Thomas Neumuth

Validation of Knowledge Acquisition for Surgical Process Models

OBJECTIVE Surgical Process Models (SPMs) are models of surgical interventions. The objectives of this study are to validate acquisition methods for Surgical Process Models and to assess the performance of different observer populations. DESIGN The study examined 180 SPM of simulated Functional Endoscopic Sinus Surgeries (FESS), recorded with observation software. About 150,000 single measurements in total were analyzed. MEASUREMENTS Validation metrics were used for assessing the granularity, content accuracy, and temporal accuracy of structures of SPMs. RESULTS Differences between live observations and video observations are not statistically significant. Observations performed by subjects with medical backgrounds gave better results than observations performed by subjects with technical backgrounds. Granularity was reconstructed correctly by 90%, content by 91%, and the mean temporal accuracy was 1.8 s. CONCLUSION The study shows the validity of video as well as live observations for modeling Surgical Process Models. For routine use, the authors recommend live observations due to their flexibility and effectiveness. If high precision is needed or the SPM parameters are altered during the study, video observations are the preferable approach.

Analysis of surgical intervention populations using generic surgical process models.

PurposeAccording to differences in patient characteristics, surgical performance, or used surgical technological resources, surgical interventions have high variability. No methods for the generation and comparison of statistical ‘mean’ surgical procedures are available. The convenience of these models is to provide increased evidence for clinical, technical, and administrative decision-making.MethodsBased on several measurements of patient individual surgical treatments, we present a method of how to calculate a statistical ‘mean’ intervention model, called generic Surgical Process Model (gSPM), from a number of interventions. In a proof-of-concept study, we show how statistical ‘mean’ procedure courses can be computed and how differences between several of these models can be quantified. Patient individual surgical treatments of 102 cataract interventions from eye surgery were allocated to an ambulatory or inpatient sample, and the gSPMs for each of the samples were computed. Both treatment strategies are exemplary compared for the interventional phase Capsulorhexis.ResultsStatistical differences between the gSPMs of ambulatory and inpatient procedures of performance times for surgical activities and activity sequences were identified. Furthermore, the work flow that corresponds to the general recommended clinical treatment was recovered out of the individual Surgical Process Models.ConclusionThe computation of gSPMs is a new approach in medical engineering and medical informatics. It supports increased evidence, e.g. for the application of alternative surgical strategies, investments for surgical technology, optimization protocols, or surgical education. Furthermore, this may be applicable in more technical research fields, as well, such as the development of surgical workflow management systems for the operating room of the future.

Structured recording of intraoperative surgical workflows

Surgical Workflows are used for the methodical and scientific analysis of surgical interventions. The approach described here is a step towards developing surgical assist systems based on Surgical Workflows and integrated control systems for the operating room of the future. This paper describes concepts and technologies for the acquisition of Surgical Workflows by monitoring surgical interventions and their presentation. Establishing systems which support the Surgical Workflow in operating rooms requires a multi-staged development process beginning with the description of these workflows. A formalized description of surgical interventions is needed to create a Surgical Workflow. This description can be used to analyze and evaluate surgical interventions in detail. We discuss the subdivision of surgical interventions into work steps regarding different levels of granularity and propose a recording scheme for the acquisition of manual surgical work steps from running interventions. To support the recording process during the intervention, we introduce a new software architecture. Core of the architecture is our Surgical Workflow editor that is intended to deal with the manifold, complex and concurrent relations during an intervention. Furthermore, a method for an automatic generation of graphs is shown which is able to display the recorded surgical work steps of the interventions. Finally we conclude with considerations about extensions of our recording scheme to close the gap to S-PACS systems. The approach was used to record 83 surgical interventions from 6 intervention types from 3 different surgical disciplines: ENT surgery, neurosurgery and interventional radiology. The interventions were recorded at the University Hospital Leipzig, Germany and at the Georgetown University Hospital, Washington, D.C., USA.

Acquisition of process descriptions from surgical interventions

The recording and analysis of process descriptions from running surgical interventions is a very new and promising field named Surgical Workflows. Surgical Workflows fulfill two major objectives: they form the base of scientific evaluation and rapid prototyping of surgical assist systems, and they pave the road for the entering of workflow management systems into the operating room for intraoperative support of the surgeon. In this paper we describe how process descriptions from surgical interventions can be obtained for Surgical Process Modelling (SPM) as a specific domain of Business Process Modelling (BPM). After the introduction into the field of Surgical Workflows and the motivation of the research efforts, we deal with theoretical considerations about surgical interventions and the identification of classifications. Based on that, we propose the extendable structure for computational data acquisition support and conclude with use cases. The presented approach was applied to more than 200 surgical interventions of 10 different intervention types from otorhinolaryngology, neurosurgery, heart surgery, eye surgery, and interventional radiology, and it represents an ongoing project.

Modeling surgical processes: A four-level translational approach

MOTIVATION The precise and formal specification of surgical interventions is a necessary requirement for many applications in surgery, including teaching and learning, quality assessment and evaluation, and computer-assisted surgery. Currently, surgical processes are modeled by following various approaches. This diversity lacks a commonly agreed-upon conceptual foundation and thus impedes the comparability, the interoperability, and the uniform interpretation of process data. OBJECTIVE However, it would be beneficial if scientific models, in the same context, shared a coherent conceptual and formal mathematical basis. Such a uniform foundation would simplify the acquisition and exchange of data, the transition and interpretation of study results, and the transfer and adaptation of methods and tools. Therefore, we propose a generic, formal framework for specifying surgical processes, which is presented together with its design methodology. METHODS The methodology follows a four-level translational approach and comprises an ontological foundation for the formal level that orients itself by linguistic theories. RESULTS A unifying framework for modeling surgical processes that is ontologically founded and formally and mathematically precise was developed. The expressive power and the unifying capacity of the presented framework are demonstrated by applying it to four contemporary approaches for surgical process modeling by using the common underlying formalization. CONCLUSIONS The presented four-level approach allows for capturing the knowledge of the surgical intervention formally. Natural language terms are consistently translated to an implementation level to support research fields where users express their expert knowledge about processes in natural language, but, in contrast to this, statistical analysis or data mining need to be performed based on mathematically formalized data sets. The availability of such a translational approach is a valuable extension for research regarding the operating room of the future.

RECORDING OF SURGICAL PROCESSES: A STUDY COMPARING SENIOR AND JUNIOR NEUROSURGEONS DURING LUMBAR DISC HERNIATION SURGERY

BACKGROUND: Evaluating surgical practice in the operating room is difficult, and its assessment is largely subjective. OBJECTIVE: Recording of standardized spine surgery processes was conducted to ascertain whether any significant differences in surgical practice could be observed between senior and junior neurosurgeons. METHODS: Twenty-four procedures of lumbar discectomies were consecutively recorded by a senior neurosurgeon. In 12 cases, surgery was entirely performed by a senior neurosurgeon with the aid of a resident, and in the 12 remaining cases, surgery was performed by a resident with the aid of a senior neurosurgeon. The data recorded were general parameters (operating time for the whole procedure and for each step), and general and specific parameters of the surgeons activities (number of manual gestures, number and duration of actions performed, use of the instruments, and use of interventions on anatomic structures). The Mann-Whitney U test was used for comparison between the 2 groups of neurosurgeons. RESULTS: The operating time was statistically lower for the group of senior surgeons. The seniors statistically demonstrated greater economy in time and in gestures during the closure step, for sewing and for the use of scissors, needle holders, and forceps. The senior surgeons statistically worked for a shorter time on the skin and used fewer manual gestures on the thoracolumbalis fascia. The number of changes in microscope position was also statistically lower for this group. CONCLUSION: There is a relationship between surgical practice, as determined by a method of objective measurement using observation software, and surgical experience: gesture economy evolves with seniority.

Intervention time prediction from surgical low-level tasks

OBJECTIVE Effective time and resource management in the operating room requires process information concerning the surgical procedure being performed. A major parameter relevant to the intraoperative process is the remaining intervention time. The work presented here describes an approach for the prediction of the remaining intervention time based on surgical low-level tasks. MATERIALS AND METHODS A surgical process model optimized for time prediction was designed together with a prediction algorithm. The prediction accuracy was evaluated for two different neurosurgical interventions: discectomy and brain tumor resections. A repeated random sub-sampling validation study was conducted based on 20 recorded discectomies and 40 brain tumor resections. RESULTS The mean absolute error of the remaining intervention time predictions was 13 min 24s for discectomies and 29 min 20s for brain tumor removals. The error decreases as the intervention progresses. DISCUSSION The approach discussed allows for the on-line prediction of the remaining intervention time based on intraoperative information. The method is able to handle demanding and variable surgical procedures, such as brain tumor resections. A randomized study showed that prediction accuracies are reasonable for various clinical applications. CONCLUSION The predictions can be used by the OR staff, the technical infrastructure of the OR, and centralized management. The predictions also support intervention scheduling and resource management when resources are shared among different operating rooms, thereby reducing resource conflicts. The predictions could also contribute to the improvement of surgical workflow and patient care.

Online recognition of surgical instruments by information fusion

PurposeAutomatic online recognition of surgical instruments is required to monitor instrument use for surgical process modeling. A system was developed and tested using available technologies.MethodsA recognition system was developed using RFID technology to identify surgical activities. Information fusion for online recognition of surgical process models was conceived as a layer model to abstract information from specific sensor technologies. Redundant, complementary, and cooperative sensor signal fusion was used in the layer model to increase the surgical instrument recognition rate. Several different information fusion strategies were evaluated for situation recognition abilities in a mock-up environment based on simulations of surgical processes.ResultsThis information fusion system was able to reliably detect, identify, and localize surgical instruments in an interventional suite. A combination of information fusion strategies was able to achieve a correct classification rate of 97% and was as effective as observer-based acquisition methods.ConclusionDifferent information fusion strategies for the recognition of surgical instruments were evaluated, showing that redundant, complementary, and cooperative information fusion is feasible for recognition of surgical work steps. A combination of sensor- and observer-based modeling strategies provides the most robust solution for surgical process models.

Similarity metrics for surgical process models

OBJECTIVE The objective of this work is to introduce a set of similarity metrics for comparing surgical process models (SPMs). SPMs are progression models of surgical interventions that support quantitative analyses of surgical activities, supporting systems engineering or process optimization. METHODS AND MATERIALS Five different similarity metrics are presented and proven. These metrics deal with several dimensions of process compliance in surgery, including granularity, content, time, order, and frequency of surgical activities. The metrics were experimentally validated using 20 clinical data sets each for cataract interventions, craniotomy interventions, and supratentorial tumor resections. The clinical data sets were controllably modified in simulations, which were iterated ten times, resulting in a total of 600 simulated data sets. The simulated data sets were subsequently compared to the original data sets to empirically assess the predictive validity of the metrics. RESULTS We show that the results of the metrics for the surgical process models correlate significantly (p<0.001) with the induced modifications and that all metrics meet predictive validity. The clinical use of the metrics was exemplarily, as demonstrated by assessment of the learning curves of observers during surgical process model acquisition. CONCLUSION Measuring similarity between surgical processes is a complex task. However, metrics for computing the similarity between surgical process models are needed in many uses in the field of medical engineering. These metrics are essential whenever two SPMs need to be compared, such as during the evaluation of technical systems, the education of observers, or the determination of surgical strategies. These metrics are key figures that provide a solid base for medical decisions, such as during validation of sensor systems for use in operating rooms in the future.

Multi-site study of surgical practice in neurosurgery based on surgical process models

Surgical Process Modelling (SPM) was introduced to improve understanding the different parameters that influence the performance of a Surgical Process (SP). Data acquired from SPM methodology is enormous and complex. Several analysis methods based on comparison or classification of Surgical Process Models (SPMs) have previously been proposed. Such methods compare a set of SPMs to highlight specific parameters explaining differences between populations of patients, surgeons or systems. In this study, procedures performed at three different international University hospitals were compared using SPM methodology based on a similarity metric focusing on the sequence of activities occurring during surgery. The proposed approach is based on Dynamic Time Warping (DTW) algorithm combined with a clustering algorithm. SPMs of 41 Anterior Cervical Discectomy (ACD) surgeries were acquired at three Neurosurgical departments; in France, Germany, and Canada. The proposed approach distinguished the different surgical behaviors according to the location where surgery was performed as well as between the categorized surgical experience of individual surgeons. We also propose the use of Multidimensional Scaling to induce a new space of representation of the sequences of activities. The approach was compared to a time-based approach (e.g. duration of surgeries) and has been shown to be more precise. We also discuss the integration of other criteria in order to better understand what influences the way the surgeries are performed. This first multi-site study represents an important step towards the creation of robust analysis tools for processing SPMs. It opens new perspectives for the assessment of surgical approaches, tools or systems as well as objective assessment and comparison of surgeons expertise.