Mikkel B. Stegmann

Segmentation of anatomical structures in chest radiographs using supervised methods: a comparative study on a public database.

The task of segmenting the lung fields, the heart, and the clavicles in standard posterior-anterior chest radiographs is considered. Three supervised segmentation methods are compared: active shape models, active appearance models and a multi-resolution pixel classification method that employs a multi-scale filter bank of Gaussian derivatives and a k-nearest-neighbors classifier. The methods have been tested on a publicly available database of 247 chest radiographs, in which all objects have been manually segmented by two human observers. A parameter optimization for active shape models is presented, and it is shown that this optimization improves performance significantly. It is demonstrated that the standard active appearance model scheme performs poorly, but large improvements can be obtained by including areas outside the objects into the model. For lung field segmentation, all methods perform well, with pixel classification giving the best results: a paired t-test showed no significant performance difference between pixel classification and an independent human observer. For heart segmentation, all methods perform comparably, but significantly worse than a human observer. Clavicle segmentation is a hard problem for all methods; best results are obtained with active shape models, but human performance is substantially better. In addition, several hybrid systems are investigated. For heart segmentation, where the separate systems perform comparably, significantly better performance can be obtained by combining the results with majority voting. As an application, the cardio-thoracic ratio is computed automatically from the segmentation results. Bland and Altman plots indicate that all methods perform well when compared to the gold standard, with confidence intervals from pixel classification and active appearance modeling very close to those of a human observer. All results, including the manual segmentations, have been made publicly available to facilitate future comparative studies.

Eye typing using Markov and active appearance models

We propose a non-intrusive eye tracking system intended for the use of everyday gaze typing using web cameras. We argue that high precision in gaze tracking is not needed for on-screen typing due to natural language redundancy. This facilitates the use of low-cost video components for advanced multi-modal interactions based on video tracking systems. Robust methods are needed to track the eyes using web cameras due to the poor image quality. A realtime tracking scheme using a mean-shift color tracker and an Active Appearance Model of the eye is proposed. It is possible from this model to infer the state of the eye such as eye corners and the pupil location under scale and rotational changes.

Multi-band Modelling of Appearance

Earlier work has demonstrated generative models capable of synthesising near photo-realistic grey-scale images of objects. These models have been augmented with colour information, and recently with edge information. This paper extends the active appearance model framework by modelling the appearance of both derived feature bands and an intensity band. As a special case of feature-band augmented appearance modelling we propose a dedicated representation with applications to face segmentation. The representation addresses a major problem within face recognition by lowering the sensitivity to lighting conditions. Results show that the localisation accuracy of facial features is considerably increased using this appearance representation under diffuse and directional lighting and at multiple scales.

Corpus callosum atrophy is associated with mental slowing and executive deficits in subjects with age-related white matter hyperintensities: the LADIS Study

Background: Previous research has indicated that corpus callosum atrophy is associated with global cognitive decline in neurodegenerative diseases, but few studies have investigated specific cognitive functions. Objective: To investigate the role of regional corpus callosum atrophy in mental speed, attention and executive functions in subjects with age-related white matter hyperintensities (WMH). Methods: In the Leukoaraiosis and Disability Study, 567 subjects with age-related WMH were examined with a detailed neuropsychological assessment and quantitative magnetic resonance imaging. The relationships of the total corpus callosum area and its subregions with cognitive performance were analysed using multiple linear regression, controlling for volume of WMH and other confounding factors. Results: Atrophy of the total corpus callosum area was associated with poor performance in tests assessing speed of mental processing—namely, trail making A and Stroop test parts I and II. Anterior, but not posterior, corpus callosum atrophy was associated with deficits of attention and executive functions as reflected by the symbol digit modalities and digit cancellation tests, as well as by the subtraction scores in the trail making and Stroop tests. Furthermore, semantic verbal fluency was related to the total corpus callosum area and the isthmus subregion. Conclusions: Corpus callosum atrophy seems to contribute to cognitive decline independently of age, education, coexisting WMH and stroke. Anterior corpus callosum atrophy is related to the frontal-lobe-mediated executive functions and attention, whereas overall corpus callosum atrophy is associated with the slowing of processing speed.

Unsupervised motion-compensation of multi-slice cardiac perfusion MRI

This paper presents a novel method for registration of single and multi-slice cardiac perfusion MRI. Utilising off-line computer intensive analyses of variance and clustering in an annotated training set, the presented method is capable of providing registration without any manual interaction in less than a second per frame. Changes in image intensity during the bolus passage are modelled by a slice-coupled active appearance model, which is augmented with a cluster analysis of the training set. Landmark correspondences are optimised using the MDL framework due to Davies et al. Image search is verified and stabilised using perfusion specific prior models of pose and shape estimated from training data. Qualitative and quantitative validation of the method is carried out using 2000 clinical quality, short-axis, perfusion MR slice images, acquired from 10 freely breathing patients with acute myocardial infarction. Despite evident perfusion deficits and varying image quality in the limited training set, a leave-one-out cross-validation of the method showed a mean point to curve distance of 1.25+/-0.36 pixels for the left and right ventricle combined. We conclude that this learning-based method holds great promise for the automation of cardiac perfusion investigations, due to its accuracy, robustness and generalisation ability.

Clinical significance of corpus callosum atrophy in a mixed elderly population.

Corpus callosum (CC) is the main tract connecting the hemispheres, but the clinical significance of CC atrophy is poorly understood. The aim of this work was to investigate clinical and functional correlates of CC atrophy in subjects with age-related white matter changes (ARWMC). In 569 elderly subjects with ARWMC from the Leukoaraiosis And DISability (LADIS) study, the CC was segmented on the normalised mid-sagittal magnetic resonance imaging (MRI) slice and subdivided into five regions. Correlations between the CC areas and subjective memory complaints, mini mental state examination (MMSE) score, history of depression, geriatric depression scale (GDS) score, subjective gait difficulty, history of falls, walking speed, and total score on the short physical performance battery (SPPB) were analyzed. Significant correlations between CC atrophy and MMSE, SPPB, and walking speed were identified, and the CC areas were smaller in subjects with subjective gait difficulty. The correlations remained significant after correction for ARWMC grade. In conclusion, CC atrophy was independently associated with impaired global cognitive and motor function in subjects with ARWMC.

Sparse modeling of landmark and texture variability using the orthomax criterion

In the past decade, statistical shape modeling has been widely popularized in the medical image analysis community. Predominantly, principal component analysis (PCA) has been employed to model biological shape variability. Here, a reparameterization with orthogonal basis vectors is obtained such that the variance of the input data is maximized. This property drives models toward global shape deformations and has been highly successful in fitting shape models to new images. However, recent literature has indicated that this uncorrelated basis may be suboptimal for exploratory analyses and disease characterization. This paper explores the orthomax class of statistical methods for transforming variable loadings into a simple structure which is more easily interpreted by favoring sparsity. Further, we introduce these transformations into a particular framework traditionally based on PCA; the Active Appearance Models (AAMs). We note that the orthomax transformations are independent of domain dimensionality (2D/3D etc.) and spatial structure. Decompositions of both shape and texture models are carried out. Further, the issue of component ordering is treated by establishing a set of relevant criteria. Experimental results are given on chest radiographs, magnetic resonance images of the brain, and face images. Since pathologies are typically spatially localized, either with respect to shape or texture, we anticipate many medical applications where sparse parameterizations are preferable to the conventional global PCA approach.

Mid-sagittal plane and mid-sagittal surface optimization in brain MRI using a local symmetry measure

This paper describes methods for automatic localization of the mid-sagittal plane (MSP) and mid-sagittal surface (MSS). The data used is a subset of the Leukoaraiosis And DISability (LADIS) study consisting of three-dimensional magnetic resonance brain data from 62 elderly subjects (age 66 to 84 years). Traditionally, the mid-sagittal plane is localized by global measures. However, this approach fails when the partitioning plane between the brain hemispheres does not coincide with the symmetry plane of the head. We instead propose to use a sparse set of profiles in the plane normal direction and maximize the local symmetry around these using a general-purpose optimizer. The plane is parameterized by azimuth and elevation angles along with the distance to the origin in the normal direction. This approach leads to solutions confirmed as the optimal MSP in 98 percent of the subjects. Despite the name, the mid-sagittal plane is not always planar, but a curved surface resulting in poor partitioning of the brain hemispheres. To account for this, this paper also investigates an optimization strategy which fits a thin-plate spline surface to the brain data using a robust least median of squares estimator. Albeit computationally more expensive, mid-sagittal surface fitting demonstrated convincingly better partitioning of curved brains into cerebral hemispheres.

The AAM-API: An Open Source Active Appearance Model Implementation

This paper presents a public domain implementation of the Active Appearance Model framework and gives examples using it for segmentation and analysis of medical images. The software is open source, designed with efficiency in mind, and has been thoroughly tested and evaluated in several medical and non-medical applications.

Motion-compensation of cardiac perfusion MRI using a statistical texture ensemble

This paper presents a novel method for segmentation of cardiac perfusion MRI. By performing complex analyses of variance and clustering in an annotated training set off-line, the presented method provides real-time segmentation in an on-line setting. This renders the method feasible for e.g. analysis of large image databases or for live non rigid motion-compensation in modern MR scanners. Changes in image intensity during the bolus passage is modelled by an Active Appearance Model augmented with a cluster analysis of the training set and priors on pose and shape. Preliminary validation of the method is carried out using 250 MR perfusion images, acquired without breath-hold from five subjects. Quantitative and qualitative results show high accuracy, given the limited number of subjects.