig. 10) in their skillset and incompetence before the study task, both groups positively perceived their performance after the study, as shown in Table 9. The non-MRCS showed a much higher confidence in Task Execution Satisfaction despite not performing well, as they had much larger average incision depths and failed to reach task proficiency. The MRCS study participants scored lower (50%) for their task execution satisfaction as they could obtain real-time feedback on their task execution performance due to the MRCS guided aids. In addition, nearly 90% of the MRCS condition participants highlighted using the MRCS as an effective aid in achieving their tasks compared to 70% of the non-MRCS condition participants.

Table 9

Survey of study participants' self-evaluation
Description	Non-MRCS	MRCS
User Surgical Task Execution Satisfaction (Post-Test Satisfaction)	70%	50%
User Assessment of Effectiveness of Test Setup (medium of helpfulness)	70%	90%

The study participants also improved their self-confidence in task planning and execution from the training. The MRCS condition participants showed remarkable task execution confidence from the training exercise, from 5% before the human study to a 65% approval rating from learning in that environment to become somewhat competent in task planning and execution, as shown in Fig. 12.

The non-MRCS study participants did not see a greater than 50% change in any category, with only a 10% change moving into a somewhat competent confidence in their task execution confidence, as shown in Fig. 13.

To better understand the training effectiveness among the participants, they completed a rating scale of 1–10 to quantify their responses on study participation and how practical the study had been with the results displayed in Table 10. In this survey, the MRCS study participants felt that the study had been more valuable in helping them make an effective hernia surgical plan than the non-MRCS participants. They noted correctly identifying and planning their route for that procedure and recognizing the different anatomy around the target site. The study also positively impacted their understanding of task planning and execution compared to the non-MRCS study participants.

A 2-sample t-test compares the means of the two groups' self-perceived performance rating pre- and post-task execution, as shown in Table 10. With a p-value threshold of 0.05, neither group of participants had a statistical significance of their self-perceived performance pre- and post-task execution nor between each other at post-task execution, as shown in Table 10. The p-value for the MRCS condition for the pre-and post-task execution comparison is 0.097, and the p-value for the non-MRCS pre- and post-task execution is 0.234.

Since these (both with and without MRCS) are new environments for these participants, they have difficulty perceiving their improvement, despite the MRCS participants acknowledging in the post-task completion surveys that the MRCS specifically helped them much better than the non-MRCS participants, as shown in Fig. 12 and Fig. 13.

It can also be seen from Table 10 that the lack of significance between the two groups, MRCS and non-MRCS, when the participants are asked to self-evaluate their performance in the human study, could signal that they are both not better yet when the actual performance, in terms of vertical depth incision, between the two groups is compared, the MRCS Study group outperformed the non-MRCS study group as shown in Table 7.

Table 10

Study participants' self-evaluation ratings of pre-and post-task execution (TE) performance
Sample	N	Mean	StDev	P-value
Pre_TE_MRCS	20	3.80	2.55	0.097
Post_TE_MRCS	20	5.20	2.65	0.097
Pre_TE_NoMRCS	20	3.75	2.59	0.234
Post_TE_NoMRCS	20	4.65	2.08	0.234
Post_TE_MRCS	20	5.20	2.65	0.470
Post_TE_NoMRCS	20	4.65	2.08	0.470

The study participants are then surveyed, post-task execution, for the perceived workload. Using the raw unweighted scores, the MRCS study participants were higher than the non-MRCS study participants in five of the six sub-scales and overall workload (Fig. 14). This is likely due to the headset's weight and the time required of the MRCS study participants to walk through the tutorial and further execute the task. Despite this additional workload, they could still outperform their non-MRCS study counterparts. The MRCS study participants only tested lower on the temporal workload as they did not feel the task required much time to master and execute compared to the non-MRCS study participants.

There is no statistical significance between the two groups of study participants in perceived mental, physical, and frustration workload when a threshold p-value of 0.05 is considered, as shown in Table 11. However, the p-values of Temporal, Performance, and Effort are lower than the threshold of 0.05 at 0.027, 0.046, and 0.021, respectively. The difference in the Performance and Effort workload contribution by the MRCS may be attributed to the time required to master and perfect their technique, as the stream of images requires these study participants to engage more with the system. However, the temporal workload required to complete the task is higher for the non-MRCS as they need more guidance in expectation of the task. They have no image in front of their task interaction guiding them on the outcome of their task.

Table 11

2 sample t-test for perceived workload between the two study groups of graduate engineering Students for Human study.
Variable	Student type	N	Mean	Standard Deviation	P-value
Mental	MRCS	20	31.3	21.1	0.729
Mental	Non-MRCS	20	28.8	24.1	0.729
Physical	MRCS	20	24.5	19.6	0.708
Physical	Non-MRCS	20	22.0	22.3	0.708
Temporal	MRCS	20	13.3	16.8	0.027
Temporal	Non-MRCS	20	26.3	18.8	0.027
Performance	MRCS	20	43.8	26.2	0.046
Performance	Non-MRCS	20	29.0	18.0	0.046
Effort	MRCS	20	41.3	19.7	0.021
Effort	Non-MRCS	20	25.8	21.0	0.021
Frustration	MRCS	20	25.5	24.6	0.238
Frustration	Non-MRCS	20	17.0	20.0	0.238

3.2. Human Subjects Study II – Medical Residents (Qualitative)

Seven Medical residents are asked to perform the same study as the MRCS condition with the same constraints and time. Their responses regarding using the MRCS are recorded and analyzed to capture their feedback correctly. The first level of feedback is from the medical resident's interaction with the 3D bioprint Specimen. The responsiveness to the user action of incision is well received, as well as the ability of the post-treatment of the bioprint to reflect diseased tissue, making it hard for the medical residents to know when to stop at the precise depth point at the start of the test execution. This is reflected in their user vertical incision depth performance as, on average, their depth is beyond 2.5mm at 4.5mm for four of the seven residents. According to their feedback, three of the seven residents in the initial response exceeded and cut through the total height of the 3D Bioprint specimen as they seemed to underestimate the simulated diseased tissue's tactile response before understanding its texture and adjusting.

The medical residents also supported the image manipulation tutorials and anatomy identification portion of the task execution exercise. They highlighted the training as critically instrumental in their thought process on planning and approaching the surgical site for task execution. The medical residents also expressed concern for the motion allowed within the virtual image, as their surgical technique execution required specific access to the surgical site. Moving forward, any user-activated virtual image interaction would need additional definitions of multiple degrees of freedom beyond translation, resizing, and rotating that the current virtual image entails. The medical residents also expressed concern about the contrast of the virtual image with the environment, making it harder to visualize at certain angles. This could be due to the lighting in the room, but this is a potential area for improvement in the future.

Regarding using the MRCS, the medical residents responded positively when asked to gauge their confidence in using it, as shown in Fig. 15. Of the seven medical residents, only one indicated they are still very incompetent when it came to understanding how to surgically plan and execute the specific human study procedure, as opposed to three before the start of the human study. The most significant jump in confidence in the surgical plan and surgical execution happened when at least five medical residents evaluated themselves as having been at least somewhat competent after interaction with the MRCS. The only student who had marked themselves as very competent before the study is now downgraded to a somewhat competent mindset. This specific downgrade of a self-assessment can be considered as a user understanding that they might not be as informed regarding surgical planning and execution as they initially thought, as they had overstated their previous knowledge base. Interaction with the MRCS training portion pre-task execution could have led to this potential self-realization of their limited skill set.

A comparison of the Post TE confidence rating between the non-medical graduate engineering students of Study 1 and medical residents in this study shows no statistical significance between the two groups at a p-value of 0.571 (Table 12). Both groups were confident in having executed the task using the MRCS platform.

Table 12

Medical resident and graduate engineering student self-evaluation ratings of post-task execution (TE) performance.
Study participant	N	Mean	Standard Deviation	P-Value
Post-TE graduate engineering student	20	5.20	2.65	0.571
Post-TE medical resident	7	4.43	3.10	0.571

The workload rating by the medical residents highlighted that the effort and frustration required to perform the surgical task is relevant and is reflected in the Workload ratings as shown in Fig. 16. This is similar to the engineering graduate students, as both sets required time to acclimate to the MRCS platform's user interface when targeting the 2.5mm vertical incision.

Table 13

2 Sample T-Test NASA TLX workload comparison between medical residents and graduate engineering students using the MRCS.
Variable	Type	N	Mean	StDev	P-Value
Mental	Graduate engineering student	20	31.3	21.1	0.868
Mental	Medical resident	7	30.0	15.0	0.868
Physical	Graduate engineering student	20	24.5	19.6	0.175
Physical	Medical resident	7	14.3	14.8	0.175
Temporal	Graduate engineering student	20	13.3	16.8	0.870
Temporal	Medical resident	7	12.1	14.4	0.870
Performance	Graduate engineering student	20	43.8	26.2	0.815
Performance	Medical resident	7	41.4	20.6	0.815
Effort	Graduate engineering student	20	41.3	19.7	0.521
Effort	Medical resident	7	47.9	23.4	0.521
Frustration	Graduate engineering student	20	25.5	24.6	0.019
Frustration	Medical resident	7	51.4	20.8	0.019

The temporal workload is highest for non-MRCS engineering graduate students. This is probably because they had to rely on their intuition to target a vertical incision depth and had no marker to tell them exactly when to stop. With the task's iterative nature, this realization could have led them to believe they felt rushed and hurried to perform the job. The MRCS participants (Graduate and Medical residents) benefitted from the MRCS, letting them know exactly when to stop. Still, the frustration with using the MRCS is highlighted in the comparison with the graduate students, as shown in Table 13, as the p-value is 0.019 and is statistically significant. This may be because the environment differs from what they expect or experience. There is no statistical significance between these two groups in the other five types of workloads when engaged with the MRCS.

3.2.1. Proctor Evaluation

The medical residents demonstrated familiarity with the patient positioning aspect of the exercise. They knew how and when to identify areas needing more proper guidance and instruction for learners. Despite the additional fidelity required in these scenarios, the current setup is adequate for personnel with limited exposure to medical training as a learning and training tool to become proficient.

The students can assess the surgical site and area for proper access and, despite the virtual imagery's confines, can also place the virtual specimen in the correct position for entry. Assessing the surgical site for entry denotes the student physically engaging with the site to ensure they have a visual and tactile feel of the tissue response. This assessment happened throughout the surgical site setup and task execution. However, the students did not augment their technique to ensure that they were accurate in their incision depth, as it can be seen from the video review that they had not achieved dexterous tissue manipulation. This can be further rectified with additional engagement with the MRCS, as the iterative exposure can undoubtedly lead to decreased timeline for skill acquisition.

From the data analysis, proficiency is achieved based on the difference in average depths between the participants who used the MRCS and those who did not. This is made possible because of the guided path that the AR imagery plays in their training to become proficient, as shown in the results section, with their much lower standard deviation and more accurate results closer to the target value (Fig. 11). This can be said to help them learn better as the non-MRCS condition participants relied heavily on not only the training they received from the instructor with the measurement aids but also on their intuition on what a proper incision depth should be. It is possible that the MRCS condition participants failed to meet the absolute minimum zero difference point due to the current setup that did not allow them to go beyond the 40 iterations in the task to a recommended 80 iterations. Increasing engagement with the MRCS beyond the 40 iterations and within the targeted timeframe could strengthen its effect, especially considering that the first study participants had no formal medical training before the study. It could be regarded that novices without medical training need a probable "test period" when starting a task.

The effectiveness of the MRCS platform as a training tool is presented in two formats. The first format is the MRCS condition participants, determining it is an effective tool for learning about the task and its associated penalties in real-time. This is reflected in their self-perceived acknowledgment of improved competence regarding that specific task and how helpful they found the MRCS in achieving the assigned task. The second effectiveness format came from the interaction with the MRCS by medical residents who understand human anatomy and pathologies. Their positive response towards the bio-printed specimen as a simulated tissue, whose deformation is matched by the virtual imagery as guidance, could improve their understanding and task execution despite their additional requests for refinement around the virtual image manipulation.

Several responses from both study group participants suggest improving the setup, as there are limitations. The headset component of the MRCS platform could be physically lighter to lessen the extra workload required from the participants. The MRCS Platform in the medical resident study displayed contrast image issues, making it difficult to see the image response to the simulated tissue interaction at certain angles.

The medical residents rated the overall workload higher than the novice MRCS graduate students. The non-MRCS graduate students had the lowest overall mental workload among these three groups. Despite this higher rating of the perceived workload by the medical residents, it does not impact their performance; this rating could be just perception.

The medical residents explicitly mentioned frustration with the additional effort required initially to perform the task. This does not negate the platform as a training tool but raises the opportunity to address comfort and ease of use in adapting to these new learning environments.

This work demonstrates the ability of the Mixed Reality Combination System (MRCS) as a training system to not only guide a user's navigation as they pre-plan the task execution through image visualization and interaction but also track the task execution to quantify their skill set in achieving task completion and gaining performance improvement.

The MRCS uses the updated tracking information from the user's interaction with the bioprint for haptic feedback during task execution to provide guided imagery based on the user's updated position during this task execution. This overall interaction provides an iterative learning platform that a user needs to achieve the desired proficiency, as evidenced by the difference in performance between those using and those not using the platform. This demonstration of improved user performance could be translated to similar benefits in other fields where a user must achieve proficiency through guided training like construction, agriculture, healthcare, and robotics.

Acknowledgment

This work is partially supported by the Office of Naval Research (ONR) under grant N00014-17-1- 2566. The authors thank Anatomage Inc. for supplying their virtual cadaver models for use in the MRCS platform.

Author contributions statement

E.K, M.S, J.C, and P.L conceived the experiment(s),

E.K, J.B, and A.H built the software models and physical materials required for the experiment and study participants.

E.K conducted the study and wrote the manuscript.

E.K., M.S., J.C., and P.L. analyzed the results.

All authors reviewed the manuscript.

ORCID IDs

Andrew Hudson https://orcid.org/0000-0002-4148-5708

Jennifer Bone https://orcid.org/0000-0001-9170-3235

Adam Feinberg https://orcid.org/0000-0003-3338-5456

Ernest Kabuye https://orcid.org/0000-0003-0314-9946

Jonathan Cagan https://orcid.org/0000-0002-3935-9219

Philip LeDuc https://orcid.org/0000-0002-5342-8567

Data availability statement

The data acquired during this study are available from the corresponding author upon reasonable request. ([email protected])

Competing interests

The authors declare no competing interests.

Agius A, Calleja N, Camenzuli C, et al (2018) Perceptions of first-year medical students towards learning anatomy using cadaveric specimens through peer teaching. Anat Sci Educ 11:346–357. https://doi.org/10.1002/ase.1751
Ammar SA (2009) Management of Giant Ventral Hernia by Polypropylene Mesh and Host Tissue Barrier: Trial of Simplification. J Clin Med Res 1:226. https://doi.org/10.4021/JOCMR2009.10.1268
Anderson JG, Abrahamson K (2017) Your Health Care May Kill You: Medical Errors. Stud Health Technol Inform 234:13–17
Armstrong JPK, Burke M, Carter BM, et al (2016) 3D Bioprinting Using a Templated Porous Bioink. Adv Healthc Mater 5:1724–30. https://doi.org/10.1002/adhm.201600022
Asencio F, Carbó J, Ferri R, et al (2021) Laparoscopic Versus Open Incisional Hernia Repair: Long-Term Follow-up Results of a Randomized Clinical Trial. World J Surg 45:. https://doi.org/10.1007/s00268-021-06164-7
Balogun JA, Adekanmbi A, Balogun FM (2019) Recognition and Disclosure of Medical Errors Among Residents in Surgical Specialties in a Tertiary Hospital in Ibadan. World J Surg 43:717–722. https://doi.org/10.1007/s00268-018-4836-y
Balta JY, Lamb C, Soames RW (2015) A pilot study comparing the use of Thiel- and formalin-embalmed cadavers in the teaching of human anatomy. Anat Sci Educ 8:86–91. https://doi.org/10.1002/ase.1470
Bari A, Khan RA, Rathore AW (2016) Medical errors; causes, consequences, emotional response and resulting behavioral change. Pak J Med Sci 32:. https://doi.org/10.12669/pjms.323.9701
Barsom EZ, Graafland • M, Schijven • M P (2016) Systematic review on the effectiveness of augmented reality applications in medical training. Surg Endosc 30:. https://doi.org/10.1007/s00464-016-4800-6
Bing EG, Parham GP, Cuevas A, et al (2019) Using Low-Cost Virtual Reality Simulation to Build Surgical Capacity for Cervical Cancer Treatment. J Glob Oncol 2019:1–7. https://doi.org/10.1200/JGO.18.00263
Blasier RB (2009) The Problem of the Aging Surgeon: When Surgeon Age Becomes a Surgical Risk Factor. Clin Orthop Relat Res 467:402–411. https://doi.org/10.1007/s11999-008-0587-7
Bowyer MW, Kuhls DA, Haskin D, et al (2013) Advanced surgical skills for exposure in trauma (ASSET): The first 25 courses. Journal of Surgical Research 183:553–558. https://doi.org/10.1016/j.jss.2013.02.005
Chen X, Xu L, Wang Y, et al (2015) Development of a surgical navigation system based on augmented reality using an optical see-through head-mounted display. J Biomed Inform 55:124–31. https://doi.org/10.1016/j.jbi.2015.04.003
Choi Y-B, Lee I-S (2018) Incisional and Ventral Hernia Repair. The Journal of Minimally Invasive Surgery 21:5–12. https://doi.org/10.7602/jmis.2018.21.1.5
Chuah YJ, Peck Y, Lau JEJ, et al (2017) Hydrogel based cartilaginous tissue regeneration: recent insights and technologies. Biomater Sci 5:613–631. https://doi.org/10.1039/C6BM00863A
COLEMAN R, KOGAN I (1998) An improved low-formaldehyde embalming fluid to preserve cadavers for anatomy teaching. J Anat 192:443–446. https://doi.org/10.1046/j.1469-7580.1998.19230443.x
Datta R, Chon SH, Dratsch T, et al (2020) Are gamers better laparoscopic surgeons? Impact of gaming skills on laparoscopic performance in “Generation Y” students. PLoS One 15:. https://doi.org/10.1371/journal.pone.0232341
Encinosa WE, Hellinger FJ (2008) The impact of medical errors on ninety-day costs and outcomes: An examination of surgical patients. Health Serv Res 43:2067–2085. https://doi.org/10.1111/j.1475-6773.2008.00882.x
Fried MP, Kaye RJ, Gibber MJ, et al (2012) Criterion-Based (Proficiency) Training to Improve Surgical Performance. Arch Otolaryngol Head Neck Surg 138:1024. https://doi.org/10.1001/2013.jamaoto.377
Gabrysz-Forget F, Young M, Zahabi S, et al (2020) Surgical Errors Happen, but Are Learners Trained to Recover from Them? A Survey of North American Surgical Residents and Fellows. J Surg Educ 77:1552–1561. https://doi.org/10.1016/j.jsurg.2020.05.031
Ghosh SK (2017) Cadaveric dissection as an educational tool for anatomical sciences in the 21st century. Anat Sci Educ 10:286–299. https://doi.org/10.1002/ase.1649
Giannotti D, Patrizi G, Di Rocco G, et al (2013) Play to Become a Surgeon: Impact of Nintendo WII Training on Laparoscopic Skills. PLoS One 8:. https://doi.org/10.1371/journal.pone.0057372
Gottlieb-Vedi E, Mackenzie H, van Workum F, et al (2019) Surgeon Volume and Surgeon Age in Relation to Proficiency Gain Curves for Prognosis Following Surgery for Esophageal Cancer. Ann Surg Oncol 26:497–505. https://doi.org/10.1245/s10434-018-6869-8
Gupta A, Lawendy B, Goldenberg MG, et al (2021) Can video games enhance surgical skills acquisition for medical students? A systematic review. Surgery (United States) 169
HAMMOND JW, QUEALE WS, KIM TK, MCFARLAND EG (2003) SURGEON EXPERIENCE AND CLINICAL AND ECONOMIC OUTCOMES FOR SHOULDER ARTHROPLASTY. The Journal of Bone and Joint Surgery-American Volume 85:2318–2324. https://doi.org/10.2106/00004623-200312000-00008
Hart SG (2006) Nasa-Task Load Index (NASA-TLX); 20 Years Later. Proceedings of the Human Factors and Ergonomics Society Annual Meeting 50:904–908. https://doi.org/10.1177/154193120605000909
Hart SG, Staveland LE (1988) Development of NASA-TLX (Task Load Index): Results of Empirical and Theoretical Research. Advances in Psychology 52:. https://doi.org/10.1016/S0166-4115(08)62386-9
Hettig J, Engelhardt S, Hansen C, Mistelbauer G (2018) AR in VR: assessing surgical augmented reality visualizations in a steerable virtual reality environment. Int J Comput Assist Radiol Surg 13:1717–1725. https://doi.org/10.1007/s11548-018-1825-4
Hoonakker P, Carayon P, Gurses AP, et al (2011) Measuring workload of ICU nurses with a questionnaire survey: the NASA Task Load Index (TLX). IIE Trans Healthc Syst Eng 1:131–143. https://doi.org/10.1080/19488300.2011.609524
Kabuye E, Hellebrekers T, Bobo J, et al (2022) Tracking of Scalpel Motions With an Inertial Measurement Unit System. IEEE Sens J 22:4651–4660. https://doi.org/10.1109/JSEN.2022.3145312
Kabuye E, LeDuc P, Cagan J (2023) A mixed reality system combining augmented reality, 3D bio-printed physical environments and inertial measurement unit sensors for task planning. Virtual Real 1:1–14. https://doi.org/10.1007/s10055-023-00777-0
Kanevsky J, Safran T, Zammit D, et al (2019) Making Augmented and Virtual Reality Work for the Plastic Surgeon. Ann Plast Surg 82:363–368. https://doi.org/10.1097/SAP.0000000000001594
Kennel L, Martin DMA, Shaw H, Wilkinson T (2018) Learning anatomy through Thiel- vs. formalin-embalmed cadavers: Student perceptions of embalming methods and effect on functional anatomy knowledge. Anat Sci Educ 11:166–174. https://doi.org/10.1002/ase.1715
Kim DH, Kim Y, Park JS, Kim SW (2019) Virtual reality simulators for endoscopic sinus and skull base surgery: The present and future. Clin Exp Otorhinolaryngol 12
Kowalczyk KA, Majewski A (2021) Analysis of surgical errors associated with anatomical variations clinically relevant in general surgery. Review of the literature. Translational Research in Anatomy 23:100107. https://doi.org/10.1016/j.tria.2020.100107
Kuhls DA, Risucci DA, Bowyer MW, Luchette FA (2013) Advanced surgical skills for exposure in trauma: a new surgical skills cadaver course for surgery residents and fellows. J Trauma Acute Care Surg 74:664–70. https://doi.org/10.1097/TA.0b013e31827d5e20
Lam CK, Sundaraj K, Sulaiman MN (2014) Computer-based virtual reality simulator for phacoemulsification cataract surgery training. Virtual Real 18:281–293. https://doi.org/10.1007/S10055-014-0251-3/TABLES/2
Landers R (2015) Reducing Surgical Errors: Implementing a Three-Hinge Approach to Success. AORN J 101:657–665. https://doi.org/10.1016/J.AORN.2015.04.013
Lee A, Hudson AR, Shiwarski DJ, et al (2019) 3D bioprinting of collagen to rebuild components of the human heart. Science (1979) 365:482–487. https://doi.org/10.1126/science.aav9051
Lee ARY Bin, Tan JHI, Ng HJH, Rajaratnam V (2021) The ageing surgeon workforce: A qualitative study of interprofessional perspectives. Int J Healthc Manag 14:. https://doi.org/10.1080/20479700.2020.1815294
Mackenzie H, Miskovic D, Ni M, et al (2013) Clinical and educational proficiency gain of supervised laparoscopic colorectal surgical trainees. Surg Endosc 27:2704–2711. https://doi.org/10.1007/S00464-013-2806-X
Makary MA, Daniel M (2016) Medical error-the third leading cause of death in the US. BMJ 353:i2139. https://doi.org/10.1136/bmj.i2139
Markar SR, Ni M, Mackenzie H, et al (2020) The effect of time between procedures upon the proficiency gain period for minimally invasive esophagectomy. Surg Endosc 34:. https://doi.org/10.1007/s00464-019-06692-3
Mirdamadi E, Tashman JW, Shiwarski DJ, et al (2020) FRESH 3D Bioprinting a Full-Size Model of the Human Heart. ACS Biomater Sci Eng 6:6453–6459. https://doi.org/10.1021/acsbiomaterials.0c01133
Misiakos EP, Patapis P, Zavras N, et al (2015) Current trends in laparoscopic ventral hernia repair. Journal of the Society of Laparoendoscopic Surgeons 19
M.W. VE, J.J. G, P. S, L.A. S (2018) Comparison of a novel cadaver model (Fix for Life) with the formalin-fixed cadaver and manikin model for suitability and realism in airway management training. Anesth Analg 127:
Noor N, Shapira A, Edri R, et al (2019) 3D Printing of Personalized Thick and Perfusable Cardiac Patches and Hearts. Advanced Science 6:1900344. https://doi.org/10.1002/advs.201900344
Novoa NM, Varela G (2020) Monitoring surgical quality: the cumulative sum (CUSUM) approach. Mediastinum 4:4–4. https://doi.org/10.21037/med.2019.10.01
Pham JC, Aswani MS, Rosen M, et al (2012) Reducing medical errors and adverse events. Annu Rev Med 63:447–63. https://doi.org/10.1146/annurev-med-061410-121352
Phillips AW, Dent B, Navidi M, Griffin SM (2016) Surgical Proficiency in the Era of Centralization. J Clin Oncol 34:3939–3940. https://doi.org/10.1200/JCO.2016.68.1817
Pratt P, Ives M, Lawton G, et al (2018) Through the HoloLens^TM looking glass: augmented reality for extremity reconstruction surgery using 3D vascular models with perforating vessels. Eur Radiol Exp 2:2. https://doi.org/10.1186/s41747-017-0033-2
Rastogi P, Kandasubramanian B (2019) Review of alginate-based hydrogel bioprinting for application in tissue engineering. Biofabrication 11:042001. https://doi.org/10.1088/1758-5090/ab331e
Robinson DA, Piekut DT, Hasman L, Knight PA (2020) Cadaveric Simulation Training in Cardiothoracic Surgery: A Systematic Review. Anat Sci Educ 13:413–425. https://doi.org/10.1002/ase.1908
Sahoo DR, Biswal T (2021) Alginate and its application to tissue engineering. SN Appl Sci 3:1–19. https://doi.org/10.1007/S42452-020-04096-W/FIGURES/5
Sakai D, Joyce K, Sugimoto M, et al (2020) Augmented, virtual and mixed reality in spinal surgery: A real-world experience. Journal of Orthopaedic Surgery 28:230949902095269. https://doi.org/10.1177/2309499020952698
Sánchez-Margallo JA, Plaza de Miguel C, Fernández Anzules RA, Sánchez-Margallo FM (2021) Application of Mixed Reality in Medical Training and Surgical Planning Focused on Minimally Invasive Surgery. Front Virtual Real 2:144. https://doi.org/10.3389/frvir.2021.692641
Satkunasivam R, Klaassen Z, Ravi B, et al (2020) Relation between surgeon age and postoperative outcomes: A population-based cohort study. CMAJ 192:. https://doi.org/10.1503/cmaj.190820
Scott DJ (2008) Proficiency-Based Training for Surgical Skills. Semin Colon Rectal Surg 19:72–80. https://doi.org/10.1053/j.scrs.2008.02.003
Selcuk İ, Tatar I, Huri E (2019) Cadaveric anatomy and dissection in surgical training. Turk J Obstet Gynecol 16:72–75. https://doi.org/10.4274/tjod.galenos.2018.15931
Sheikh AH, Barry DS, Gutierrez H, et al (2016) Cadaveric anatomy in the future of medical education: What is the surgeons view? Anat Sci Educ 9:203–208. https://doi.org/10.1002/ase.1560
Stapenhorst F, Garrido dos Santos M, Prestes JP, et al (2021) Bioprinting: A promising approach for tissue regeneration. Bioprinting 22:. https://doi.org/10.1016/J.BPRINT.2021.E00130
Sylos Labini M, Gsaxner C, Pepe A, et al (2019) Depth-Awareness in a System for Mixed-Reality Aided Surgical Procedures. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 11645 LNAI:716–726. https://doi.org/10.1007/978-3-030-26766-7_65
Thunnissen FM, Schreurs BW, Latenstein CSS, et al (2021) Evaluation of a shared decision-making strategy with online decision aids in surgical and orthopaedic practice: study protocol for the E-valuAID, a multicentre study with a stepped-wedge design. BMC Med Inform Decis Mak 21:110. https://doi.org/10.1186/s12911-021-01467-0
Van Den Bos J, Rustagi K, Gray T, et al (2011) The $17.1 billion problem: the annual cost of measurable medical errors. Health Aff (Millwood) 30:596–603. https://doi.org/10.1377/hlthaff.2011.0084
Van Der Meijden OAJ, Schijven MP (2009) The value of haptic feedback in conventional and robot-assisted minimal invasive surgery and virtual reality training: A current review. Surg Endosc
van Workum F, Fransen L, Luyer MD, Rosman C (2018) Learning curves in minimally invasive esophagectomy. World J Gastroenterol 24:4974–4978. https://doi.org/10.3748/wjg.v24.i44.4974
Våpenstad C, Hofstad EF, Langø T, et al (2013) Perceiving haptic feedback in virtual reality simulators. Surg Endosc 27:2391–2397. https://doi.org/10.1007/s00464-012-2745-y
Vles MD, Terng NCO, Zijlstra K, et al (2020) Virtual and augmented reality for preoperative planning in plastic surgical procedures: A systematic review. Journal of Plastic, Reconstructive & Aesthetic Surgery 73:1951–1959. https://doi.org/10.1016/j.bjps.2020.05.081
Waerlop F, Rashidian N, Marrannes S, et al (2021) Thiel embalmed human cadavers in surgical education: Optimizing realism and long-term application. Am J Surg 221
Wan L, Skoko J, Yu J, et al (2017) Mimicking Embedded Vasculature Structure for 3D Cancer on a Chip Approaches through Micromilling. Scientific Reports 2017 7:1 7:1–8. https://doi.org/10.1038/s41598-017-16458-3
Weininger G, Mori M, Brooks C, et al (2020) Association Between Cardiac Surgeons’ Number of Years in Practice and Surgical Outcomes in New York Cardiac Centers. JAMA Netw Open 3:e2023671. https://doi.org/10.1001/jamanetworkopen.2020.23671
Wentink M, Stassen LPS, Alwayn I, et al (2003) Rasmussen’s model of human behavior in laparoscopy training. Surgical Endoscopy and Other Interventional Techniques 17:1241–1246. https://doi.org/10.1007/S00464-002-9140-Z/FIGURES/3
Yiasemidou M, Gkaragkani E, Glassman D, Biyani CS (2018) Cadaveric simulation: a review of reviews. Irish Journal of Medical Science (1971 -) 187:827–833. https://doi.org/10.1007/s11845-017-1704-y
Yu JZ, Korkmaz E, Berg MI, et al (2017) Biomimetic scaffolds with three-dimensional undulated microtopographies. Biomaterials 128:109–120. https://doi.org/10.1016/j.biomaterials.2017.02.014
Zheng K, Du D (2021) Recent advances of hydrogel-based biomaterials for intervertebral disc tissue treatment: A literature review. J Tissue Eng Regen Med 15:299–321. https://doi.org/10.1002/term.3172
Hand–eye coordination: video games and laparoscopic surgery. Nat Clin Pract Urol 2:268–268. https://doi.org/10.1038/ncpuro0202

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Surgical Training Outcomes Using a Mixed Reality Combination System

Status:

Version 1

Abstract

Figures

1. Introduction

2. Study Methodologies

2.1. Human Subjects Study Participants Setup

2.2. Human Study Surgical Training and Task Execution Design

2.2.1. User Proficiency

2.2.2. User Effectiveness

2.2.3. Human Subjects Study Task Training Setup

3. Results

3.1. Human Subjects Study I – Engineering Graduate Students (Quantitative)

3.2. Human Subjects Study II – Medical Residents (Qualitative)

3.2.1. Proctor Evaluation

4. Discussion

5. Conclusion

Declarations

References

Additional Declarations

Status:

Version 1