I. Introduction... 7
I.1. Essentials of laparoscopic surgery... 8
I.2. Psychomotor challenges... 9
I.3. Skills to be a proficient surgeon... 11
I.4. Evolution of surgical education... 12
I.5. Objective assessment methods... 16
I.5.1. Structured rating systems for objective evaluation... 16
I.5.2. Instrument motion analysis... 20
I.5.3. A good assessment tool... 22
I.6. Problem statement and aim of this thesis... 23
I.7. Thesis outline... 26
II. Research hypotheses and objectives... 27
II.1. Research hypotheses... 28
II.1.1. First instrument motion analysis method... 28
II.1.2. Second instrument motion analysis method... 28
II.1.3. Third instrument motion analysis method... 28
II.2. Objectives... 29
II.2.1. General objective... 29
II.2.2. Specific objectives... 29
III. State of the art... 31
III.1. Introduction... 32
III.2. Surgical simulators... 32
III.2.1. Physical simulators or box trainers... 32
III.2.2. Virtual simulators... 37
III.2.3. Hybrid simulators... 40
III.3. Objective assessment methods based on motion analysis... 41
III.3.1. Extracorporeal methods... 43
III.3.2. Intracorporeal methods... 48
IV. Laparoscopic hybrid simulator with stereoscopic tracking system... 55
IV.1. Introduction... 56
IV.2. Material and methods... 58
IV.2.1. System description... 58
IV.2.2. Stereo vision system... 59
IV.2.3. Laparoscopic instrument tracking... 64
IV.2.4. Technical validation... 65
IV.3. Results... 67
IV.3.1. Relative position error... 68
IV.3.2. Accumulated distance error... 68
IV.3.3. Face validation... 69
VI.4. Discussion... 70
VI.4.1. Chapter conclusions... 74
V. Optical pose tracker for assessment of MIS technical skills... 75
V.1. Introduction... 76
V.2. Technical validation... 78
V.2.1. First design... 78
V.2.2. Second design... 83
V.3. Validation for MIS skills assessment... 88
V.3.1. Experimental validation... 89
V.3.2. MIS skills assessment... 91
V.3.3. Statistical analysis... 93
V.3.4. Results... 94
V.4. Installing the motion assessment system inside an OR setting... 103
V.4.1. Analysis of the position of the camera system... 103
V.4.2. Feasibility study for the use of the motion analysis method in a OR setting... 104
V.5. Discussion... 106
V.5.1. Chapter conclusions... 113
VI. Video-based instrument tracking method... 115
VI.1. Introduction... 116
VI.2. Materials... 118
VI.3. Laparoscopic instrument tracking method... 119
VI.3.1. Materials and methods... 120
VI.3.2. Results... 126
VI.4. Method to insert multimedia support content... 128
VI.4.1. Technical validation... 132
VI.4.2. Results... 133
VI.5. Discussion... 135
VI.5.1. Chapter conclusions... 138
VII. General discussion... 141
VIII. Conclusions and future works... 147
VIII.1. Research hypotheses’ verification... 148
VIII.1.1. General hypothesis... 148
VIII.1.2. First instrument motion analysis method... 149
VIII.1.3. Second instrument motion analysis method... 149
VIII.1.4. Third instrument motion analysis method... 151
VIII.2. Main contributions... 151
VIII.4. Future works... 155
VIII.4.1. Concerning the first method based on stereoscopic techniques... 155
VIII.4.2. Concerning the second method based on an optical pose tracker... 156
VIII.4.3. Concerning the third method based on a classifier... 157
Annex I: Virtual environment for tracking methods of laparoscopic instruments... 171
Annex II: Demographic questionnaire... 175
Annex III: GOALS rating scale and suturing checklist... 177
Annex IV: Face validity questionnaire... 181
Annex V: Camera position in the operating room for an external tracker... 183
Annex VI: Learning curves in intracorporeal suture... 189
Glossary of abbreviations and acronyms... 192
Minimally invasive surgery (MIS) has become in many surgical specialties and procedures the gold-standard choice due to its efficiency and benefits towards patient safety. However, the introduction of laparoscopic surgery has led to the need of developing new surgical skills different to those required for open surgery, with a significant learning curve to perform a safe laparoscopic surgery. Traditional subjective assessment methods of trainees are no longer adequate for surgical training. Reduced working hours as well as demands from surgeons and associations mean that more objective assessment tools that can accredit surgeons as technically competent are required.
Evidence exists to validate motion analysis for laparoscopic technical skills assessment. Motion analysis of the laparoscopic instruments seeks to determine aspects that indicate the difference between surgeon's level of surgical dexterity. However, at the moment there is not an extended method to be used with current available training systems used in training laboratories as well as for the OR. Therefore, further efforts are needed on developing cost-effective training and evaluation systems to teach and objectively assess the next generation of laparoscopic surgeons regardless of the training scenario.
Taking into account the needs discussed above, the purpose of this thesis is to design, develop and validate three novel motion analysis methods focused on the use of real laparoscopic instruments during laparoscopic performance. These methods are based on computer vision techniques attempting to not interfere with the surgical practice. They will be presented in an evolutionary way from methods for exclusive use in a box trainer to solutions with the potential of being used in actual OR setting.
The first method introduces a tracking approach of laparoscopic instruments based on stereoscopic vision techniques. A stereoscopic camera system is implemented and installed inside a box trainer, arranged to identify both laparoscopic instruments from two different angles and capture their motion. Technical validation shows that the laparoscopic instruments movements can be accurately recorded by the tracking method in a box trainer setting, providing a reliable source of information. Surgeons reported that the instrument tracking method does not interrupt the normal use of laparoscopic instruments as well as the performance of training tasks inside the box trainer.
The second method is built around a commercially available optical pose tracker. This method seeks to address some limitations presented in the previous approach such as the possibility of using the instrument tracking method in other training settings. Two approaches for the system design are presented with their corresponding technical validation. Accuracy results for the tracking method of laparoscopic instruments show that it can be used as a tool for objective assessment of MIS technical skills based on motion analysis. Usefulness of the method as an assessment tool has been evaluated positively by surgeons. Construct validity has been confirmed for grasping, cut and suturing tasks in a box trainer setting. Concurrent validity with regard to parameters scored by GOALS assessment system has been verified for both cut with non-dominant hand and suturing tasks. Moreover, feasibility of using the method in an OR setting during the dissection and suturing tasks has been proved.
The third approach implements a video-based tracking method of laparoscopic instruments based on the endoscopic video as the only source of information. This instrument tracking method attempts to deal with some of the challenges of the previous approaches such as the need for a camera system inside the simulator or the use of artificial markers on the laparoscopic instruments. Additionally, an assistance system to provide visual support contents during the performance of laparoscopic training tasks is presented. Results have shown the feasibility of identifying and tracking in real time a laparoscopic scissors during the performance of laparoscopic training tasks using only the endoscopic video as a source of information. The tracking system has been proven to be robust under different conditions of illumination, speed of movements and position of the instrument, reporting high accuracy with respect to the different validation scenarios in a box trainer setting. Feasibility of embedding multimedia support content in the working area displayed by the camera has been proved as well as keeping this content stable with regard to the position initially indicated by the user with a high success rate.
Overall, this PhD corroborates the research hypothesis regarding the use of three different video-based tracking technologies for motion analysis of laparoscopic instruments, the use of instrument motion analysis for MIS technical skills assessment and the relationship between motion-related assessment metrics and quality of technical performance in laparoscopic training. These presented methods provide a tool to objectively assess MIS technical performance and a support to train novice surgeons in MIS techniques. The findings of this thesis encourage us to continue researching in improving these methods to be introduced as part of an actual laparoscopic training program, both inside and outside the OR.
La cirugía de mínima invasión se ha convertido en la opción de referencia en muchas especialidades y procedimientos quirúrgicos debido a su eficacia y beneficios para el paciente. Sin embargo, la introducción de técnicas quirúrgicas como la laparoscopia ha llevado a la necesidad de desarrollar nuevas destrezas quirúrgicas distintas de la cirugía convencional, lo cual requiere una importante curva de aprendizaje para poder llevar a cabo procedimientos quirúrgicos seguros. En este sentido, los métodos subjetivos tradicionales de evaluación no son adecuados para un entrenamiento quirúrgico eficaz. La reducción en el número de horas de trabajo, y por tanto de formación, junto con la demanda de los cirujanos y asociaciones, llevan a la necesidad de desarrollar más herramientas de evaluación objetivas que puedan acreditar a los cirujanos como técnicamente competentes.
Existen evidencias para considerar el análisis del movimiento una opción válida en la evaluación de las destrezas técnicas laparoscópicas. El análisis de los movimientos del instrumental laparoscópico como método de evaluación busca determinar aspectos diferenciadores respecto a los niveles de destreza quirúrgica de un cirujano. Sin embargo, por el momento no existe un método estandarizado para su uso con los sistemas actuales de entrenamiento quirúrgico, así como en el quirófano. Por consiguiente, es necesario llevar a cabo más esfuerzos en el desarrollo de sistemas eficaces de entrenamiento y evaluación que permitan la formación y la evaluación objetiva de las nuevas generaciones de cirujanos laparoscopistas.
Teniendo en cuenta los requerimientos anteriores, como finalidad de esta tesis se plantea diseñar, desarrollar y validar tres novedosos métodos de análisis del uso de instrumental laparoscópico real para su aplicación durante las primeras fases de formación laparoscópica. Estos métodos están basados en técnicas de visión por computador, buscando interferir lo menos posible en el flujo normal de las actividades formativas. Estas tres soluciones serán presentadas de un modo evolutivo, desde métodos que únicamente puede ser empleados en un simulador de entrenamiento laparoscópico, hasta soluciones con el potencial de ser aplicados a un entorno quirúrgico real.
La primera solución presentada muestra un método de seguimiento de instrumental laparoscópico basado en técnicas de visión estereoscópica. Para ello, un sistema de cámara estéreo es instalado en el interior de un simulador para poder registrar la posición tridimensional de los movimientos del la punta de los instrumentos laparoscópicos. Los resultados de las pruebas de validación muestran que el método de seguimiento de instrumental es capaz de registrar con precisión los movimientos del instrumental laparoscópico dentro del simulador de entrenamiento, proporcionando una fuente fiable de información. Por otro lado, los cirujanos declararon que el sistema no interrumpe en el uso habitual de los instrumentos laparoscópicos ni en el desarrollo de las actividades de entrenamiento en el interior del simulador.
El segundo método presentado aplica un dispositivo de seguimiento óptico comercial a la práctica laparoscópica. Este método afronta algunas de las limitaciones presentes en el método anterior tales como la posibilidad de ser utilizado en con otros sistema de entrenamiento laparoscópico. Para el desarrollo de esta nueva solución al seguimiento de instrumental laparoscópico se presentan dos posibles diseños con sus respectivas validaciones técnicas. Se obtienen resultados positivos en cuanto a la precisión del sistema en el registro de los movimientos del instrumental laparoscópico. Los cirujanos valoraron de forma positiva la utilidad del método de seguimiento presentado como herramienta para llevar a cabo una evaluación objetiva de las destrezas técnicas quirúrgicas. Los resultados obtenidos confirman la validez constructiva de un conjunto de métricas basadas en el uso del instrumental laparoscópico durante las tareas de agarre, corte y sutura en simulador. Por otro lado, las métricas evaluadas muestran validez concurrente durante las tareas de corte con la mano no dominante y sutura intracorpórea respecto a los resultados obtenidos en los parámetros evaluados por el método de evaluación GOALS. Se ha demostrado la viabilidad en el uso del método de seguimiento de instrumental en un entorno quirúrgico real para el cálculo de un conjunto de métricas objetivas de evaluación durante la ejecución de las tareas de disección y sutura intracorpórea, sin mostrar oclusiones e interrupciones del flujo de trabajo normal.
La tercera solución presenta un método de seguimiento de instrumental laparoscópico basado en el análisis de la imagen del vídeo endoscópico como única fuente de información. Esta solución pretende solventar algunas limitaciones de los dos métodos anteriores tales como la necesidad de instalar un sistema de cámaras en el interior del simulador o el uso de marcas artificiales en los instrumentos laparoscópicos. Por otro lado, también se presenta un sistema de asistencia a la formación que permite insertar contenidos multimedia de apoyo durante la ejecución de tareas formativas en las primeras fases de entrenamiento laparoscópico. Los resultados demuestran la viabilidad del método de seguimiento de instrumental para poder identificar y registrar en tiempo real los movimientos de la punta de una tijera laparoscópica, haciendo uso del vídeo endoscópico. Del mismo modo, los resultados muestran un comportamiento robusto del método de seguimiento bajo diferentes condiciones de iluminación y velocidad en el uso del instrumental. El método muestra una alta precisión en la localización de la punta del instrumental en diferentes entornos de validación dentro del simulador de entrenamiento. Por otro lado, se ha demostrado la viabilidad del sistema de asistencia a la formación para la inclusión de contenido de apoyo en el área de trabajo mostrado por la cámara endoscópica, así como una alta tasa de éxito para mantener dicho contenido de un modo estable respecto a la posición inicialmente marcada por el usuario.
En general, esta tesis corrobora las hipótesis investigadas en cuanto al uso de tres tipos de técnicas diferentes para el análisis del uso del instrumental laparoscópico, la aplicación del análisis de los movimientos del instrumental para la evaluación objetiva de destrezas técnicas quirúrgicas y la relación entre métricas basadas en el uso del instrumental quirúrgico con la calidad de la ejecución técnica durante la formación laparoscópica. Los métodos presentados proporcionan varias herramientas para poder llevar a cabo una evaluación objetiva de las destrezas técnicas quirúrgicas basada en el análisis del uso del instrumental laparoscópico. Las conclusiones de esta tesis fomentan a seguir investigando en la mejora de estos métodos de modo que puedan ser empleados en programas formativos en cirugía laparoscópica.
Minimally invasive surgery is a highly demanding surgery concerning technical requirements for surgeons, which must be trained in order to perform safe surgical interventions. Traditional surgical education in minimally invasive surgery is commonly based on subjective criteria to quantify and evaluate surgical abilities, which could be imprecise and lead to the incorrect accreditation of the surgeon's skills. Surgical associations, surgeons and authors are increasingly demanding the development of more objective assessment methods and tools that can accredit surgeons as technically competent. This chapter presents the foundations of this PhD as well as the justification of the research and the aim of this thesis.
I.1. Essentials of laparoscopic surgery
The term minimally invasive surgery (MIS) was introduced by John Wickman to describe the emerging therapeutic approach designed to minimise the traumatic offense to the patient by surgical allied interventional procedures (Cuschieri, 2005). MIS can be applied to the abdomen (laparoscopy), chest (thorascopy), joints (arthroscopy), gastrointestinal tract (coloscopy), uterus (hysteroscopy), and blood vessels (angioscopy). As we are mainly concerned with the abdominal applications of the technique, most of the material presented in this thesis pertains to laparoscopy.
In laparoscopy, a workspace inside the patient and visualization of the abdominal cavity is possible due to the pneumoperitoneum generated by infusing CO 2 gas through an inserted needle, which provides surgeon space to perform the surgical procedure. This surgical procedure is performed through small incisions on the patient's abdomen where a laparoscope is inserted throw one of these access ports to provide the images from the surgical field (Figure 1 and Figure 2) (Uson et al., 2013). This kind of surgery tries to reduce the invasiveness for the patient, providing them with several benefits such as reduction of tissue trauma and post-operative pain (Cordera et al., 2003), better aesthetic results (Seitz et al., 2008), and less post-operative periods (Delaney et al., 2008) with a consequent reduction of the healthcare costs. These benefits have made these surgical techniques the focus of attention and the choice in many surgical procedures, which were usually performed by open surgery.
Additionally, in recent years the growing trend towards reduction of traumatic damage to the patients and reduction of the invasiveness of surgical interventions, which seeks to decrease the postoperative pain, visible scarring and complications such as wound infection, has led to the development of new MIS techniques including Laparo-Endoscopic Single-Site Surgery (LESS) and Natural Orifice Translumenal Endoscopic Surgery (NOTES) (Díaz-Güemes et al., 2013; Sánchez-Margallo FM et al., 2012).
I.2. Psychomotor challenges
The evolution from open surgery to MIS requires learning and training new skills, surgical manoeuvres (Usón et al., 2013), and the way to use a new set of surgical tools (Sánchez-Margallo FM et al., 2010a). Surgical experience acquired for open surgery need not be related to the laparoscopic performance (Figert et al., 2001). During the early years, the popularity of laparoscopic surgery led to an alarming increase in medical errors in common open procedures such as cholecystectomy, due in great part to an insufficient preparation in the required skills for this type of surgeries (Hiemstra, 2012; Tsuda et al., 2009). Consequently, learning programmes started refocusing their goals to the adoption of the specific skills needed in MIS.
Laparoscopy requires surgeons to perform tasks following two-dimensional video images of the operative field from the laparoscope in which only the distal part of the laparoscopic instruments are visible (Figure 1). This leads to some psychomotor challenges such as loss of depth perception and 2D interpretation of 3D structures. Besides, laparoscopic instruments are longer than traditional instruments, and consequently the surgeon’s hands are far from the working ends of the instruments, which can cause hand tremors after long periods of use. Surgeons have to relearn the skills of physical manipulation in order to deal with the loss of the sense of touch and the reduced degrees of freedom (DoF) of laparoscopic instruments. Movements are reduced from 6 DoF in open surgery to 4 DoF in laparoscopic procedures (Figure 3). In addition, during the performance of this surgical technique appears the fulcrum effect, in which directional movements of the surgeon’s hand result in contrary deflections of the working end of the laparoscopic instrument, creating a disparity between visual and real feedback.
The aforementioned limitations together with the unergonomic design of laparoscopic instruments entail also some ergonomic problems for the surgeon (Pérez-Duarte et al., 2013; Sánchez-Margallo FM et al., 2010a). In conclusion, laparoscopic surgery is a technically demanding discipline, which imposes significant psychomotor challenges on surgeons (Table 1), which have to be trained, assessed and certified (van Hove et al., 2010; Pellen et al., 2009; Aggarwal and Darzi, 2006; Sokollik et al., 2004).
I.3. Skills to be a proficient surgeon
There are three main qualities in the development of a surgeon as competent in laparoscopic surgery; these are cognitive skills, technical skills, and surgical judgment (van Hove et al., 2010; Doyle et al., 2007; Dankelman et al., 2005; Miller et al., 1990; Bloom et al., 1956) (Figure 4). The cognitive skills are the theoretical background required to perform a surgery. Technical skills consist of manual and technical abilities required in the OR in order to apply the previous knowledge. Finally, a competent surgeon must be able to apply his good criteria and knowledge when facing the different events that may occur in the OR. These skills are known as surgical judgement, which are the most difficult to acquire due to they are developed almost exclusively through experience and acquired knowledge. Although a competent surgeon possesses knowledge, judgement, decisiveness, and professionalism, proficiency in technical skills seems to be fundamental to perform surgery safety (Feldman et al., 2004). To guarantee safe use of laparoscopic instruments with difficult handling, in a limited working area, and with limited tactile perception, training of the operative skills is crucial.
I.4. Evolution of surgical education
I.4.1. Traditional surgical learning
Since the beginning of the 20th Century, traditional surgical education in MIS has followed the William Halsted’s paradigm, who proposed an intensive training programme based on the axiom “see one, do one, teach one” (Halsted, 1904). This learning method is based on subjective criteria, where the teaching process relies mainly on a mentor-apprentice relationship, and more importantly, takes place in the OR with real patients.
During a traditional learning programme, an experienced surgeon, who observes the evolution of the trainee, makes the evaluation of his surgical skills. The assessment is usually performed by means of general reports fulfilled by the supervisor, such as the In Training Evaluation Reports (ITER) (Sidhu et al., 2004). Then, trainees are allowed to contribute to the surgical intervention under the supervision of an experienced surgeon (EAES, 1994; Reznick, 1993; Halsted, 1904). However, throughout this training methodology technical skills cannot be precisely assessed and it could be affected by bias (Fried and Feldman, 2008; Smith et al., 2001). Consequently, this subjective method to quantify and evaluate surgical abilities could be potentially unsafe for the patient (Fried and Feldman, 2008; Aggarwal et al., 2007; Verdaasdonk et al., 2007). For this reason, authors, surgeons and associations (e.g., Acreditation Council for Graduate Medical Education) are increasingly demanding the development of more objective assessment tools that can accredit surgeons as technically competent (van Hove et al., 2010; Roberts et al., 2006; Ritchie, 2004; Park and Witzke, 2002; Darzi et al., 1999; Satava, 1999). Nevertheless, nowadays there is not a universally extended or recommended system to be used as an objective evaluation tool of technical skills in laparoscopic surgery (van Hove et al., 2010; Cuschieri, 2005).
I.4.2. New surgical learning approaches
The first time a surgeon faces the psychomotor challenges in MIS (Table 1) it should not be on a living patient, and it seems reasonable that as much skill acquisition as possible should be moved into training settings where mistakes do not compromise patient safety (Park and Lee, 2011). Animal models and human cadavers provide excellent training opportunities for this aim despite their high cost. Human cadavers offer accurate anatomy for learning MIS skills (Sánchez-Margallo FM et al., 2010b). However, problems with conserving the tissue and lack of bleeding in the case of damaging tissues and vessels are the main reason for using in many cases live animal models instead of human cadavers. Few places, however, have animal housing to make animal model as a regular part of a training curriculum. In addition to this, there are ethical considerations in regard to acquiring surgical skills by using live animals and in Europe there is an increase in restrictive laws on the use of experimental animal models.
The aforementioned shortcomings of training on patients, animals, and human cadavers have led to a growing interest in laboratories with formal curricula, specially designed to teach technical skills during the earlier stages of training in laparoscopic surgery. With this new model of surgical education, basic surgical skills are learned and practiced on inanimate models and simulators, with the aim of better preparing trainees for the operating room experience and away from the clinical responsibility of putting a patient at risk. The use of surgical simulators for proficiency-based training programs are a suitable solution as a paradigm shift in this kind of surgical education models (Gallagher et al., 2005). There is evidence that surgical skills level can affect clinical outcomes, and more effective surgical training and evaluation could have a significant impact on health care (Reznick and MacRae, 2006).
Nowadays, there are several training platforms (either academic prototypes or commercial products) that make acquisition of these new cognitive and psychomotor skills easier. Surgical simulators have made great advances for learning surgical skills. Several studies have reported the increasing of a surgeon’s surgical abilities training with a surgical simulator (Stelzer et al., 2009; Sturm et al., 2008). They provide different surgical scenarios and software to assess surgical practice (Oropesa et al., 2011a), and therefore they are the opportunity to move the learning curves out of the operating room.
Broadly, simulators can be classified into virtual reality simulators, box trainers (physical simulators) and hybrid simulators. At the moment, the realism of the haptic sense as well as surgical scenarios in virtual simulators is still insufficient. Besides, these simulators are not always affordable for all surgical training centres or hospitals. Realistic force and tactile feedback are fundamental parts for successful laparoscopic training and result in improved skills practice transfer to residents (Botdem and Jakimowicz, 2009), particularly compared to training without haptic feedback (Sándor et al., 2010). Surgeons consider box trainers suitable training aid for practising surgical skills outside the operating room (Stylopoulos et al., 2003). They are a very economically reasonable way to train fundamental surgical tasks and procedures with a completely real haptic sensation (Enciso et al., 2012; Sánchez-Margallo FM et al., 2009a; Fried GM, 2008). Nevertheless, current available training systems lack of a standardized assessment methodology, having into account that standardization is a key characteristic of all successful examinations and educational aids.
In order to carry out an adequate performance of laparoscopic surgery it is important to progress from the basic principles to advances skills through its steep learning curve (Usón et al., 2013). The sequential training model follow by the Jesús Usón Minimally Invasive Surgery Centre (JUMISC) is an example of a training program structured in four levels, in which gradually the trainee progresses from basic to advanced training (Figure 5). (L1) At the first level surgeon learn basic and advanced surgical skills using surgical simulators and inanimate models; (L2) then, they learn anatomical protocols and advanced surgical procedures on animal models; (L3) During the third level training of advanced procedural skills are combined with the use of Information and Communication Technologies (ICT) support; (L4) and finally at the last level, surgeons are ready to apply all the knowledge and surgical skills to the practice in the OR. Accreditation at each training level should be achieved before moving on to the next level (Satava et al., 2003; Park and Witzke, 2002). However, there is a general lack of consensus on the criteria that should officially define the tasks, metrics and assessment methods to employ in laparoscopic training programs.
I.5. Objective assessment methods
I.5.1. Structured rating systems for objective evaluation
I.5.1.1. Rating scales
Several methods have been proposed to deal with the lack of standardized objective assessment tools of surgical skills, most of them based on structured grading (Chipman and Schmitz, 2009; Doyle et al., 2007; Vassiliou et al., 2005). These methods attempt to standardize evaluation through rated checklists on an inanimate benchtop model. One of the most generic type is the Global Rating System (GRS) (Doyle et al., 2007; Larson et al., 2005), which is a nonblinded evaluation method consisting of a number of items as general markers of technical skills that could be applied to a wide variety of procedures, but not for specific tasks. This method has some limitations such as the need of an instructed examiner during the evaluation procedure and, due to its nonblinded approach, it is potentially affected by bias. Video assessment methods using GRS provide a more objective blinded rating process and with the same scoring criteria as live rating systems (Driscoll et al., 2008; Adrales et al., 2004). The Operative Component Rating Scale (OCRS) is another video assessment method based on an operation-specific rating form. The assessment of a surgical procedure is performed with respect to technical skills on a five-point rating scale. However, also a group of experienced surgeons is needed to define the procedural component to be assessed and to rate the recorded videos (Dath et al., 2004).
The Objective Structured Clinical Examination (OSCE) was described by Harden et al. (Harden et al., 1975) in order to avoid many of disadvantages associated with traditional examination of clinical competence. A typical OSCE consists of a series of stations through which trainees rotate. At each station the trainee is asked to perform a specific evaluation task, which is assessed by an experienced surgeon using checklists or global rating scales. This method is mainly focused on the assessment of procedural knowledge and attitude of the trainee towards the patient, but without a specific part for technical skills assessment (Sidhu et al., 2004).
The Objective Structured Assessment of Technical Skills (OSATS) was developed following the success of the OSCE method in assessing surgical competences (Reznick et al., 1997; Martin et al., 1997). This is a performance-based assessment of technical skills, in which trainees perform a series of time-limited surgical tasks at each station (Chipman and Schmitz, 2009). It was one of the first objective assessment methods of technical skills. It involves the use of two scoring systems: a task-specific checklist score and a global rating scale of overall performance. Although OSATS was first validated for being use with a benchtop model, it also can be applied to physical laparoscopic simulators (Broe et al., 2006) and in the operating theatre (Aggarwal et al., 2007; Larsen et al., 2008).
A more specific version of OSATS method is the Global Operative Assessment of Laparoscopic Skills (GOALS), which was developed by Vassiliou et al. (Vassiliou et al., 2005) to provide a method based on a global rating system for evaluation of operative performance during a laparoscopic procedure (Sroka et al., 2010). This assessment tool rates five performance-based domains such as depth perception, bimanual dexterity, efficiency, tissue handling, and autonomy. Chang et al. (Chang et al., 2007) used GOALS as assessment tool for surgeons and residents using an intraoperative-recorded videotape of the laparoscopic procedure and a blinded review.
These objective assessment methods however have some limitations such as the high resources required, the subjectivity issues of human observers, the possible ambiguity of their scores, and the need of the presence of an experienced supervisor, which drive the necessity to develop more automated evaluation systems.
I.5.1.2. Assessment programmes
In the late 1990s, the Society of American Gastrointestinal Endoscopic Surgeons (SAGES) began development of the Fundamentals of Laparoscopic Surgery (FLS), a comprehensive program designed to teach the cognitive and psychomotor skills in laparoscopic surgery (Edelman et al., 2010; Mashaud et al., 2010; Sroka et al., 2010; Derossis et al., 1998). Nowadays it is one of the most extended objective scoring systems for evaluation of surgical skills. This program consists of web-based study guides and hands-on manual skills practice and training (Figure 6 and Figure 7). The assessment of the cognitive component is a timed multiplechoice exam performed via computer. It tests the understanding and application of the basic fundamentals of laparoscopy with emphasis on clinical judgment or intraoperative decision-making. The manual skills assessment is based on the McGill Inanimate System for Training and Evaluation of Laparoscopic Skills (MISTELS) (Vassiliou et al., 2006; Fraser et al., 2003), which is a series of five simulation tasks with an objective scoring system based on proficiency-based criteria set by expert performance. The tasks used for laparoscopic skill evaluation are peg transfer, pattern cutting, placement of a ligating loop, and intracorporeal and extracorporeal knot tying (Figure 7). This system assesses each individual task using time and accuracy (number of errors) as evaluation parameters. After the FLS program was implemented, several studies examined its feasibility, reliability and validity, and demonstrated that the program did have advantages in laparoscopic training (Sroka et al., 2010; McCluney et al., 2007).
The great success of FLS has led it to be implemented in the USA and Canada. The Royal Australian College of Surgeons has also incorporated FLS into its training programs for all surgical residents (Soper and Fried, 2008). The need to take a box trainer-based exam at a specialized test centre demonstrates the major disadvantage of this system as well as the measurement of performance and objective evaluation metrics, which require an independent evaluator.
In Europe, the European Association for Endoscopic Surgery (EAES) developed the Laparoscopic Surgical Skills programme (LSS), which is a programme that offers a standard for comprehensive assessment for training and education in laparoscopic surgery within a multi-level curriculum (Buzink et al., 2012). This curriculum is focused on training and assessment of basic laparoscopic skills and fundamental laparoscopic procedures as well as specific advanced laparoscopic procedures. Within each level, the LSS assessment comprises a sequence of tests to evaluate the surgeon’s proficiency in cognitive skills, surgical technical skills and judgement (Figure 8).
With a web-based knowledge test, adequate acquaintance of the theory presented in the course regarding procedures, techniques, instrumentation and ergonomics is tested. Passing this exam is a requirement to be admitted to the simulator assessment and scenario-based assessment. Simulator assessment certifies that participants have achieved a sufficient level of psychomotor and technical surgical skills to start performing the specific index procedures in the clinical setting under the supervision of an acknowledged trainer. Scenario-based assessment is conducted to assess the knowledge and judgment skills with regard to the index procedures. This assessment consists of an a web-based exam with real clinical situations. After successful completion of the simulator assessment and scenario-based assessment, evidence for sufficient expertise in supervised clinical procedures is required. Each LSS level is therefore concluded by two steps for clinical performance assessment. The Global Assessment Score forms (Wyles et al., 2011) are used to assess overall workplace performance of the procedures, after which the Competency Assessment Tool (Miskovic, 2013) is used to analyse the video recordings.
1.5.2. Instrument motion analysis
In order to investigate surgeons’ laparoscopic technical abilities during surgical interventions, it seems reasonable to observe carefully the movements of the laparoscopic instruments. Measuring competence merely by setting time targets for certain procedures could be inaccurate and insufficient. A fast surgeon is not necessarily a good surgeon. Counting the number of procedures performed has also been used as a tool to accredit surgeons, but it does not have to tell about how well the surgeon operates (Darzi et al., 1999).
Traditional psychology literature suggests that when subjects learn a complex motor task, as they become more competent at that task, they also become more efficient in the movements they use to complete it. At the beginning, they may be inaccurate and with uncoordinated movements; but as they learn and become more competent at the task, they achieve more efficient, coordinate, and smooth movements (Rosenbaum, 2009). The three-stage theory of motor skills acquisition has been globally accepted (Sándor et al., 2010). During the first stage of training, the learner has to understand the task and create a plan to perform it, but this stage also allows the learner to make errors (cognitive stage). With improved practice, the learner reaches the integrative stage, when the task is performed more fluently. Finally, in the autonomous stage, the learner no longer needs to think about how to execute the task and can therefore concentrate exclusively on the actual task (Reznik, 1993).
Instrument motion analysis has been used successfully in several studies to address the assessment of surgeon’s psychomotor skills (Oropesa et al., 2013a; Climent and Hexsel, 2012; Pagador et al., 2012; Chmarra et al., 2010; Datta et al., 2007). Motion analysis can typically be performed both in VR simulators and in real-life training (box trainers/OR). Systems for the latter case are cheaper assessment alternatives, enabling natural feedback. Assessment methods based on instrument motion analysis can be used in a wide range of surgical and training settings from box trainers to the OR. Therefore, the range of laparoscopic surgical tasks in which these assessment tools can be used is extensive.
Motion analysis of surgical instruments requires track, record, and evaluate this motion information. To develop a complete objective assessment method for surgical technical skills based on instrument motion analysis four fundamental phases should be achieved (Figure 9):
1. To describe the tasks and metrics to quantify the surgeon’s technical skills.
2. To define and implement a method to record these metrics (tracking method).
3. To develop an evaluation method based on the recorded metrics that can identify the surgeon's level of surgical dexterity.
4. To validate the method and its reliability as a surgical assessment tool.
The need to track laparoscopic instrument results from the significant learning curve required performing safe laparoscopic techniques, and the need to provide an objective assessment of the surgeon's technical skills. Computational analysis of instrument motion offers the potential to assess technical skills more objectively and cost effectively than structured human grading, complementing its results. This provides objective criteria, in which students can reassess and correct their technical performance as necessary.
A number of important aspects are considered when designing a motion analysis method of laparoscopic instruments. Some of the most important are to ensure that it provides trainee with force and visual feedback. In a clinical context, these two aspects cannot be separated. One of the main reasons is because, despite the limited force feedback in laparoscopic surgery, surgeons learn to interpret visual information in order to supplement the sense of force. Other important issues to have in mind are portability and system accuracy. These methods should be used in different training settings, considering that there are as many as three environments in which surgical motion analysis may occur: in a virtual reality trainer, box trainer, or in the operating room. Finally, system accuracy is essential when performing the instrument tracking, as it is the basic information to compute the motion-related assessment metrics.
Different technologies have been applied to develop motion analysis methods of laparoscopic instruments for objective assessment such as mechanical (Gunther et al., 2007), acoustic (Sokollik et al., 2004), infrared (Estebanez et al., 2011a), electromagnetic (Pagador et al., 2012), sensor-based (Chmarra et al., 2010), and video-based technologies (Oropesa et al., 2013a). A laparoscopic instrument motion analysis should seek to minimize its size and be as non-obstructive as possible, as external wires and nonsurgical equipment in the operating room could affect surgeons’ performance and hinder their natural movements.
Since medical education extends over the lifetime of the surgeon an evaluation method of technical skills, if reliable and valid, could be used as a tool in monitoring of the resident’s progress and regular certification. A standardized assessment of technical skills would be helpful as a feedback tool for trainees, allowing them to identify their technical skills’ deficits earlier.
I.5.3. A good assessment tool
Prior to implementation of a new assessment tool in a training curriculum, feasibility, validity and reliability of this tool should be determined (Aggarwal et al., 2007; McDougall, 2007). The feasibility term measures whether the assessment process is capable of being done by the tool being tested. The reliability quality verifies whether the information provided by the assessment tool is reliable and accurate enough. Finally, the validity of an assessment tool verifies that it measures what it intends to measure (Aggarwal et al., 2007; McDougall, 2007; Gallagher et al., 2003).
There are a variety of aspects to validity, which can be broadly classified into subjective (face, content) and objective (construct, concurrent, predictive). Face validity determines the realism of the assessment tool, the extent to which it resembles real life situations and whether it is considered useful for assessment. Content validity reviews whether the contents used in the assessment cover correctly the skills the system attempts to measure. Construct validity indicates whether the assessment tool is able to discriminate between various levels of expertise. Construct validity is a mandatory step in determining the validity of an assessment tool to verify the hypothesis that a higher surgical trainee will consistently perform better than a less experienced trainee. This validation test is usually done by measuring through the assessment method performance in two or more groups that are hypothesized to differ in the surgical skills to be assessed (e.g., experts surgeons and novices). In discriminative validity the test scores on the assessment tool under validation and the scores achieved on another instrument purposing to measure the same concept are related. If validation is performed close in time or simultaneously over the same data, it is known as concurrent validity. Additionally, if the instrument against it is being validated can be considered a gold standard this validity is known as criterion validity. Predictive validity is defined as the extent to which the scores on the assessment tool predict actual performance.
I.6. Problem statement and aim of this thesis
The presented PhD work is conducted under the research line “Advanced systems for surgical training” of the Jesús Usón Minimally Invasive Surgery Centre (JUMISC) in Cáceres, Spain. The main objective of this research line is to design and develop collaborative platforms, training and assessment devices and other technologies in order to improve both the training process and surgical skills acquisition.
Work of this research line is performed in close collaboration with the Robotics and Artificial Vision Laboratory (ROBOLAB) at the Escuela Politécnica, Universidad de Extremadura (Cáceres, Spain) and the Bioengineering and Telemedicine Centre (GBT) at the Escuela Técnica Superior de Ingenieros de Telecomunicación (ETSI), Universidad Politécnica de Madrid (Madrid, Spain). ROBOLAB is a laboratory devoted to conducting research in Intelligent Mobile Robotics and Computer Vision techniques. GBT aims to teach and research on the application of telecommunications technology and multimedia in the field of telemedicine and medical imaging.
The first step to develop a surgical training program is clearly to identify the educational outcomes and objectives. Through assessment tools, training programs should determine how well their trainees are achieving these objectives (Sidhu et al., 2004). The research under this PhD is focused on the design, development and validation of novel motion analysis methods of laparoscopic instruments as tools for surgical technical skills assessment .
Laparoscopic surgery requires new technical skills for the surgeon, which must be trained and assessed in order to become a proficient surgeon, enabling to perform safe surgical interventions. Traditional subjective assessment methods of trainee are no longer adequate for surgical training (van Hove et al., 2010; Fried and Feldman, 2008; Darzi et al., 1999). Reduced working hours (Pearsons et al., 2011) and demands form surgeons and associations (Roberts et al., 2006; Ritchie, 2004; Smith, 1998) mean that more objective assessment tools that can accredit surgeons as technically competent are required. Evaluation of surgeons’ psychomotor skills during their training program is an essential part of the complete assessment of their surgical proficiency and the lack of a standardized method reinforces the research of this thesis.
Evidence exists to validate motion analysis for use in laparoscopic technical skills assessment (Mason et al., 2013; Oropesa et al., 2011a). These methods compute metrics to quantify motion information concerning the use of the surgical instrument to establish the surgeon's level of surgical dexterity. They seek to determine aspects that indicate the difference between performances at various levels of proficiency. Nevertheless, at the moment there is not an extended method to be used with current available training systems used in training laboratories as well as in the OR.
Within this context and taking into account the needs discussed above, we propose three novel motion analysis methods focused on the use of real laparoscopic instruments during surgical performance. These methods are based on computer vision techniques attempting to interfere as little as possible with the surgical practise. The organizational structure followed in this thesis is showed in Figure 10, where each proposed motion analysis method for laparoscopic instruments is presented. They will be shown in an evolutionary way from methods for exclusive use in a box trainer to solutions with the potential of being used in actual OR setting:
· The first method will introduce a tracking system of laparoscopic instruments based on stereoscopic techniques, which is installed inside a box trainer.
· The second method will adapt a commercially available third-generation optical pose tracker for laparoscopic practice.
· Finally, the last method will implement a video-based tracking method of laparoscopic instruments based on the endoscopic image as the source of information.
Research will be focused on the design, implementation and validation of all the aforementioned approaches for motion analysis of laparoscopic practise. All these systems will be installed, tested and validated for laparoscopic training using a box trainer setup. The research work addressed in this thesis will be the first steps to achieve, as a future goal, a method that can be used in an actual laparoscopic training program.
Besides their use as tools for motion analysis, laparoscopic instrument tracking methods offer a wide range of possibilities in surgery. They are an essential component of image and video guided surgery (IVGS) systems (Sánchez-González et al., 2011). These systems are useful for surgeon when the surgical instruments are outside the field of view, when it is difficult to identify in the image due to the artefact that it produces, or when the instrument cannot be readily detected by the imaging modality (Perrin et al., 2009). In visual servoing applications, this technology could be used to guide robotic arms using visual feedback from the camera system (Ryu et al., 2013; Staub et al., 2010; Groeger et al., 2008).
I.7. Thesis outline
This thesis is structured according to each method proposed, and therefore each main chapter will cover the practical design, implementation and validation of one motion analysis method. Thus, the outline will be the following:
Chapter I has established the foundations of this PhD, focusing on the relevant aspects of training and assessment of laparoscopic skills, and more specifically concerning technical skills. In addition, justification of the research and the aim of this thesis have been presented.
Chapter II will present the hypotheses that drive this research, as well as the specific objectives to meet in this PhD work.
Chapter III will present the state of the art of methods and technologies for objective assessment of laparoscopic technical skills.
Chapter IV will present the first approach of motion analysis method of laparoscopic instruments. This method implemented inside a box trainer is based on stereoscopic vision techniques. A technical validation of the described method as well as the user’s evaluation will be presented.
Chapter V will present the second method for motion analysis, consisting on the application of a optical pose tracker for the laparoscopic practice. The method provides motion information of laparoscopic instruments as a mean of subsequent evaluation of the surgeon’s technical skills. Face, construct and concurrent validation of this method will be carried out.
Chapter VI will present the third approach of laparoscopic instrument motion analysis and an assistance tool for the first stages of training in laparoscopic surgery. The motion analysis is a video-based tracking method of laparoscopic instruments tip using the endoscopic camera image as the only source of information.
Chapter VII will present the conclusions extracted from this research. Each hypothesis presented in Chapter II will be discussed, and the main contributions of this PhD will be presented. Finally, guidelines and directions for future works based on the results will be given.