Celina Imielinska

Methodology for evaluating image-segmentation algorithms

The purpose of this paper is to describe a framework for evaluating image segmentation algorithms. Image segmentation consists of object recognition and delineation. For evaluating segmentation methods, three factors - precision (reproducibility), accuracy (agreement with truth, validity), and efficiency (time taken) - need to be considered for both recognition and delineation. To assess precision, we need to choose a figure of merit, repeat segmentation considering all sources of variation, and determine variations in figure of merit via statistical analysis. It is impossible usually to establish true segmentation. Hence, to assess accuracy, we need to choose a surrogate of true segmentation and proceed as for precision. In determining accuracy, it may be important to consider different landmark areas of the structure to be segmented depending on the application. To assess efficiency, both the computational and the user time required for algorithm and operator training and for algorithm execution should be measured and analyzed. Precision, accuracy, and efficiency are interdependent. It is difficult to improve one factor without affecting others. Segmentation methods must be compared based on all three factors. The weight given to each factor depends on application.

Hybrid Segmentation of Anatomical Data

We propose new hybrid methods for automated segmentation of radiological patient data and the Visible Human data. In this paper, we integrate boundary-based and region-based segmentation methods which amplifies the strength but reduces the weakness of both approaches. The novelty comes from combining a boundary-based method, the deformable model-based segmentation with region-based segmentation methods, the fuzzy connectedness and Voronoi Diagram-based segmentation, to develop hybrid methods that yield high precision, accuracy and efficiency. This work is a part of a NLM funded effort to provide a fully implemented and tested Visible Human Project Segmentation and Registration Toolkit (Insight).

Segmentation and Evaluation of Adipose Tissue from Whole Body MRI Scans

Accurate quantification of total body and the distribution of regional adipose tissue using manual segmentation is a challenging problem due to the high variation between manual delineations. Manual segmentation also requires highly trained experts with knowledge of anatomy. We present a hybrid segmentation method that provides robust delineation results for adipose tissue from whole body MRI scans. A formal evaluation of accuracy of the segmentation method is performed. This semi-automatic segmentation algorithm reduces significantly the time required for quantification of adipose tissue, and the accuracy measurements show that the results are close to the ground truth obtained from manual segmentations.

Modeling Real-Time 3-D Lung Deformations for Medical Visualization

In this paper, we propose a physics-based and physiology-based approach for modeling real-time deformations of 3-D high-resolution polygonal lung models obtained from high-resolution computed tomography (HRCT) images of normal human subjects. The physics-based deformation operator is nonsymmetric, which accounts for the heterogeneous elastic properties of the lung tissue and spatial-dynamic flow properties of the air. An iterative approach is used to estimate the deformation with the deformation operator initialized based on the regional alveolar expandability, a key physiology-based parameter. The force applied on each surface node is based on the airflow pattern inside the lungs, which is known to be based on the orientation of the human subject. The validation of lung dynamics is done by resimulating the lung deformation and comparing it with HRCT data and computing force applied on each node derived from a 4-D HRCT dataset of a normal human subject using the proposed deformation operator and verifying its gradient with the orientation.

Post-carotid endarterectomy neurocognitive decline is associated with cerebral blood flow asymmetry on post-operative magnetic resonance perfusion brain scans

Abstract Objective: Up to 25% of patients experience subtle declines in post-operative neurocognitive function following, otherwise uncomplicated, carotid endarterectomy (CEA). We sought to determine if post-CEA neurocognitive deficits are associated with cerebral blood flow (CBF) abnormalities on post-operative MR perfusion brain scans. Methods: We enrolled 22 CEA patients to undergo a battery of neuropsychometric tests pre-operatively and on post-operative day 1 (POD 1). Neurocognitive dysfunction was defined as a two standard deviation decline in performance in comparison to a similarly aged control group of lumbar laminectomy patients. All patients received MR perfusion brain scans on POD 1 that were analysed for asymmetries in CBF distribution. One patient experienced a transient ischemic attack within 24 hours before the procedure and was excluded from our analysis. Results: Twenty-nine percent of CEA patients demonstrated neurocognitive dysfunction on POD 1. One hundred percent of those patients with cognitive deficits demonstrated CBF asymmetry, in contrast to only 27% of those patients without cognitive impairment. Post-CEA cognitive dysfunction was significantly associated with CBF abnormalities (RR=3.75, 95% CI: 1.62–8.67, p=0.004). Conclusion: Post-CEA neurocognitive dysfunction is significantly associated with post-operative CBF asymmetry. These results support the hypothesis that post-CEA cognitive impairment is caused by cerebral hemodynamic changes. Further work exploring the relationship between CBF and post-CEA cognitive dysfunction is needed.

Enabling a continuum of virtual environment experiences

We define a virtual environment as a set of surroundings that appear to a user through computer-generated sensory stimuli. The level of immersion-or sense of being in another world-that a user experiences within a VE relates to how much stimuli the computer delivers to the user. Thus, one can classify VEs along a virtuality continuum, which ranges from the real world to an entirely computer-generated environment. We present a technology that allows seamless transitions between levels of immersion in VEs. Milgram and Kishino (1994) first proposed the concept of a virtuality continuum in the context of visual displays. The concept of a virtuality continuum extends to multimodal VEs, which combine multiple sensory stimuli, including 3D sound and haptic capability, leading to a multidimensional virtuality continuum. Emerging applications will benefit from multiple levels of immersion, requiring innovative multimodal technologies and the ability to traverse the multidimensional virtuality continuum.

Physically-based Deformation of High-Resolution 3D Lung Models for Augmented Reality based Medical Visualization

Visualization tools using Augmented Reality Environments are effective in applications related to medical training, prognosis and expert interaction. Such medical visualization tools can also provide key visual insights on the physiology of deformable anatomical organs (e.g. lungs). In this paper we propose a deformation method that facilitates physically-based elastostatic deformations of 3D highresolution polygonal models. The implementation of the deformation method as a pre-computation approach is shown for a 3D high-resolution lung model. The deformation is represented as an integration of the applied force and the local elastic property assigned to the 3D lung model. The proposed deformation method shows faster convergence to equilibrium as compared to other physically-based simulation methods. The proposed method also accounts for the anisotropic tissue elastic properties. The transfer functions are formulated in such a way that they overcome stiffness effects during deformations.

Distributed Augmented Reality With 3-D Lung Dynamics—A Planning Tool Concept

Augmented reality (AR) systems add visual information to the world by using advanced display techniques. The advances in miniaturization and reduced hardware costs make some of these systems feasible for applications in a wide set of fields. We present a potential component of the cyber infrastructure for the operating room of the future: a distributed AR-based software-hardware system that allows real-time visualization of three-dimensional (3-D) lung dynamics superimposed directly on the patients body. Several emergency events (e.g., closed and tension pneumothorax) and surgical procedures related to lung (e.g., lung transplantation, lung volume reduction surgery, surgical treatment of lung infections, lung cancer surgery) could benefit from the proposed prototype

A Novel Drill Set for the Enhancement and Assessment of Robotic Surgical Performance

BACKGROUND There currently exist several training modules to improve performance during video-assisted surgery. The unique characteristics of robotic surgery make these platforms an inadequate environment for the development and assessment of robotic surgical performance. METHODS Expert surgeons (n=4) (>50 clinical robotic procedures and >2 years of clinical robotic experience) were compared to novice surgeons (n=17) (<5 clinical cases and limited laboratory experience) using the da Vinci Surgical System. Seven drills were designed to simulate clinical robotic surgical tasks. Performance score was calculated by the equation Time to Completion + (minor error) x 5 + (major error) x 10. The Robotic Learning Curve (RLC) was expressed as a trend line of the performance scores corresponding to each repeated drill. RESULTS Performance scores for experts were better than novices in all 7 drills (p<0.05). The RLC for novices reflected an improvement in scores (p<0.05). In contrast, experts demonstrated a flat RLC for 6 drills and an improvement in one drill (p=0.027). CONCLUSION This new drill set provides a framework for performance assessment during robotic surgery. The inclusion of particular drills and their role in training robotic surgeons of the future awaits larger validation studies.

Asymmetry Analysis in Rodent Cerebral Ischemia Models

RATIONALE AND OBJECTIVES An automated method for identification and segmentation of acute/subacute ischemic stroke, using the inherent bi-fold symmetry in brain images, is presented. An accurate and automated method for localization of acute ischemic stroke could provide physicians with a mechanism for early detection and potentially faster delivery of effective stroke therapy. MATERIALS AND METHODS Segmentation of ischemic stroke was performed on magnetic resonance (MR) images of subacute rodent cerebral ischemia. Eight adult male Wistar rats weighing 225-300 g were anesthetized with halothane in a mix of 70% nitrous oxide/30% oxygen. Animal core temperature was maintained at 37 degrees C during the entire surgical procedure, including occlusion of the middle cerebral artery (MCA) and the 90-minute post-reperfusion period. To confirm cerebral ischemia, transcranial measurements of cerebral blood flow were performed with laser-Doppler flowmetry, using 15-mm flexible fiberoptic Doppler probes attached to the skull over the MCA territory. Animal MR scans were performed at 1.5 T using a knee coil. Three experts performed manual tracing of the stroke regions for each rat, using the histologic-stained slices to guide delineation of stroke regions. A strict tracing protocol was followed that included multiple (three) tracings of each stroke region. The volumetric MR image data were processed for each rat by computing the axis of symmetry and extracting statistical dissimilarities. A nonparametric Wilcoxon rank sum test operating on paired windows in opposing hemispheres identified seeds in the pixels exhibiting statistically significant bi-fold mirror asymmetry. Two brain reference maps were used for analysis: an absolute difference map (ADM) and a statistical difference map (SDM). Although an ADM simply displays the absolute difference by subtracting one brain hemisphere from its reflection, SDM highlights regions by labeling pixels exhibiting statistically significant asymmetry. RESULTS To assess the accuracy of the proposed segmentation method, the surrogate ground truth (the stroke tracing data) was compared to the results of our proposed automated segmentation algorithm. Three accuracy segmentation metrics were utilized: true-positive volume fraction (TPVF), false-positive volume fraction (FPVF), and false-negative volume fraction (FNVF). The mean value of the TPVF for our segmentation method was 0.8877; 95% CI 0.7254 to 1.0500; the mean FPVF was 0.3370, 95% CI -0.0893 to 0.7633; the mean FNVF was 0.1122, 95% CI -0.0502 to 0.2747. CONCLUSIONS Unlike most segmentation methods that require some degree of manual intervention, our segmentation algorithm is fully automated and highly accurate in identifying regions of brain asymmetry. This approach is attractive for numerous neurologic applications where the operators intervention should be minimal or null.