Patient-specific Anatomical Models Research Articles

A magnetic resonance spectroscopy driven initialization scheme for active shape model based prostate segmentation

Segmentation of the prostate boundary on clinical images is useful in a large number of applications including calculation of prostate volume pre- and post-treatment, to detect extra-capsular spread, and for creating patient-specific anatomical models. Manual segmentation of the prostate boundary is, however, time consuming and subject to inter- and intra-reader variability. T2-weighted (T2-w) magnetic resonance (MR) structural imaging (MRI) and MR spectroscopy (MRS) have recently emerged as promising modalities for detection of prostate cancer in vivo. MRS data consists of spectral signals measuring relative metabolic concentrations, and the metavoxels near the prostate have distinct spectral signals from metavoxels outside the prostate. Active Shape Models (ASM's) have become very popular segmentation methods for biomedical imagery. However, ASMs require careful initialization and are extremely sensitive to model initialization. The primary contribution of this paper is a scheme to automatically initialize an ASM for prostate segmentation on endorectal in vivo multi-protocol MRI via automated identification of MR spectra that lie within the prostate. A replicated clustering scheme is employed to distinguish prostatic from extra-prostatic MR spectra in the midgland. The spatial locations of the prostate spectra so identified are used as the initial ROI for a 2D ASM. The midgland initializations are used to define a ROI that is then scaled in 3D to cover the base and apex of the prostate. A multi-feature ASM employing statistical texture features is then used to drive the edge detection instead of just image intensity information alone. Quantitative comparison with another recent ASM initialization method by Cosio showed that our scheme resulted in a superior average segmentation performance on a total of 388 2D MRI sections obtained from 32 3D endorectal in vivo patient studies. Initialization of a 2D ASM via our MRS-based clustering scheme resulted in an average overlap accuracy (true positive ratio) of 0.60, while the scheme of Cosio yielded a corresponding average accuracy of 0.56 over 388 2D MR image sections. During an ASM segmentation, using no initialization resulted in an overlap of 0.53, using the Cosio based methodology resulted in an overlap of 0.60, and using the MRS-based methodology resulted in an overlap of 0.67, with a paired Student's t-test indicating statistical significance to a high degree for all results. We also show that the final ASM segmentation result is highly correlated (as high as 0.90) to the initialization scheme.

Magnitude and Direction of Pulsatile Displacement Forces Acting on Thoracic Aortic Endografts

To assess 3-dimensional (3D) pulsatile displacement forces (DF) acting on thoracic endografts using 3D computational techniques. A novel computational method to quantitate the pulsatile 3D flow and pressure fields and aortic wall dynamics in patient-specific anatomical models based on cardiac-gated computed tomography (CT) scans was used to construct simulations of the proximal and mid-descending thoracic aorta. Endografts of varying lengths and diameters were implanted in these patient-specific models. The magnitude and direction of the DF vector were calculated and expressed in Newtons (N). This DF included the effects of both the pressure and shearing stresses of blood. The magnitude of DF increased with endografts of increasing diameter and length. A 36-mm endograft in the mid-descending aorta had a mean DF of 21.7 N with a peak systolic DF of 27.8 N and an end-diastolic DF of 16.7 N. Conversely, a 30-mm endograft in the proximal descending aorta had a mean DF of 14.9 N, with peak systolic and end-diastolic DFs of 18.9 and 11.5, respectively. The orientation of the DF acting on the endograft varied depending on aortic angulation and tortuosity; in general, the vector was perpendicular to the greater curvature of the endograft rather than along the downstream longitudinal centerline axis of the aorta as is commonly believed. The DF vector pointed primarily in the cranial direction for the proximal descending endograft and in the sideways direction for the mid-descending endograft simulation. Furthermore, it was shown that elevated pressure plays an important role in the magnitude and direction of DF; an increase in mean blood pressure resulted in an approximately linearly proportional increase in DF. The orientation of the DF varies depending on curvature and location of the endograft, but in all instances, it is in the cranial rather than caudal direction on axial imaging. This is counter to the intuitive notion that displacement forces act in the downward direction of blood flow. Therefore, we postulate that migration of thoracic endografts may be different from abdominal endografts since it may involve upward rather than downward movement of the graft. Computational methods can enhance the understanding of the magnitude and orientation of the loads experienced in vivo by thoracic aortic endografts and therefore improve their design and performance.

Automated identification of anatomical landmarks on 3D bone models reconstructed from CT scan images

Identification of anatomical landmarks on skeletal tissue reconstructed from CT/MR images is indispensable in patient-specific preoperative planning (tumour referencing, deformity evaluation, resection planning, and implant alignment and anchoring) as well as intra-operative navigation (bone registration and instruments referencing). Interactive localisation of landmarks on patient-specific anatomical models is time-consuming and may lack in repeatability and accuracy. We present a computer graphics-based method for automatic localisation and identification (labelling) of anatomical landmarks on a 3D model of bone reconstructed from CT images of a patient. The model surface is segmented into different landmark regions (peak, ridge, pit and ravine) based on surface curvature. These regions are labelled automatically by an iterative process using a spatial adjacency relationship matrix between the landmarks. The methodology has been implemented in a software program and its results (automatically identified landmarks) are compared with those manually palpated by three experienced orthopaedic surgeons, on three 3D reconstructed bone models. The variability in location of landmarks was found to be in the range of 2.15–5.98 mm by manual method (inter surgeon) and 1.92–4.88 mm by our program. Both methods performed well in identifying sharp features. Overall, the performance of the automated methodology was better or similar to the manual method and its results were reproducible. It is expected to have a variety of applications in surgery planning and intra-operative navigation.

Training on N.O.T.E.S.: From history we learn

Surgical errors occurring early in the learning curve of laparoscopic surgery providers delayed the uptake and progress of minimally invasive surgery (MIS) for years. This taught us a valuable lesson; innovations in surgical techniques should not be rapidly implemented until all aspects including applicability, feasibility and safety have been fully tested. In 2005, the Natural Orifice Surgery Consortium for Assessment and Research (NOSCAR) published a white paper highlighting the barriers to NOTES development and identifying key elements for its progress. One of these elements is the training of future providers. Proficiency-based, virtual reality simulation will offer a feasible alternative to animal testing once the safety and efficacy parameters of NOTES are established. Recent advances in imaging including computed tomography (CT) scanning, magnetic resonance imaging (MRI) scanning, and ultrasound (US) scanning can offer improved image registration and real-time tracking. Combining these advanced imaging technologies with the newly designed virtual reality simulators will result in a fully comprehensive simulation curriculum which will offer a unique facility for future NOTES providers to train anytime, anywhere, and as much as they need to in order to achieve the pre-set proficiency levels for a variety of NOTES procedures. Furthermore they will incorporate patient-specific anatomical models obtained from patient imaging and uploaded onto the simulator to ensure face reliability and validity assurance. Training in a clean, safe environment with proximate feedback and performance analysis will help accelerate the learning curve and therefore improve patients' safety and outcomes in order to maximize the benefits of innovative access procedures such as NOTES.

Patient-Tailored Cerebral Arterial Model for Simulating Neurovascular Intervention (2nd Report, In Vitro Reproduction of Membranous Vascular Structure Accompanied with Elastic Deformation)

In this paper, we propose an in vitro patient-specific anatomical model of human cerebral artery, a hardware platform for simulating endovascular intervention for making diagnosis and surgical trainings. Proposed vessel model hollowly reproduces 3-dimensional configuration of vascular lumens with elastic membranous configuration as a transparent silicone structure, using fluoroscopic information, which is obtained with CT (Computed Tomography), MRI (Magnetic Resonance Imaging) and other imaging modalities. This vessel model also reproduces the physical characteristics of artery tissue, such as elastic modulus, Poisson's ratio and friction coefficient, within respectively 10% of errors. Consequently, it allows practically simulating procedures, which is accompanied by realistic vascular deformation, realistic feel of operating catheters and realistic behavior of surgical instruments (e.g. slip/stick motion and unwinding motion). Presented vasculature model should provide a helpful platform for interventionalists to make preliminary simulation, for young trainees to practice surgical procedures and for developers to evaluate surgical robots, device and other various medical instruments.

An In Vitro Patient-Specific Biological Model of the Cerebral Artery Reproduced with a Membranous Configuration for Simulating Endovascular Intervention

We propose an in vitro patient-specific anatomical model of the human cerebral artery and its simulation of endovascular intervention, a potent treatment modality for cerebrovascular diseases. Our proposed model reproduces the 3-dimensional vasculature lumen, using computed tomography (CT) and magnetic resonance (MR) fluoroscopic information, within a thin artery-like membranous configuration having material properties close to arterial tissue. This cerebral arterial model reproduces an exceedingly realistic surgical feel, dynamic vascular deformation and, other important aspects involving endovascular intervention, realizing a highly realistic surgical simulation. We also propose another vasculature model that reproduces the subarachnoid space around the cerebral arteries. This version simulates endovascular intervention realistically. The model is compatible with current major imaging modalities such as CT, MR, and transcranial Doppler (TDC), and should provide effective platforms for applications, such as diagnosis, surgical planning, medical training, hemodynamic analysis and medical system development and evaluation, especially surgical robots.

Robotic whole body stereotactic radiosurgery: clinical advantages of the Cyberknife® integrated system

Radiosurgery is defined as the delivery of high doses of ionising radiation, in mono- or hypo- fractionated treatments, to destroy tumours or focal areas of pathology. The clinical requirements of designing a radiosurgical treatment system include providing: a) a highly precise beam delivery to targets located throughout the body, b) a highly conformal dose distribution, c) the ability to irradiate both small and/or large complex-shaped lesions while minimising the dose to adjacent radiosensitive tissues and d) the ability to interactively track lesion motion due to normal patient motion. To accomplish this, the CyberKnife radiosurgery system has pioneered in this area by taking advantage of the inherent geometrical targeting precision of a commercial arm-based robotic system carrying a compact X-band linear accelerator and integrated with X-ray imaging and visualisation feedback systems. The arm-mounted linear accelerator, equipped with patient specific anatomical models, registered to the patient in real-time with image guidance, dynamically and safely delivers conformal and homogeneous radiation for therapeutic benefit. This paper details the components of the CyberKnife system and their integration in the clinical workflow of radiosurgery.

Identifying and reducing errors with surgical simulation.

The major determinant of a patient's safety and outcome is the skill and judgment of the surgeon. While knowledge base and decision processing are evaluated during residency, technical skills-which are at the core of the profession-are not evaluated. Innovative state of the art simulation devices that train both surgical tasks and skills, without risk to patients, should allow for the detection and analysis of errors and "near misses". Studies have validated the use of a sophisticated endoscopic sinus surgery simulator (ES3) for training residents on a procedural basis. Assessments are proceeding as to whether the integration of a comprehensive ES3 training programme into the residency curriculum will have long term effects on surgical performance and patient outcomes. Using various otolaryngology residencies, subjects are exposed to mentored training on the ES3 as well as to minimally invasive trainers such as the MIST-VR. Technical errors are identified and quantified on the simulator and intraoperatively. Through a web based database, individual performance can be compared against a national standard. An upgraded version of the ES3 will be developed which will support patient specific anatomical models. This advance will allow study of the effects of simulated rehearsal of patient specific procedures (mission rehearsal) on patient outcomes and surgical errors during the actual procedure. The information gained from these studies will help usher in the next generation of surgical simulators that are anticipated to have significant impact on patient safety.

A procedure for computing patient-specific anatomical models for finite element-based surgical simulation

This paper presents ongoing research on a semi-automatic method for computing, from CT and MR data, patient-specific anatomical models used in surgical simulation. Up to now, virtual anatomical models for surgical simulation have been highly task-specific and obtained by highly elaborated computer–user interaction, including segmentation and meshing. We propose a minimally supervised procedure for extracting from a set of MR and CT scans a highly descriptive anatomical mesh, leading not merely to one generic model, but to a family of patient-specific models. This procedure integrates a distortion-tolerant mutual information MR-CT registration, a new tissue classification exploiting global spatial cues, a simplex-based surface meshing model to identify and triangulate the relevant anatomical boundaries, and a final almost-regular tetrahedralization of tissue volumes surrounded by these boundaries. This procedure is designed to be extensible to other surgical applications.

Computers in imaging and guided surgery

The authors review the main technical issues in computer-integrated surgery (CIS) systems. They illustrate with examples of working systems the state of the art in the field and provide perspectives on deployment and future developments. They discuss the structure of CIS systems. At the core is a computer (or network of computers) running various modeling and analysis processes, including image and sensor processing, creation and manipulation of patient-specific anatomical models, surgical planning, visualization, monitoring, and control of surgical processes. After receiving information about the patient from medical imaging devices, some CIS systems act directly on the patient using specialized robots or other computer controlled therapy devices.

Patient-specific anatomic models from three dimensional medical image data for clinical applications in surgery and endoscopy.

Virtual surgery and endoscopy use computer-generated volume renderings and/or models created from 3D medical image scans (CT or MRI) of individual patients. The patient's anatomy, including organs and other internal structures of interest, are then traversed in a virtual "fly-through," giving nearly the same visual impression as if the corresponding real organ was being examined intraoperatively, or as if an actual video or fiberoptic endoscopic procedure was being performed. Such virtual examinations may provide capabilities and information not possible or available in physical examinations. The potential is to provide a noninvasive computer-aided treatment plan or diagnostic screening procedure to augment or replace conventional invasive procedures. With sophisticated image processing and computational analysis, it is possible to perform realistic and useful simulations of surgical and endoscopic procedures, including "virtual dissection and resection" and "virtual biopsy." Surgical margins can be accurately assessed and differential tissue diagnoses made based upon spectral or other information contained in the patient-specific images and models.

Patient-specific anatomic models. Geometric surface generation from three-dimensional medical images using a specified polygonal budget.

Virtual reality offers the promise of highly interactive, natural control of the visualization process, greatly enhancing the scientific value of the data produced by medical imaging systems. Due to the computational and real time update requirements of virtual reality interfaces, however, the complexity of polygonal surfaces which can be displayed is limited. In this paper, we present a novel method for the production of a polygonal surface containing a pre-specified number of polygons from volumetric data. To preserve surface detail, we extract a set of curvature weights from the volumetric data and use these weights as the input vectors to a 2-D Kohonen network. The adaptation of the network to the input vectors results in a display surface that preserves useful detail relative to the number of polygons used.

Numerical simulation of annular-phased arrays of dipoles for hyperthermia of deep-seated tumors.

We have used the finite-difference time domain (FDTD) method to calculate the SAR distributions from an annular-phased array of eight dipole antennas coupled through water "boluses" in anatomically based three-dimensional models of the human body. We evaluated the effect of tapered bolus chambers, frequency (100-120 MHz), dipole length (17-30 cm), and phase and amplitude of power to the various dipoles on the ability to focus energy in the region of deep-seated tumors in the prostate and the liver. Assuming tumor conductivity and permittivity to be similar or slightly higher than surrounding normal tissues, calculations indicate that adjustment of the noted parameters should result in considerable improvement in focusing of SAR distributions in tumor-bearing regions. If such calculations can be shown to correctly predict empirical measurements from complex inhomogeneous (although not necessarily anatomically correct) phantoms, they may be useful for hyperthermia treatment planning based on patient-specific anatomic models.