RATIONALE AND OBJECTIVES: We present a new method for automatic brain extraction on structural magnetic resonance images, based on a multi-atlas registration framework. MATERIALS AND METHODS: Our method addresses fundamental challenges of multi-atlas approaches. To overcome the difficulties arising from the variability of imaging characteristics between studies, we propose a study-specific template selection strategy, by which we select a set of templates that best represent the anatomical variations within the data set. Against the difficulties of registering brain images with skull, we use a particularly adapted registration algorithm that is more robust to large variations between images, as it adaptively aligns different regions of the two images based not only on their similarity but also on the reliability of the matching between images. Finally, a spatially adaptive weighted voting strategy, which uses the ranking of Jacobian determinant values to measure the local similarity between the template and the target images, is applied for combining coregistered template masks. RESULTS: The method is validated on three different public data sets and obtained a higher accuracy than recent state-of-the-art brain extraction methods. Also, the proposed method is successfully applied on several recent imaging studies, each containing thousands of magnetic resonance images, thus reducing the manual correction time significantly. CONCLUSIONS: The new method, available as a stand-alone software package for public use, provides a robust and accurate brain extraction tool applicable for both clinical use and large population studies.

Recent neuroanatomical pattern classification studies have attempted to individually classify cases with psychotic disorders using morphometric MRI data in an automated fashion. However, this approach has not been tested in population-based samples, in which variable patterns of comorbidity and disease course are typically found. We aimed to evaluate the diagnostic accuracy (DA) of the above technique to discriminate between incident cases of first-episode schizophrenia identified in a circumscribed geographical region over a limited period of time, in comparison with next-door healthy controls. Sixty-two cases of first-episode schizophrenia or schizophreniform disorder and 62 age, gender and educationally-matched controls underwent 1.5 T MRI scanning at baseline, and were naturalistically followed-up over 1 year. T1-weighted images were used to train a high-dimensional multivariate classifier, and to generate both spatial maps of the discriminative morphological patterns between groups and ROC curves. The spatial map discriminating first-episode schizophrenia patients from healthy controls revealed a complex pattern of regional volumetric abnormalities in the former group, affecting fronto-temporal-occipital gray and white matter regions bilaterally, including the inferior fronto-occipital fasciculus, as well as the third and lateral ventricles. However, an overall modest DA (73.4%) was observed for the individual discrimination between first-episode schizophrenia patients and controls, and the classifier failed to predict 1-year prognosis (remitting versus non-remitting course) of first-episode schizophrenia (DA=58.3%). In conclusion, using a "real world" sample recruited with epidemiological methods, the application of a neuroanatomical pattern classifier afforded only modest DA to classify first-episode schizophrenia subjects and next-door healthy controls, and poor discriminative power to predict the 1-year prognosis of first-episode schizophrenia.

Cross-subject image registration is the building block for many cardiac studies. In the literature, it is often handled by voxel-wise registration methods. However, studies are lacking to show which methods are more accurate and stable in this context. Aiming at answering this question, this paper evaluates 12 popular registration methods and validates a recently developed method DRAMMS [16] in the context of cross-subject cardiac registration. Our dataset consists of short-axis end-diastole cardiac MR images from 24 subjects, in which non-cardiac structures are removed. Each registration method was applied to all 552 image pairs. Registration accuracy is approximated by Jaccard overlap between deformed expert annotation of source image and the corresponding expert annotation of target image. This accuracy surrogate is further correlated with deformation aggressiveness, which is reflected by minimum, maximum and range of Jacobian determinants. Our study shows that DRAMMS [16] scores high in accuracy and well balances accuracy and aggressiveness in this dataset, followed by ANTs [13], MI-FFD [14], Demons [15], and ART [12]. Our findings in cross-subject cardiac registrations echo those findings in brain image registrations [7].

In this paper, we adapt spectral signatures for capturing morphological changes over time. Advanced techniques for capturing temporal shape changes frequently rely on first registering the sequence of shapes and then analyzing the corresponding set of high dimensional deformation maps. Instead, we propose a simple encoding motivated by the observation that small shape deformations lead to minor refinements in the spectral signature composed of the eigenvalues of the Laplace operator. The proposed encoding does not require registration, since spectral signatures are invariant to pose changes. We apply our representation to the shapes of the ventricles extracted from 22 cine MR scans of healthy controls and Tetralogy of Fallot patients. We then measure the accuracy score of our encoding by training a linear classifier, which outperforms the same classifier based on volumetric measurements.