Acta Neurochir DOI 10.1007/s00701-013-1918-3
TECHNICAL NOTE - NEUROSURGERY TRAINING
A neurosurgical phantom-based training system with ultrasound simulation Andrea Müns & Constanze Mühl & Robert Haase & Hendrik Möckel & Claire Chalopin & Jürgen Meixensberger & Dirk Lindner
Received: 12 July 2013 / Accepted: 9 October 2013 # Springer-Verlag Wien 2013
Abstract Background Brain tumor surgeries are associated with a high technical and personal effort. The required interactions between the surgeon and the technical components, such as neuronavigation, surgical instruments and intraoperative imaging, are complex and demand innovative training solutions and standardized evaluation methods. Phantombased training systems could be useful in complementing the existing surgical education and training. Methods A prototype of a phantom-based training system was developed, intended for standardized training of important aspects of brain tumor surgery based on real patient data. The head phantom consists of a three-part construction that includes a reusable base and adapter, as well as a changeable module for single use. Training covers surgical planning of the optimal access path, the setup of the navigation system including the registration of the head phantom, as well as the navigated craniotomy with real instruments. Tracked instruments during the simulation and predefined access paths constitute the basis for the essential objective training feedback. Results The prototype was evaluated in a pilot study by assistant physicians at different education levels. They performed a complete simulation and a final assessment using an evaluation questionnaire. The analysis of the questionnaire showed the evaluation result as “good” for the phantom A. Müns (*) : C. Mühl : J. Meixensberger : D. Lindner Department of Neurosurgery, University Hospital Leipzig, Liebigstraße 20, 04103 Leipzig, Germany e-mail: [email protected] R. Haase : H. Möckel PHACON GmbH, Leipzig, Germany C. Chalopin : J. Meixensberger ICCAS, University of Leipzig, Leipzig, Germany
construction and the used materials. The learning effect concerning the navigated planning was evaluated as “very good”, as well as having the effect of increasing safety for the surgeon before planning and conducting craniotomies independently on patients. Conclusions The training system represents a promising approach for the future training of neurosurgeons. It aims to improve surgical skill training by creating a more realistic simulation in a non-risk environment. Hence, it could help to bridge the gap between theoretical and practical training with the potential to benefit both physicians and patients. Keywords Neurosurgical training . Head phantom . Tumor resection . Ultrasound simulation . Ultrasound phantom
Introduction Background During brain tumor surgeries, various imaging modalities, such as magnetic resonance imaging (MRI) or intraoperative ultrasound (iUS), neuronavigation systems and microscopes need to be coordinated for effective use and interpretation. The interactions are complex and should ideally be learned during a standardized training that can be verified objectively. Mistakes can have serious consequences, and teaching during surgery results in longer operating times and may increase the overall risk to the patient [3]. Surgical simulation and skill training offer an opportunity to teach and practice in a non-risk environment where surgeons can develop and refine skills through harmless repetition [12]. There is enormous potential to address patient safety, risk management concerns, operating room management and work hour requirements with more efficient and effective training methods [1]. Surgical organizations are calling for
Acta Neurochir
methods to ensure the maintenance of skills, advance surgical training, and credential surgeons as technically competent [3]. State of the art training systems Nowadays, a navigation system is part of the standard equipment in neurosurgery. The suitability for training purposes is limited since the system cannot give a feedback to the planned intervention and a standardized failure analysis is hardly possible. Patient positioning, incision, craniotomy and trepanation can neither be measured nor evaluated. The state of the art simulation systems in the neurosurgical field are virtual reality-based systems [16], which use force feedback [10, 11, 18, 19], partly in combination with augmented reality [2]. The literature currently describes models that assist in procedure planning, augment the visual-spatial learning of complex surgical approaches and simulate technical components of neurosurgical procedures [9, 12, 13]. Simulation environments combine a graphic interface with a graphic display and the corresponding software, such as Dextroscope [8], Cranial Base Surgery Simulators [4, 20], ROBO-SIM [14, 15] and ImmersiveTouch [11]. Undoubtedly virtual reality-based simulators make a major contribution to the training of future neurosurgeons. Nevertheless, such systems are currently limited by the computational complexity of accurate tissue deformation, the arduous process of manually segmenting volume-rendered models, and the great expense of sophisticated haptic interfaces [12]. The restricted hand-eye coordination makes the training more unrealistic. Further present concepts in neurosurgical training are live surgeries [17] or training on animal cadavers [5]. Investigating the scientific literature, no reference could be found describing a standardized phantom for neurosurgical training that allows a preoperative planning for different tumor locations and the simulation of the intervention based on patient data. Another challenging task for resident neurosurgeons is the handling and mental compounding of ultrasound images during neurosurgical procedures. Their safe application and interpretation require a lot of training and experience, which could be partly gained from an ultrasound simulation tool.
Hardware The head phantom shows a quite realistic representation of a human skull with frontotemporal muscle, skin and dura. The three-part construction is based on a reusable system for the automatic recognition of modules and risk structures. The second part provides a reusable specific head adapter (Fig. 1) that carries the corresponding frontotemporal module designed for single use only (Fig. 2). The design process of the phantom was based on the MNI152 brain template created by the Montreal neurological institute (MNI), which represents the average of 152 different MRI scans [6]. 3D-construction datasets were created for bone, skin and muscle, segmented from the brain template (Fig. 3). The bone structure and the molds for the soft tissue structures were produced using rapid prototyping methods. Skin, dura and brain were modeled by different silicon materials, chosen in material studies performed to find the optimal materials according to structure, haptics, color and tear strength. During the simulation, all instruments are tracked by a twocamera system to improve the visibility and enlarge the covered tracking area. The hardware setup of the training system allows the user to use real instruments for drilling and milling. Datasets To increase the flexibility, students can train on different tumor locations in the frontotemporal region with the same phantom. During the simulation, a visualization of the patient’s dataset
Material and methods Idea The objective of our study was to create a neurosurgical phantom-based training system for the purpose of teaching techniques for the planning of tumor resections and the realization of the corresponding craniotomy. Further requirements included a standardized evaluation method for measuring the progress by learning curves and an integrated ultrasound simulation tool, based on patient data.
Fig. 1 Neurosurgical training system, consisting of a three-part head phantom (base system, adapter and changeable module), tracking cameras and connected laptop with installed navigation software
Acta Neurochir Fig. 2 Module for the frontotemporal access, consisting of a bony structure, skin, frontotemporal muscle and dura. a outside view; b inside view
with respect to the head phantom is required. To satisfy this requirement, a method was developed that integrates real patient datasets into the predefined structures of the MNI template. Besides suitable tumor localization, an adequate patient dataset consists of good quality MRI and iUS datasets. The navigation system (SonoNavigator, LOCALITE, St. Augustin, Germany) provides the matrices to transform both datasets into the same spatial coordinates. The patient’s MRI dataset is registered by a linear and a non-linear algorithm from the FSL package (FMRIB software library) [7] to the MNI template. The computed transformations can also be applied to the iUS dataset since it is defined by the same coordinate system (Fig. 3). A predefined mask separates the brain structure from the remaining structures in the MRI dataset before merging the inner part of the patient’s dataset with the outer part of the template. The border between both datasets remains visible at closer inspection, but is completely sufficient for training purposes (Fig. 4). The big advantage of this method is that the head phantom must not be modified at all because creating new cases is just a question of image processing. At the same time, anonymity of the patient can be guaranteed. Software and simulation The simulation starts with choosing a dataset depending on the type of access to be practiced. Based on this dataset, the entry and target have to be defined.
The next steps include phantom positioning, adjustment of the tracking cameras and the marker-based phantom registration. The registration will be accepted only if the target registration error does not exeed 1 mm (threshold is adjustable). A second registration method, based on anatomical landmarks, will be implemented to make the registration process more realistic. The subsequent skin incision, as well as the preparation of the frontotemporal muscle are performed using a tracked surgical knife. An Aesculap System (Microspeed uni, Aesculap AG, Tuttlingen, Germany) is used for tracked drilling and milling, while milling can also be accompanied by realistic flushing. Before the bone flap can be removed, the dura needs to be separated through the drilling holes. The current development of the software comprises the detection of injuries of certain nerves and arteries during the opening process, as well as the implementation of the evaluation concept. For each dataset, a predefined master access path, including entry, target, trajectory, skin incision line, drilling holes and trepanation line, provides the basis for the automatic evaluation of the simulation. Since the tracking cameras are able to record the instrument handling during the simulation, the conducted steps can be related to the planning as well as to the master access path. The current integration of the ultrasound simulation tool realizes the identification of the tumor in a final ultrasound examination, which can only be successful with a correct placed craniotomy. The ultrasound
Fig. 3 Patient’s MRI dataset with superimposed iUS dataset before applying the computed transformation to register on MNI template
Acta Neurochir Fig. 4 Merged dataset constructed from the patient’s dataset and the MNI template. Bone, skin and frontotemporal muscle were segmented from the template and are highlighted. The inner part of the MR Dataset was taken from the patient’s dataset and deformed in different registration steps
simulation is based on the patient's intraoperatively acquired iUS dataset, which has also been transformed to fit on the merged MRI dataset. During the simulated acquisition of iUS, a tracked ultrasound dummy probe needs to be swept across the phantom's brain surface. Simultaneously, the corresponding slides from the transformed patient's iUS dataset are computed and visualized, which gives the impression of a real ultrasound examination, on the condition that the craniotomy was placed adequately.
Finally, a questionnaire was filled out by the participants containing questions related to the used materials, concerning structure, haptics, color, tear strength, removability and cutting sensation. In a second part, questions with regard to the ergonomic comfort and learning effect had to be answered. Every point was evaluated with a metric scale between one and five, where one indicated 'very good'. At the end, personal comments, additions and suggestions for improvements were welcomed.
Pilot study
Results
The prototype was evaluated in a small clinical setup. Five residents between the second and seventh year (average 4.8 years, standard deviation 2.28) were asked to perform the simulation. None of them were involved in the development process. The handling of the training system was explained in a short introduction and no specifications were made concerning methods or directions for accessing the tumor. In order to establish comparability, all simulations were performed with the same dataset characterized by a metastasis localized in the temporal lobe. The required time was measured for every single step and the simulation was finished upon skin suture.
The craniotomy was accomplished successfully in all simulations. The average time of the simulation was 23.4 minutes (sd (standard deviation) 7.9 min). The secondyear resident took the longest time, with 37.2 minutes. We assume a correlation between the level of specialist training and time needed for the simulation, but the case number is, of course, too small for a statistical proof. Skin opening was carried out via a curved incision in two simulations and by a linear incision in three simulations. The frontotemporal muscle was injured in three cases during the skin incision. Two drilling holes were placed in all simulations. The size of the bone flap varied between approximately 3 and 6 cm in diameter. The dura was injured
Acta Neurochir Table 1 Questionnaire, part one: evaluation of materials used to construct the head phantom
Overall impression Phantom Skin
Bone
Avg n=5
Std Dev n=5
Color Width haptics
2.20 2.20 1.40 3.40
0.45 1.10 0.55 0.55
Cutting Sensation Tear strength Removability from bone Adhesive residues bone Removability from muscle Adhesive residues muscle Skin suture TOTAL average Color Haptics Authenticity drilling Authenticity milling TOTAL average
3.40 2.20 2.00 1.60 2.50 2.00 2.60 2.33 1.40 1.40 1.20 1.20 1.3
1.14 0.45 1.22 0.89 0.58 0.71 0.89 0.81 0.55 0.55 0.45 0.45 0.5
Muscle
Dura
Avg n=5
Std Dev n=5
Color Structure Width
1.40 2.80 2.20
0.55 0.84 0.84
Haptics Tear strength Removability from bone Adhesive residues bone TOTAL average Color Structure Haptics Width Tear strength Removability from bone Adhesive residues bone TOTAL average
3.00 2.60 2.20 2.00 2.31 2.80 3.00 3.00 2.80 3.20 2.20 1.20 2.60
1.22 0.55 1.64 0.71 0.90 1.30 1.22 0.71 1.48 0.84 1.10 0.45 1.01
Structure, haptics, color, tear strength, removability and cutting sensation had to be evaluated by five test persons with a metric scale between 1 and 5, where 1 indicates 'very good'. The table shows average and standard deviation
in two cases. It was clearly recognizable that the finesse demonstrated in performing each of these single steps increased with the level of education. For the simulated structures of bone, muscle, skin and dura, different parameters (such as structure, haptics, color, tear strength, removability and cutting sensation) were assessed (Table 1). The first overall impression of the head phantom was rated with average of 2.2 (sd 0.45). The rating for the bone structure showed the best results, with an average of 1.3 (sd 0.5). The poorest rating obtained was for the dura, with an average of 2.6 (sd 1).
The ergonomics and learning effects were assessed in the second part of the questionnaire and showed very good results (Table 2). The assessment of the increase of safety before conducting the first interventions independently on patients was rated with an average of 1.0 (sd 0). Few critical comments were made related to the locking mechanism and the stability of the ball joint construction for fixing the head phantom in a plastic tray. Each simulation was supervised by the same specialist for neurosurgery. At the end of the training, he made an assessment including a highlighting of possible errors and improvement opportunities.
Table 2 Questionnaire, part two: evaluation of ergonomic comfort and learning effect by all five test persons and a metric scale from 1 to 5, where 1 indicates 'very good' Avg Std Dev n=5 n=5 Ergonomics related to Patient positioning Visualization in software Covered area tracking cameras Handling pointer Handling craniotome Handling trepan TOTAL average The table shows average and standard deviation
1.60 1.00 2.00 1.20 1.40 1.40 1.43
0.55 0.00 0.71 0.45 0.89 0.89 0.58
Avg Std Dev n=5 n=5 Learning effect related to Planning with navigation system Patient positioning Navigation system Incision line Craniotomy Trepanation TOTAL average
1.20 2.6 1.00 2.00 1.60 2.00 1.73
0.45 0.55 0.00 1.73 0.89 0.71 0.72
Acta Neurochir
Discussion This proof of concept study identified strengths and weaknesses of the training systems, and the results of the questionnaire look very promising. The combination of theory and practice creates the opportunity to teach and practice neurosurgical procedures outside of the operating room, but nevertheless in a quite realistic environment. An important question to ask is whether human performance can be improved through the use of a neurosurgical training environment and whether that improvement can be measured [3]. Therefore, a larger group of residents must statistically examine the efficacy of the training system. A further extended study has to validate the learning curves, which can be achieved by repetitive exercises and a standardized evaluation method to measure the expected improvement. Nevertheless, it could be shown that the feedback from all test participants was very positive related to the convenience in specialist training for neurosurgery. As described in the Introduction, the current state of the art level training in the neurosurgical field mainly takes place on virtual reality-based systems, during live surgeries and in training on animal cadavers. The advantages of phantom-based training systems compared with virtual reality-based systems include the realistic tactile head phantom, which can be used with real instruments, the hand-to-eye-coordination during the simulation, as well as the probable lower investment costs. On the other hand, the limitation to the single use of the changeable modules results in higher operational costs and leads to greater strain of material resources. However, it should not be considered as a competing product with virtual training systems, but rather as complementary system that may close the gap between training on a virtual-based training system and training on patients. Current and future development will focus on improving and extending the functioning of the proposed training system. Further development is required, especially in implementing the ultrasound simulation tool and the suggested evaluation concept. The locking mechanism was already improved, while the challenge of a more stable solution for the ball joint still needs to be addressed. Future development incorporates the construction of two further modules for the occipital and the parietal region to extend the possibilities for the training of different access paths. A simulation of risk structures, such as important nerves and blood vessels, will be integrated with the aim to notice a corresponding injury and include it in the training evaluation. In general, the further development of phantom-based training systems may have the potential to improve surgical education in order to address risk management concerns, patient safety and operating room management by more effective training methods.
Acknowledgment The described project was co-financed by the European Union under the European Regional Development Fund (EFRE, project number 14220/2466), while PHACON GmbH (Leipzig, Germany) was involved as a project partner. All authors declare no financial or personal conflict of interest regarding the material discussed in the article. Conflicts of interest None.
References 1. Aggarwal R, Darzi A (2005) Organising a surgical skills centre. Minim Invasive Ther Allied Technol 14(4):275–279 2. Alaraj A, Charbel FT, Birk D, Tobin M, Luciano C, Banerjee PP, Rizzi S, Sorenson J, Foley K, Slavin K, Roitberg B (2013) Role of cranial and spinal virtual and augmented reality simulation using immersive touch modules in neurosurgical training. Neurosurgery 72(l):A115–A123 3. Alaraj A, Lemole MG, Finkle JH, Yudkowsky R, Wallace A, Luciano C, Banerjee PP, Rizzi SH, Charbel FT (2011) Virtual reality training in neurosurgery: review of current status and future applications. Surg Neurol Int 2:52 4. Bernardo A, Preul MC, Zabramski JM, Spetzler RF (2003) A threedimensional interactive virtual dissection model to simulate transpetrous surgical avenues. Neurosurgery 52(3):499–505 5. Csókay A, Papp A, Imreh D, Czabajszky M, Valálik I, Antalfi B (2013) Modelling pathology from autolog fresh cadaver organs as a novel concept in neurosurgical training. Acta Neurochir PMID: 23624637 6. Evans AC, Janke AL, Collins DL, Baillet S (2012) Brain templates and atlases. Neuroimage 62(2):911–922 7. Jenkinson M, Beckmann CF, Behrens TEJ, Woolrich MW, Smith SM (2011) Fsl. Neuroimage 62(2):782–790 8. Kockro RA, Serra L, Tseng-Tsai Y, Chan C, Yih-Yian S, Gim-Guan C, Lee E, Hoe LY, Hern N, Nowinski WL (2000) Planning and simulation of neurosurgery in a virtual reality environment. Neurosurgery 46(1):118–135 9. Kockro RA, Stadie A, Schwandt E, Reisch R, Charalampaki C, Ng I, Yeo TT, Hwang P, Serra L, Perneczky A (2007) A collaborative virtual reality environment for neurosurgical planning and training. Neurosurgery 61(5 Suppl 2):379–391 10. Lemole GM Jr, Banerjee PP, Luciano C, Neckrysh S, Charbel FT (2007) Virtual reality in neurosurgical education: part-task ventriculostomy simulation with dynamic visual and haptic feedback. Neurosurgery 61(1):142–148 11. Luciano C, Banerjee P, Lemole GM Jr, Charbel F (2006) Second generation haptic ventriculostomy simulator using the immersivetouch system. Stud Health Technol Inf 119:343–348 12. Malone HR, Syed ON, Downes MS, D'Ambrosio AL, Quest DO, Kaiser MG (2010) Simulation in neurosurgery: a review of computerbased simulation environments and their surgical applications. Neurosurgery 67(4):1105–1116 13. Radetzky A, Rudolph M (2001) Simulating tumour removal in neurosurgery. Int J Med Inform 64:461–472 14. Radetzky A, Rudolph M, Starkie S, Davies B, Auer LM (2000) Robo-sim: a simulator for minimally invasive neurosurgery using an active manipulator. Stud Health Technol Inf 77:1165–1169 15. Radetzky A, Schröcker F, Auer LM (2000) Improvement of surgical simulation using dynamic volume rendering. Stud Health Technol Inf 70:272–278 16. Robison RA, Liu CY, Apuzzo ML (2011) Man, mind, and machine: the past and future of virtual reality simulation in neurologic surgery. World Neurosurg 76(5):419–430
Acta Neurochir 17. Roser F, Pfister G, Tatagiba M, Ebner FH (2013) Live surgery in neurosurgical training courses: essential infrastructure and technical set-up. Acta Neurochir. doi:10.1007/s00701-012-1578-8 18. Sato D, Kobayashi R, Kobayashi A, Fujino S, Uchiyama M (2006) Soft tissue pushing operation using a haptic interface for simulation of brain tumor resection. J Robot Mechatron 18(5):634–642
19. Spicer MA, van Velsen MV, Caffrey JP, Apuzzo ML (2004) Virtual reality neurosurgery: a simulator blueprint. Neurosurgery 54(4):783–797 20. Stredney D, Wiet GJ, Bryan J, Sessanna D, Murakami J, Schmalbrock P, Powell K, Welling B (2002) Temporal bone dissection simulation— an update. Stud Health Technol Inf 85:507–513
TECHNICAL NOTE - NEUROSURGERY TRAINING
A neurosurgical phantom-based training system with ultrasound simulation Andrea Müns & Constanze Mühl & Robert Haase & Hendrik Möckel & Claire Chalopin & Jürgen Meixensberger & Dirk Lindner
Received: 12 July 2013 / Accepted: 9 October 2013 # Springer-Verlag Wien 2013
Abstract Background Brain tumor surgeries are associated with a high technical and personal effort. The required interactions between the surgeon and the technical components, such as neuronavigation, surgical instruments and intraoperative imaging, are complex and demand innovative training solutions and standardized evaluation methods. Phantombased training systems could be useful in complementing the existing surgical education and training. Methods A prototype of a phantom-based training system was developed, intended for standardized training of important aspects of brain tumor surgery based on real patient data. The head phantom consists of a three-part construction that includes a reusable base and adapter, as well as a changeable module for single use. Training covers surgical planning of the optimal access path, the setup of the navigation system including the registration of the head phantom, as well as the navigated craniotomy with real instruments. Tracked instruments during the simulation and predefined access paths constitute the basis for the essential objective training feedback. Results The prototype was evaluated in a pilot study by assistant physicians at different education levels. They performed a complete simulation and a final assessment using an evaluation questionnaire. The analysis of the questionnaire showed the evaluation result as “good” for the phantom A. Müns (*) : C. Mühl : J. Meixensberger : D. Lindner Department of Neurosurgery, University Hospital Leipzig, Liebigstraße 20, 04103 Leipzig, Germany e-mail: [email protected] R. Haase : H. Möckel PHACON GmbH, Leipzig, Germany C. Chalopin : J. Meixensberger ICCAS, University of Leipzig, Leipzig, Germany
construction and the used materials. The learning effect concerning the navigated planning was evaluated as “very good”, as well as having the effect of increasing safety for the surgeon before planning and conducting craniotomies independently on patients. Conclusions The training system represents a promising approach for the future training of neurosurgeons. It aims to improve surgical skill training by creating a more realistic simulation in a non-risk environment. Hence, it could help to bridge the gap between theoretical and practical training with the potential to benefit both physicians and patients. Keywords Neurosurgical training . Head phantom . Tumor resection . Ultrasound simulation . Ultrasound phantom
Introduction Background During brain tumor surgeries, various imaging modalities, such as magnetic resonance imaging (MRI) or intraoperative ultrasound (iUS), neuronavigation systems and microscopes need to be coordinated for effective use and interpretation. The interactions are complex and should ideally be learned during a standardized training that can be verified objectively. Mistakes can have serious consequences, and teaching during surgery results in longer operating times and may increase the overall risk to the patient [3]. Surgical simulation and skill training offer an opportunity to teach and practice in a non-risk environment where surgeons can develop and refine skills through harmless repetition [12]. There is enormous potential to address patient safety, risk management concerns, operating room management and work hour requirements with more efficient and effective training methods [1]. Surgical organizations are calling for
Acta Neurochir
methods to ensure the maintenance of skills, advance surgical training, and credential surgeons as technically competent [3]. State of the art training systems Nowadays, a navigation system is part of the standard equipment in neurosurgery. The suitability for training purposes is limited since the system cannot give a feedback to the planned intervention and a standardized failure analysis is hardly possible. Patient positioning, incision, craniotomy and trepanation can neither be measured nor evaluated. The state of the art simulation systems in the neurosurgical field are virtual reality-based systems [16], which use force feedback [10, 11, 18, 19], partly in combination with augmented reality [2]. The literature currently describes models that assist in procedure planning, augment the visual-spatial learning of complex surgical approaches and simulate technical components of neurosurgical procedures [9, 12, 13]. Simulation environments combine a graphic interface with a graphic display and the corresponding software, such as Dextroscope [8], Cranial Base Surgery Simulators [4, 20], ROBO-SIM [14, 15] and ImmersiveTouch [11]. Undoubtedly virtual reality-based simulators make a major contribution to the training of future neurosurgeons. Nevertheless, such systems are currently limited by the computational complexity of accurate tissue deformation, the arduous process of manually segmenting volume-rendered models, and the great expense of sophisticated haptic interfaces [12]. The restricted hand-eye coordination makes the training more unrealistic. Further present concepts in neurosurgical training are live surgeries [17] or training on animal cadavers [5]. Investigating the scientific literature, no reference could be found describing a standardized phantom for neurosurgical training that allows a preoperative planning for different tumor locations and the simulation of the intervention based on patient data. Another challenging task for resident neurosurgeons is the handling and mental compounding of ultrasound images during neurosurgical procedures. Their safe application and interpretation require a lot of training and experience, which could be partly gained from an ultrasound simulation tool.
Hardware The head phantom shows a quite realistic representation of a human skull with frontotemporal muscle, skin and dura. The three-part construction is based on a reusable system for the automatic recognition of modules and risk structures. The second part provides a reusable specific head adapter (Fig. 1) that carries the corresponding frontotemporal module designed for single use only (Fig. 2). The design process of the phantom was based on the MNI152 brain template created by the Montreal neurological institute (MNI), which represents the average of 152 different MRI scans [6]. 3D-construction datasets were created for bone, skin and muscle, segmented from the brain template (Fig. 3). The bone structure and the molds for the soft tissue structures were produced using rapid prototyping methods. Skin, dura and brain were modeled by different silicon materials, chosen in material studies performed to find the optimal materials according to structure, haptics, color and tear strength. During the simulation, all instruments are tracked by a twocamera system to improve the visibility and enlarge the covered tracking area. The hardware setup of the training system allows the user to use real instruments for drilling and milling. Datasets To increase the flexibility, students can train on different tumor locations in the frontotemporal region with the same phantom. During the simulation, a visualization of the patient’s dataset
Material and methods Idea The objective of our study was to create a neurosurgical phantom-based training system for the purpose of teaching techniques for the planning of tumor resections and the realization of the corresponding craniotomy. Further requirements included a standardized evaluation method for measuring the progress by learning curves and an integrated ultrasound simulation tool, based on patient data.
Fig. 1 Neurosurgical training system, consisting of a three-part head phantom (base system, adapter and changeable module), tracking cameras and connected laptop with installed navigation software
Acta Neurochir Fig. 2 Module for the frontotemporal access, consisting of a bony structure, skin, frontotemporal muscle and dura. a outside view; b inside view
with respect to the head phantom is required. To satisfy this requirement, a method was developed that integrates real patient datasets into the predefined structures of the MNI template. Besides suitable tumor localization, an adequate patient dataset consists of good quality MRI and iUS datasets. The navigation system (SonoNavigator, LOCALITE, St. Augustin, Germany) provides the matrices to transform both datasets into the same spatial coordinates. The patient’s MRI dataset is registered by a linear and a non-linear algorithm from the FSL package (FMRIB software library) [7] to the MNI template. The computed transformations can also be applied to the iUS dataset since it is defined by the same coordinate system (Fig. 3). A predefined mask separates the brain structure from the remaining structures in the MRI dataset before merging the inner part of the patient’s dataset with the outer part of the template. The border between both datasets remains visible at closer inspection, but is completely sufficient for training purposes (Fig. 4). The big advantage of this method is that the head phantom must not be modified at all because creating new cases is just a question of image processing. At the same time, anonymity of the patient can be guaranteed. Software and simulation The simulation starts with choosing a dataset depending on the type of access to be practiced. Based on this dataset, the entry and target have to be defined.
The next steps include phantom positioning, adjustment of the tracking cameras and the marker-based phantom registration. The registration will be accepted only if the target registration error does not exeed 1 mm (threshold is adjustable). A second registration method, based on anatomical landmarks, will be implemented to make the registration process more realistic. The subsequent skin incision, as well as the preparation of the frontotemporal muscle are performed using a tracked surgical knife. An Aesculap System (Microspeed uni, Aesculap AG, Tuttlingen, Germany) is used for tracked drilling and milling, while milling can also be accompanied by realistic flushing. Before the bone flap can be removed, the dura needs to be separated through the drilling holes. The current development of the software comprises the detection of injuries of certain nerves and arteries during the opening process, as well as the implementation of the evaluation concept. For each dataset, a predefined master access path, including entry, target, trajectory, skin incision line, drilling holes and trepanation line, provides the basis for the automatic evaluation of the simulation. Since the tracking cameras are able to record the instrument handling during the simulation, the conducted steps can be related to the planning as well as to the master access path. The current integration of the ultrasound simulation tool realizes the identification of the tumor in a final ultrasound examination, which can only be successful with a correct placed craniotomy. The ultrasound
Fig. 3 Patient’s MRI dataset with superimposed iUS dataset before applying the computed transformation to register on MNI template
Acta Neurochir Fig. 4 Merged dataset constructed from the patient’s dataset and the MNI template. Bone, skin and frontotemporal muscle were segmented from the template and are highlighted. The inner part of the MR Dataset was taken from the patient’s dataset and deformed in different registration steps
simulation is based on the patient's intraoperatively acquired iUS dataset, which has also been transformed to fit on the merged MRI dataset. During the simulated acquisition of iUS, a tracked ultrasound dummy probe needs to be swept across the phantom's brain surface. Simultaneously, the corresponding slides from the transformed patient's iUS dataset are computed and visualized, which gives the impression of a real ultrasound examination, on the condition that the craniotomy was placed adequately.
Finally, a questionnaire was filled out by the participants containing questions related to the used materials, concerning structure, haptics, color, tear strength, removability and cutting sensation. In a second part, questions with regard to the ergonomic comfort and learning effect had to be answered. Every point was evaluated with a metric scale between one and five, where one indicated 'very good'. At the end, personal comments, additions and suggestions for improvements were welcomed.
Pilot study
Results
The prototype was evaluated in a small clinical setup. Five residents between the second and seventh year (average 4.8 years, standard deviation 2.28) were asked to perform the simulation. None of them were involved in the development process. The handling of the training system was explained in a short introduction and no specifications were made concerning methods or directions for accessing the tumor. In order to establish comparability, all simulations were performed with the same dataset characterized by a metastasis localized in the temporal lobe. The required time was measured for every single step and the simulation was finished upon skin suture.
The craniotomy was accomplished successfully in all simulations. The average time of the simulation was 23.4 minutes (sd (standard deviation) 7.9 min). The secondyear resident took the longest time, with 37.2 minutes. We assume a correlation between the level of specialist training and time needed for the simulation, but the case number is, of course, too small for a statistical proof. Skin opening was carried out via a curved incision in two simulations and by a linear incision in three simulations. The frontotemporal muscle was injured in three cases during the skin incision. Two drilling holes were placed in all simulations. The size of the bone flap varied between approximately 3 and 6 cm in diameter. The dura was injured
Acta Neurochir Table 1 Questionnaire, part one: evaluation of materials used to construct the head phantom
Overall impression Phantom Skin
Bone
Avg n=5
Std Dev n=5
Color Width haptics
2.20 2.20 1.40 3.40
0.45 1.10 0.55 0.55
Cutting Sensation Tear strength Removability from bone Adhesive residues bone Removability from muscle Adhesive residues muscle Skin suture TOTAL average Color Haptics Authenticity drilling Authenticity milling TOTAL average
3.40 2.20 2.00 1.60 2.50 2.00 2.60 2.33 1.40 1.40 1.20 1.20 1.3
1.14 0.45 1.22 0.89 0.58 0.71 0.89 0.81 0.55 0.55 0.45 0.45 0.5
Muscle
Dura
Avg n=5
Std Dev n=5
Color Structure Width
1.40 2.80 2.20
0.55 0.84 0.84
Haptics Tear strength Removability from bone Adhesive residues bone TOTAL average Color Structure Haptics Width Tear strength Removability from bone Adhesive residues bone TOTAL average
3.00 2.60 2.20 2.00 2.31 2.80 3.00 3.00 2.80 3.20 2.20 1.20 2.60
1.22 0.55 1.64 0.71 0.90 1.30 1.22 0.71 1.48 0.84 1.10 0.45 1.01
Structure, haptics, color, tear strength, removability and cutting sensation had to be evaluated by five test persons with a metric scale between 1 and 5, where 1 indicates 'very good'. The table shows average and standard deviation
in two cases. It was clearly recognizable that the finesse demonstrated in performing each of these single steps increased with the level of education. For the simulated structures of bone, muscle, skin and dura, different parameters (such as structure, haptics, color, tear strength, removability and cutting sensation) were assessed (Table 1). The first overall impression of the head phantom was rated with average of 2.2 (sd 0.45). The rating for the bone structure showed the best results, with an average of 1.3 (sd 0.5). The poorest rating obtained was for the dura, with an average of 2.6 (sd 1).
The ergonomics and learning effects were assessed in the second part of the questionnaire and showed very good results (Table 2). The assessment of the increase of safety before conducting the first interventions independently on patients was rated with an average of 1.0 (sd 0). Few critical comments were made related to the locking mechanism and the stability of the ball joint construction for fixing the head phantom in a plastic tray. Each simulation was supervised by the same specialist for neurosurgery. At the end of the training, he made an assessment including a highlighting of possible errors and improvement opportunities.
Table 2 Questionnaire, part two: evaluation of ergonomic comfort and learning effect by all five test persons and a metric scale from 1 to 5, where 1 indicates 'very good' Avg Std Dev n=5 n=5 Ergonomics related to Patient positioning Visualization in software Covered area tracking cameras Handling pointer Handling craniotome Handling trepan TOTAL average The table shows average and standard deviation
1.60 1.00 2.00 1.20 1.40 1.40 1.43
0.55 0.00 0.71 0.45 0.89 0.89 0.58
Avg Std Dev n=5 n=5 Learning effect related to Planning with navigation system Patient positioning Navigation system Incision line Craniotomy Trepanation TOTAL average
1.20 2.6 1.00 2.00 1.60 2.00 1.73
0.45 0.55 0.00 1.73 0.89 0.71 0.72
Acta Neurochir
Discussion This proof of concept study identified strengths and weaknesses of the training systems, and the results of the questionnaire look very promising. The combination of theory and practice creates the opportunity to teach and practice neurosurgical procedures outside of the operating room, but nevertheless in a quite realistic environment. An important question to ask is whether human performance can be improved through the use of a neurosurgical training environment and whether that improvement can be measured [3]. Therefore, a larger group of residents must statistically examine the efficacy of the training system. A further extended study has to validate the learning curves, which can be achieved by repetitive exercises and a standardized evaluation method to measure the expected improvement. Nevertheless, it could be shown that the feedback from all test participants was very positive related to the convenience in specialist training for neurosurgery. As described in the Introduction, the current state of the art level training in the neurosurgical field mainly takes place on virtual reality-based systems, during live surgeries and in training on animal cadavers. The advantages of phantom-based training systems compared with virtual reality-based systems include the realistic tactile head phantom, which can be used with real instruments, the hand-to-eye-coordination during the simulation, as well as the probable lower investment costs. On the other hand, the limitation to the single use of the changeable modules results in higher operational costs and leads to greater strain of material resources. However, it should not be considered as a competing product with virtual training systems, but rather as complementary system that may close the gap between training on a virtual-based training system and training on patients. Current and future development will focus on improving and extending the functioning of the proposed training system. Further development is required, especially in implementing the ultrasound simulation tool and the suggested evaluation concept. The locking mechanism was already improved, while the challenge of a more stable solution for the ball joint still needs to be addressed. Future development incorporates the construction of two further modules for the occipital and the parietal region to extend the possibilities for the training of different access paths. A simulation of risk structures, such as important nerves and blood vessels, will be integrated with the aim to notice a corresponding injury and include it in the training evaluation. In general, the further development of phantom-based training systems may have the potential to improve surgical education in order to address risk management concerns, patient safety and operating room management by more effective training methods.
Acknowledgment The described project was co-financed by the European Union under the European Regional Development Fund (EFRE, project number 14220/2466), while PHACON GmbH (Leipzig, Germany) was involved as a project partner. All authors declare no financial or personal conflict of interest regarding the material discussed in the article. Conflicts of interest None.
References 1. Aggarwal R, Darzi A (2005) Organising a surgical skills centre. Minim Invasive Ther Allied Technol 14(4):275–279 2. Alaraj A, Charbel FT, Birk D, Tobin M, Luciano C, Banerjee PP, Rizzi S, Sorenson J, Foley K, Slavin K, Roitberg B (2013) Role of cranial and spinal virtual and augmented reality simulation using immersive touch modules in neurosurgical training. Neurosurgery 72(l):A115–A123 3. Alaraj A, Lemole MG, Finkle JH, Yudkowsky R, Wallace A, Luciano C, Banerjee PP, Rizzi SH, Charbel FT (2011) Virtual reality training in neurosurgery: review of current status and future applications. Surg Neurol Int 2:52 4. Bernardo A, Preul MC, Zabramski JM, Spetzler RF (2003) A threedimensional interactive virtual dissection model to simulate transpetrous surgical avenues. Neurosurgery 52(3):499–505 5. Csókay A, Papp A, Imreh D, Czabajszky M, Valálik I, Antalfi B (2013) Modelling pathology from autolog fresh cadaver organs as a novel concept in neurosurgical training. Acta Neurochir PMID: 23624637 6. Evans AC, Janke AL, Collins DL, Baillet S (2012) Brain templates and atlases. Neuroimage 62(2):911–922 7. Jenkinson M, Beckmann CF, Behrens TEJ, Woolrich MW, Smith SM (2011) Fsl. Neuroimage 62(2):782–790 8. Kockro RA, Serra L, Tseng-Tsai Y, Chan C, Yih-Yian S, Gim-Guan C, Lee E, Hoe LY, Hern N, Nowinski WL (2000) Planning and simulation of neurosurgery in a virtual reality environment. Neurosurgery 46(1):118–135 9. Kockro RA, Stadie A, Schwandt E, Reisch R, Charalampaki C, Ng I, Yeo TT, Hwang P, Serra L, Perneczky A (2007) A collaborative virtual reality environment for neurosurgical planning and training. Neurosurgery 61(5 Suppl 2):379–391 10. Lemole GM Jr, Banerjee PP, Luciano C, Neckrysh S, Charbel FT (2007) Virtual reality in neurosurgical education: part-task ventriculostomy simulation with dynamic visual and haptic feedback. Neurosurgery 61(1):142–148 11. Luciano C, Banerjee P, Lemole GM Jr, Charbel F (2006) Second generation haptic ventriculostomy simulator using the immersivetouch system. Stud Health Technol Inf 119:343–348 12. Malone HR, Syed ON, Downes MS, D'Ambrosio AL, Quest DO, Kaiser MG (2010) Simulation in neurosurgery: a review of computerbased simulation environments and their surgical applications. Neurosurgery 67(4):1105–1116 13. Radetzky A, Rudolph M (2001) Simulating tumour removal in neurosurgery. Int J Med Inform 64:461–472 14. Radetzky A, Rudolph M, Starkie S, Davies B, Auer LM (2000) Robo-sim: a simulator for minimally invasive neurosurgery using an active manipulator. Stud Health Technol Inf 77:1165–1169 15. Radetzky A, Schröcker F, Auer LM (2000) Improvement of surgical simulation using dynamic volume rendering. Stud Health Technol Inf 70:272–278 16. Robison RA, Liu CY, Apuzzo ML (2011) Man, mind, and machine: the past and future of virtual reality simulation in neurologic surgery. World Neurosurg 76(5):419–430
Acta Neurochir 17. Roser F, Pfister G, Tatagiba M, Ebner FH (2013) Live surgery in neurosurgical training courses: essential infrastructure and technical set-up. Acta Neurochir. doi:10.1007/s00701-012-1578-8 18. Sato D, Kobayashi R, Kobayashi A, Fujino S, Uchiyama M (2006) Soft tissue pushing operation using a haptic interface for simulation of brain tumor resection. J Robot Mechatron 18(5):634–642
19. Spicer MA, van Velsen MV, Caffrey JP, Apuzzo ML (2004) Virtual reality neurosurgery: a simulator blueprint. Neurosurgery 54(4):783–797 20. Stredney D, Wiet GJ, Bryan J, Sessanna D, Murakami J, Schmalbrock P, Powell K, Welling B (2002) Temporal bone dissection simulation— an update. Stud Health Technol Inf 85:507–513
