Atlas-based segmentation is used in radiotherapy planning to accelerate the delineation of organs at risk (OARs). Atlas selection has been proposed to improve the performance of segmentation, assuming that the more similar the atlas is to the patient, the better the result. It follows that the larger the database of atlases from which to select, the better the results should be. This paper seeks to estimate a clinically achievable expected performance under this assumption. Assuming a perfect atlas selection, an extreme value theory has been applied to estimate the accuracy of single-atlas and multi-atlas segmentation given a large database of atlases. For this purpose, clinical contours of most common OARs on computed tomography of the head and neck ( N=316 ) and thoracic ( N=280 ) cases were used. This paper found that while for most organs, perfect segmentation cannot be reasonably expected, auto-contouring performance of a level corresponding to clinical quality could be consistently expected given a database of 5000 atlases under the assumption of perfect atlas selection.

The goal of this work is to reliably and accurately localize anatomical landmarks in 3-D computed tomography scans, particularly for the deformable registration of whole-body scans, which show huge variation in posture, and the spatial distribution of anatomical features. Parts-based graphical models (GM) have shown attractive properties for this task because they capture naturally anatomical relationships between landmarks. Unfortunately, standard GMs are learned from manually annotated training images and the quantity of landmarks is limited by the high cost of expert annotation. We propose a novel method that automatically learns new corresponding landmarks from a database of 3-D whole-body CT scans, using a limited initial set of expert-labeled ground-truth landmarks. The newly learned landmarks, called B-landmarks, are used to build enriched GMs. We compare our method of deformable registration based on such GM landmarks to a conventional deformable registration method and to a "baseline" state-of-the-art GM. The results show our method finds new relevant anatomical correspondences and improves by up to 35% the matching accuracy of highly variable skeletal and soft-tissue landmarks of clinical interest.

Recently, Kadir and Brady proposed a method for estimating probability density functions (PDFs) for digital signals which they call the Nonparametric (NP) Windows method. The method involves constructing a continuous space representation of the discrete space and sampled signal by using a suitable interpolation method. NP Windows requires only a small number of observed signal samples to estimate the PDF and is completely data driven. In this short paper, we first develop analytical formulae to obtain the NP Windows PDF estimates for 1D, 2D, and 3D signals, for different interpolation methods. We then show that the original procedure to calculate the PDF estimate can be significantly simplified and made computationally more efficient by a judicious choice of the frame of reference. We have also outlined specific algorithmic details of the procedures enabling quick implementation. Our reformulation of the original concept has directly demonstrated a close link between the NP Windows method and the Kernel Density Estimator.

We propose a method for reliably and accurately identifying anatomical landmarks in 3D CT volumes based on dense matching of parts-based graphical models. Such a system can be used to establish reliable correspondences in medical images which can be useful on their own or as part of more complex processing e.g. atlas building. We propose and investigate novel methods for efficiently optimizing parameters of appearance models for landmark localization in 3D images. We also investigate the trade-off between the number of model parameters and registration accuracy. We present results for the localization of 22 landmarks in clinical 3D CT volumes of cancer patients and optimization of part-specific patch scales. Over-fitting is likely due to an intrinsically high variability of the data and a limited labeled training and test set, here 83 scans, so we employ a rigorous bootstrap analysis to validate the results. The average mean and maximum registration error over all landmarks is reduced by 31% and 25% for the optimized model, compared to an empirically determined baseline. Additionally, we show a significantly improved performance over standard methods as the number of free parameters increases from an isotropic patch scale shared by all parts, to specific anisotropic patch scales learnt for each part in the model.

Recently, the Nonparametric (NP) Windows has been proposed to estimate the statistics of real 1D and 2D signals. NP Windows is accurate, because it is equivalent to sampling images at a high (infinite) resolution for an assumed interpolation model. This paper extends the proposed approach to consider joint distributions of image-pairs. Second, Green's Theorem is used to simplify the previous NP Windows algorithm. Finally, a resolution-aware NP Windows algorithm is proposed to improve robustness to relative scaling between an image pair. Comparative testing of 2D image registration was performed using translation-only and affine transformations. Although more expensive than other methods, NP Windows frequently demonstrated superior performance for bias (distance between ground truth and global maximum) and frequency of convergence. Unlike other methods, the number of samples and the number of bins have little effect on NP Windows and the prior selection of a kernel is not required.

We address the problem of estimating statistics in regions of interest (ROIs) containing both whole and partial pixels. Such ROIs arise frequently in vision problems such as segmentation and registration. For example, even where the control points of an ROI, say the vertices of a polygon, are forcibly aligned with the pixel grid, the connecting edges will rarely do so. In medical image analysis, for instance, this can be a cause of significant error. More generally, any cost function that includes statistics estimated from the image will often exhibit irregularities due to such partial pixels. Our proposed solution addresses this problem by correctly accounting for the partial pixel area. Moreover, the method has no arbitrary parameters such as bin widths or kernel sizes. It implicitly addresses the issue of independence and gives rise to continuous density estimates whose quality is, in principle at least, independent of the number of pixels in the ROI. We present results to compare our proposed method with conventional techniques such as weighted histograms and Parzen windowing.

This paper presents a flexible monocular system capable of recognising sign lexicons far greater in number than previous approaches. The power of the system is due to four key elements: (i) Head and hand detection based upon boosting which removes the need for temperamental colour segmentation; (ii) A body centred description of activity which overcomes issues with camera placement, calibration and user; (iii) A two stage classification in which stage I generates a high level linguistic description of activity which naturally generalises and hence reduces training; (iv) A stage II classifier bank which does not require HMMs, further reducing training requirements. The outcome of which is a system capable of running in real-time, and generating extremely high recognition rates for large lexicons with as little as a single training instance per sign. We demonstrate classification rates as high as 92 % for a lexicon of 164 words with extremely low training requirements outperforming previous approaches where thousands of training examples are required. 1

We present a novel non-parametric unsupervised segmentation algorithm based on region competition (Zhu and Yuille, 1996); but implemented within a level sets framework (Osher and Sethian, 1988). The key novelty of the algorithm is that it can solve N /spl ges/ 2 class segmentation problems using just one embedded surface; this is achieved by controlling the merging and splitting behaviour of the level sets according to a minimum description length (MDL) (Leclerc (1989) and Rissanen (1985)) cost function. This is in contrast to N class region-based level set segmentation methods to date which operate by evolving multiple coupled embedded surfaces in parallel (Chan et al., 2002). Furthermore, it operates in an unsupervised manner; it is necessary neither to specify the value of N nor the class models a-priori. We argue that the level sets methodology provides a more convenient framework for the implementation of the region competition algorithm, which is conventionally implemented using region membership arrays due to the lack of a intrinsic curve representation. Finally, we generalise the Gaussian region model used in standard region competition to the non-parametric case. The region boundary motion and merge equations become simple expressions containing cross-entropy and entropy terms.