1,721,045 research outputs found
Just a Touch of Digital Twin: Reality-Based Haptic Feedback for Arthroscopic Surgery Simulation
No full textThis thesis addresses haptic modeling of complex contact interactions experienced in arthroscopic surgery. Virtual surgical training simulators are promising for providing safe surgical training without causing harm to the patient and have been described as a paradigm shift in modern surgical education. Recent developments in automatic anatomic model creation are also promising in the creation of patient-specific training simulators, sometimes referred to as digital twins. This requires the estimation of patient-specific material parameters. Haptic feedback has been reported to be a central component in surgical training simulators, but often fails to meet surgeon expectations. To address this shortcoming, this research seeks to explore reality-based modeling of haptic interactions in arthroscopic surgery. The research has resulted in a collection of works that measure characteristic interactions between tissue and surgical instrument, measure constitutive tissue parameters, develop haptic models for real-time surgical training simulations, with user evaluation studies investigating the effect of the methods. The contributions are mainly within haptic technology and applications to arthroscopic surgery.
In Paper 1, the scientific literature was reviewed with respect to simulation of real-time soft tissues, haptic feedback, and estimation of biomechanical parameters. A high-level digital twin concept was also introduced. Paper 2 prototyped a signal fusion method for combining reality-based haptic force signals with finite element-based force signals using a Kalman filter. A simplified one-degree-of-freedom setup was used. Nonlinear haptics with discontinuity were found to improve face validity (p < 0.017) for a partial meniscectomy punch simulation in a user study that evaluated various haptic feedback signals. Advancing to a 3D environment, Paper 3 introduced an arthroscopic tracking instrument that enables measurement of instrument position and contact forces during ex vivo arthroscopic surgery. This instrument was used in Paper 4, together with an adaptive observer method, to es timate the elastic modulus of meniscus tissue in real-time during ex vivo surgery. The estimated moduli were compared with moduli obtained from biomechanical indentation experiments, which are considered reliable, and there were no significant differences (p = 0.585) between the methods. This showed that intraoperative parameter estimation is feasible, but the accuracy of the position sensor needs to be improved. Notably, this experiment provided knowledge on the elastic moduli of meniscus tissue which is needed for finite element-based haptic modeling. Taking advantage of this, Paper 5 presented an interactive surgical simulation of arthroscopic meniscus examination using the Simulation Open Framework Architecture (SOFA). Here, elastic moduli were used with corotational finite element simulation to produce haptic feedback using constraint-based methods. The resulting haptic force signal was verified in ex vivo meniscus examination using the arthroscopic tracking instrument and a meniscus sample that had previously been used for modulus estimation.
Finally, Paper 6 presented a haptic rendering method in which characteristic force signals are used in a 3D simulation environment. The advantage of this method is the ability to tailor the haptic force signal to the desired application. The method was implemented in SOFA, and demonstrated using characteristic force signals from arthroscopic portal creation and meniscus examination, measured using the arthroscopic tracker instrument of Paper 3 on the same ex vivo knees as those used in Paper 4. The resulting haptic forces signals were compared with those of Paper 5 in a user experiment with respect to construct and face validity. Psychophysics experiments found that neither experts nor novices could distinguish between modulus differences of 0.81 MPa, and this was the same for both methods (p < 0.05,U = 36.0; 36.0; 33.0; 31.5). The respective difference thresholds were found to be 1.80 MPa (novice, linear-elastic), 1.47 MPa (novice, reality-based), 0.99 MPa (expert, linear-elastic), and 1.39 MPa (expert, reality-based). Construct validity was established for meniscus examination, but not for arthroscopic portal creation.
In conclusion, reality-based haptic feedback has the potential to model contact interactions experienced in arthroscopic surgery with sufficient accuracy. This can be achieved through constitutive models or through characteristic contact force signals. Both methods have limitations, and the haptic rendering strategy should be carefully selected to reflect the functional tasks they are trying to model. There exist further limitations in haptic feedback for surgical simulation that represent exciting opportunities for future research.Sammendrag
Denne doktorgradsavhandlingen tar for seg modellering av haptisk tilbakemelding i kirurgiske treningssimulatorer, og handler om å gjenskape følelsen til forskjellige vevstyper i leddet. Kirurgiske treningssimulatorer har blitt fremhevet som et trygt alternativ for å øve på kirurgi uten risiko for å skade pasienter. Slike kirurgiske treningssimulatorer er også lovende for å brukes til pasientspesifikk forberedelse og planlegging, noe som krever nøyaktige vevsegenskaper. Slike pasientspesifikke modeller omtales i noen tilfeller som digitale tvillinger. Haptisk tilbakemelding i kirurgiske treningssimulatorer er ansett som en svært viktig komponent, men også noe som er vanskelig å få til. For å forbedre modellering av haptisk tilbakemelding i artroskopisimulatorer benyttes det i denne avhandlingen et rammeverk som kalles virkelighetsbasert metode. Avhandlingen består derfor av samlede verker som omhandler måling av vevsegenskaper, måling av kraftprofiler fra kontakt med vev, utvikling av metoder for haptisk simulering, og brukerundersøkelser som studerer virkningseffekten av metodene.
I den første artikkelen gjennomgår vi vitenskapelig litteratur på området og foreslår et konsept for digital tvilling med søkelys på artroskopi. I den andre artikkelen benyttes enkle prototyper til å teste ut en signalprosesseringsmetode for å kombinere haptiske signaler fra måling og simulering. Det benyttes her et Kalmanfilter. Brukere foretrakk her ulineære kraftprofiler over lineære for å simulere klipping i meniskvev (p < 0.017). I artikkel 3 presenterer vi et kirurgisk instrument med kraft- og posisjonssensorer som gjør det mulig å måle kontaktkrefter samtidig som man opererer ex vivo. I artikkel 4 bruker vi dette instrumentet til å estimere materialstivheter til meniskvev under operasjon på kadaverknær. Det benyttes her en adaptiv observatør til systemidentifikasjon. Disse materialstivhetene sammenlignes så med biomekaniske målinger fra inntrykkstester, og det ble ikke funnet noen signifikant (p = 0.585) forskjell mellom metodene. Dette viser at det er mulig å estimere materialstivheter fra målinger med kirurgiske instrumenter, men posisjonsnøyaktighet bør forbedres. I artikkel 5 bruker vi kunnskapen om materialstivheten til meniskvev til å utvikle en kirurgisk treningssimulator for artroskopisk meniskundersøkelse. Det ble her benyttet sanntids finite-elementsimulering i programvaren SOFA med en lineærelastisk materialmodell. De haptiske kraftsignalene ble verifisert mot kraftmålinger på en lateral menisk ex vivo.
I artikkel 6 presenteres det en teknikk for å bruke målte kraftsignaler til å beregne haptiske krefter ut fra tredimensjonale simuleringskontakter. Fordelen med denne teknikken er å kunne skreddersy det haptiske kraftsignalet til bruksområdet, gjerne basert på målinger. Teknikken ble implementert i SOFA, og demonstrert for etablering av kikhull i knekirurgi, samt i undersøkelse av meniskvev. Kraftsignalene kom fra målinger gjennomført på kadaverknær, og teknikken ble sammenlignet mot metoden fra artikkel 5 i en brukerundersøkelse. Her fant vi at verken erfarne kirurger eller uerfarne deltakere klarte å skille mellom forskjeller i materialstivhet på 0.81 MPa, og dette gjaldt begge metodene (p < 0.05,U = 36.0; 36.0; 33.0; 31.5).
Terskelverdiene for de ulike deltakerne ble estimert til 1.80 MPa (uerfaren, lineærelastisk metode), 1.47 MPa (uerfaren, virkelighetsbasert metode), 0.99 MPa (ekspert, lineærelastisk metode), og 1.39 MPa (ekspert, virkelighetsbasert metode). Eksperter gjorde det bedre enn uerfarne i meniskundersøkelse, men ikke i etablering av kikhull.
Avslutningsvis vil vi fremheve at virkelighetsbasert metode har potensiale for å modellere haptisk tilbakemelding i kirurgiske treningssimulatorer med tilstrekkelig nøyaktighet. Dette kan oppnås enten gjennom materialmodeller, eller gjennom bruk av kraftprofiler. Begge disse metodene har begrensninger som må tas hensyn til, og valg av metode bør tilpasses den kirurgiske operasjonen som modelleres. Det finnes i tillegg andre begrensninger i haptisk teknologi som utgjør spennende områder for fremtidig forskning
Evaluation of Path Planning and Collision Avoidance Algorithms for Autonomous Surface Vehicles
Get PDFRecently, there has been an increased focus on developing an autonomous shipping technology that is safe, trustworthy, and efficient. Some enablers of this technology are strategies to advance sustainability and reducing CO2 emissions. This is achieved by making ships more efficient, increasing safety and reducing the number of accidents caused by human errors on water, and reducing operational costs. However, there are still many challenges to face before autonomous technology on the water becomes a part of our daily life.
A safe and reliable path planning and collision avoidance method is an important component in autonomous shipping and plays a key role in incorporating this technology into our daily lives. Although numerous path planning algorithms for autonomous vessels have been developed, each with its own benefits and limitation, there is no one ultimate path planning and collision avoidance algorithm that is suitable for every vessel, in all water regions and in all scenarios. There is also no unified way of evaluating and comparing these algorithms to find the most suitable one for the chosen use case.
In this context, the main purpose of this research is to propose a strategy for a unified evaluation and comparison of path planning and collision avoidance algorithms. To achieve this goal, it is essential to first gain an understanding of path planning and collision avoidance as a part of the autonomous surface vehicle’s guidance, navigation, and control system. There are two main application cases. First, it could be used as an offline benchmarking tool to evaluate and compare the algorithms. Second, it could be used online on an actual vessel to select the safest and most efficient path in the planning phase based on the current situation.
This thesis presents an evaluation simulator platform (ESP) for evaluating path planning and collision avoidance algorithm performance for autonomous surface vehicles. In particular, the work focuses on an extended collision risk assessment (ECRA) method for evaluating the generated paths from a safety perspective. The testing results indicate that the proposed approach could be used for autonomous path planning and collision avoidance algorithm evaluation and comparison with some improvements.In reference to IEEE copyrighted material which is used with permission in this thesis, the IEEE does not endorse any of NTNU’s products or services. Internal or personal use of this material is permitted. If interested in reprinting/republishing IEEE copyrighted material for advertising or promotional purposes or for creating new collective works for resale or redistribution, please go to http://www.ieee.org/publications_standards/publications/rights/rights_link.html to learn how to obtain a License from RightsLink
Going Beyond Counting First Authors in Author Co-citation Analysis
Get PDFThe present study examines one of the fundamental aspects of author co-citation analysis (ACA) - the way co-citation
counts are defined. Co-citation counting provides the data on which all subsequent statistical analyses and mappings
are based, and we compare ACA results based on two different types of co-citation counting - the traditional type that
only counts the first one among a cited work's authors on the one hand and a non-traditional type that takes into
account the first 5 authors of a cited work on the other hand. Results indicate that the picture produced through this non-traditional author co-citation counting contains more coherent author groups and is therefore considerably clearer. However, this picture represents fewer specialties in the research field being studied than that produced through the traditional first-author co-citation counting when the same number of top-ranked authors is selected and analyzed. Reasons for these effects are discussed
Variations on the Author
Get PDF“Variations on the Author” discusses two of Eduardo Coutinho’s recent films (Um Dia na Vida, from 2010, and Últimas Conversas, posthumously released in 2015) and their contribution to the general question of documentary authorship. The director’s filmography is characterized by a consistent yet self-effacing form of authorial self-inscription: Coutinho often features as an interviewer that rather than express opinions propels discourses; an interviewer that is good at listening. This mode of self-inscription characterizes him as an author who is not expressive but who is nonetheless markedly present on the screen. In Um Dia na Vida, however, Coutinho is completely absent form the image, while Últimas Conversas, on the contrary, includes a confessional prologue that moves the director from the margins to the center of his films. This article examines the ways in which these works stand out in the filmography of a director who offers new insights into the notion of cinematic authorship
Appropriate Similarity Measures for Author Cocitation Analysis
Get PDFWe provide a number of new insights into the methodological discussion about author cocitation analysis. We first argue that the use of the Pearson correlation for measuring the similarity between authors’ cocitation profiles is not very satisfactory. We then discuss what kind of similarity measures may be used as an alternative to the Pearson correlation. We consider three similarity measures in particular. One is the well-known cosine. The other two similarity measures have not been used before in the bibliometric literature. Finally, we show by means of an example that our findings have a high practical relevance.information science;Pearson correlation;cosine;similarity measure;author cocitation analysis
Dispelling the Myths Behind First-author Citation Counts
Get PDFWe conducted a full-scale evaluative citation analysis study of scholars in the XML research field to explore just how different from each other author rankings resulting from different citation counting methods actually are, and to demonstrate the capability of emerging data and tools on the Web in supporting more realistic citation counting methods. Our results contest some common arguments for the continued
use of first-author citation counts in the evaluation of scholars, such as high correlations between author rankings by first-author citation counts and other citation
counting methods, and high costs of using more realistic citation counting methods that are not well-supported by the ISI databases. It is argued that increasingly available digital full text research papers make it possible for citation analysis studies to go beyond what the ISI databases have directly supported and to employ more
sophisticated methods
Deep learning for fault detection of guardrails
Get PDFMennesker er uendelig flinke til å trekke ut informasjon fra bilder i kompleks natur å finne og klassifisere et objekt. I løpet av de siste årene har algoritmer og metoder blitt presentert for å gjøre det samme. Avanserte algoritmer brukes til komplekse oppgaver, kjent i områder relatert til selvkjørende biler, sporing, klassifisering, osv. Feltet i maskinlæring, kalt datamaskinvisjon, går ut for å trekke ut den enorme informasjonen som finnes i bilder. De eksponentielle fremskritt i antall dataenheter i GPU-er (Graphical Computing Units), har gjort det mulig å lage algoritmer som var utenkelige for et tiår siden uten superdatamaskiner. De siste årene har det vært en økning i nevralt nettverk for å løse en rekke oppgaver, der Convolutional Neural Network er kjent for sin ytelse på bildebehandling. Også ressurssterke IT-selskaper og forskningsfakulteter som har fordelen med tilgjengelig regnekraft, har bidratt med avanserte, skreddersydde CNN-modeller for en rekke datasynoppgaver. Denne oppgaven undersøkte de nyeste CNN-modellene for å hjelpe fagarbeidere i å opprettholde rekkverk over hele Norge ved å automatisere den visuelle inspeksjonen som blir utført av disse arbeiderne. Dagens visuelle inspeksjon gjøres ved å kjøre bil i hastigheter mellom 1-15 km / t og utføre visuell inspeksjon gjennom kameraet eller bilvindu for å oppdage feilene. Dette arbeidet tar sikte på å implementere en topp moderne arkitektur, som velges ved å evaluere en rekke arkitekturer i henhold til målene som er satt i oppgaven. Eksperimenteringen ble gjort ved å samle inn data, forbehandle og implementere det nevrale nettverket. Modellen oppnådde en lovende mAP på 71%Humans are immeasurably good at extracting information from images in complex scenery
to detect and classify an object. In recent years, algorithms and methods have been presented to do the same. Advanced algorithms are used for complex tasks, famously in areas related to self-driving cars, tracking, classifying, etc. The field in machine learning called computer vision heads out to extract the vast information present in images. The exponential advances in numbers of computing units in GPUs (Graphical Computing Units), have made it possible to create algorithms that were unimaginable a decade ago
without supercomputers. Recent years have seen an increase in neural network for solving a variety of tasks, where the Convolutional Neural Network is known for its performance on image processing.
Also, resourceful IT-companies and research faculties having the advantage of available computational power, have contributed with state-of-the-art, costume tailored CNN models for a variety of computer vision tasks. This thesis investigated the state-of-the-art CNN models to aid skilled workers in maintaining guardrails across Norway to automating the visual inspection done by these workers. Today’s visual inspection is done by driving in speeds between 1-15 km/h and performing visual inspection through the camera or the
car window to detect the faults. This work sets out to implement a state-of-the-art architecture, which is chosen by evaluating a variety of architectures according to the objectives set by the thesis. The experimentation was done by collecting data, pre-processing, and implementing the neural network. The model achieved a promising mAP of 71
Deep learning for fault detection of guardrails
No full textMennesker er uendelig flinke til å trekke ut informasjon fra bilder i kompleks natur å finne og klassifisere et objekt. I løpet av de siste årene har algoritmer og metoder blitt presentert for å gjøre det samme. Avanserte algoritmer brukes til komplekse oppgaver, kjent i områder relatert til selvkjørende biler, sporing, klassifisering, osv. Feltet i maskinlæring, kalt datamaskinvisjon, går ut for å trekke ut den enorme informasjonen som finnes i bilder. De eksponentielle fremskritt i antall dataenheter i GPU-er (Graphical Computing Units), har gjort det mulig å lage algoritmer som var utenkelige for et tiår siden uten superdatamaskiner. De siste årene har det vært en økning i nevralt nettverk for å løse en rekke oppgaver, der Convolutional Neural Network er kjent for sin ytelse på bildebehandling. Også ressurssterke IT-selskaper og forskningsfakulteter som har fordelen med tilgjengelig regnekraft, har bidratt med avanserte, skreddersydde CNN-modeller for en rekke datasynoppgaver. Denne oppgaven undersøkte de nyeste CNN-modellene for å hjelpe fagarbeidere i å opprettholde rekkverk over hele Norge ved å automatisere den visuelle inspeksjonen som blir utført av disse arbeiderne. Dagens visuelle inspeksjon gjøres ved å kjøre bil i hastigheter mellom 1-15 km / t og utføre visuell inspeksjon gjennom kameraet eller bilvindu for å oppdage feilene. Dette arbeidet tar sikte på å implementere en topp moderne arkitektur, som velges ved å evaluere en rekke arkitekturer i henhold til målene som er satt i oppgaven. Eksperimenteringen ble gjort ved å samle inn data, forbehandle og implementere det nevrale nettverket. Modellen oppnådde en lovende mAP på 71
- …
