Minimal Invasive Surgeries procedures have gathered momentum in the last few decades. One subarea where the research efforts are concentrated is the field of steerable needles. These instruments can achieve curved insertion trajectories by increased maneuverability; which is a profitable feature, especially for invasive neurological surgeries. Previous work on the development of automatically controlled steering of flexible needles has shown the need to measure accurately the pose of the needle's tip. This quantity greatly influences the control actions, making the pose estimation a domain of major interest. This research was limited to a programmable bevel needle, made of four interlocked segments. Measurements were collected from NDI Aurora sensors located on each of the needle segments and after used on the estimation algorithms. The intention of this work is to estimate the complete pose of the needle, in order to provide reliable data to the control system. To fulfill these requirements, a geometric process has been designed using the framework of dual quaternions, as an optimal method to perform coordinates transformations, to determine the pose of the needle's tip. An Unscented Kalman Filter was used as an additional processing stage aiming at improving the pose estimation. The algorithms were first tested using data collected from experiments executed on 3-D printed mock-up rail that predefined the trajectory and later from experiments performed in phantom tissue. The methods based on geometric calculations successfully provides the pose of the tip for different preset mock-up rail trajectories. An additional stage composed of an Unscented Kalman Filter and a Kinematic Model of Underwater Vehicle was able to correct and estimate unmeasured states of the model.
Le procedure di chirurgia mini invasiva hanno raccolto slancio negli ultimi decenni. Una sotto-area in cui sono concentrati gli sforzi di ricerca è il campo degli aghi orientabili. Questi strumenti possono ottenere traiettorie di inserzione curve aumentando la manovrabilità; che è una caratteristica redditizia, specialmente per gli interventi neurologici invasivi. Il lavoro precedente sullo sviluppo della guida automatica controllata degli aghi flessibili ha dimostrato la necessità di misurare con precisione la posa della punta degli aghi. Questa quantità influenza notevolmente le azioni di controllo, rendendo la stima della posa un dominio di maggiore interesse. Questa ricerca era limitata ad un ago con smussatura programmabile, costituito da quattro segmenti interbloccati. Le misurazioni sono state raccolte da sensori NDI Aurora situati su ciascuno dei segmenti dell'ago e successivamente utilizzati negli algoritmi di stima. L'intenzione di questo lavoro è stimare la posa completa dell'ago, al fine di fornire dati affidabili al sistema di controllo. Per soddisfare questi requisiti, un processo geometrico è stato sviluppato utilizzando la struttura di dual-quaternion, come metodo ottimale per eseguire trasformazioni di coordinate, per determinare la posa della punta degli aghi. Inoltre, un filtro Kalman è stato utilizzato come ulteriore fase di elaborazione per migliorare la stima della posa. Gli algoritmi sono stati testati per la prima volta utilizzando i dati raccolti dagli esperimenti eseguiti su un guida mock-up stampato in 3-D che ha predefinito la traiettoria e successivamente da esperimenti eseguiti su gelatina. I metodi basati su calcoli geometrici forniscono con successo la posa della punta per diverse traiettorie di mock-up preimpostate. Uno stadio aggiuntivo composto da un filtro di Kalman e un modello cinematico di un veicolo subacqueo è stato in grado di correggere e stimare gli stati non misurati del modello.
Pose estimation for robotically assisted steerable needle
GRANADA ESLAVA, DANY FABIAN
2018/2019
Abstract
Minimal Invasive Surgeries procedures have gathered momentum in the last few decades. One subarea where the research efforts are concentrated is the field of steerable needles. These instruments can achieve curved insertion trajectories by increased maneuverability; which is a profitable feature, especially for invasive neurological surgeries. Previous work on the development of automatically controlled steering of flexible needles has shown the need to measure accurately the pose of the needle's tip. This quantity greatly influences the control actions, making the pose estimation a domain of major interest. This research was limited to a programmable bevel needle, made of four interlocked segments. Measurements were collected from NDI Aurora sensors located on each of the needle segments and after used on the estimation algorithms. The intention of this work is to estimate the complete pose of the needle, in order to provide reliable data to the control system. To fulfill these requirements, a geometric process has been designed using the framework of dual quaternions, as an optimal method to perform coordinates transformations, to determine the pose of the needle's tip. An Unscented Kalman Filter was used as an additional processing stage aiming at improving the pose estimation. The algorithms were first tested using data collected from experiments executed on 3-D printed mock-up rail that predefined the trajectory and later from experiments performed in phantom tissue. The methods based on geometric calculations successfully provides the pose of the tip for different preset mock-up rail trajectories. An additional stage composed of an Unscented Kalman Filter and a Kinematic Model of Underwater Vehicle was able to correct and estimate unmeasured states of the model.File | Dimensione | Formato | |
---|---|---|---|
2019_04_Granada_ThesisDoc.pdf
accessibile in internet per tutti
Descrizione: Thesis Doc
Dimensione
17.48 MB
Formato
Adobe PDF
|
17.48 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/147331