Abstract. Many current medical image analysis problems involve learning thousands or even millions of model parameters from extremely few samples. Employing sparse models provides an effective means for handling the curse of dimensionality, but other propitious properties beyond sparsity are typically not modeled. In this paper, we propose a simple approach, generalized sparse regularization (GSR), for incorporating domain-specific knowledge into a wide range of sparse linear models, such as the LASSO and group LASSO regression models. We demonstrate the power of GSR by building anatomically-informed sparse classifiers that additionally model the intrinsic spatiotemporal characteristics of brain activity for fMRI classification. We validate on real data and show how prior-informed sparse classifiers outperform standard classifiers, such as SVM and a number of sparse linear classifiers, both in terms of prediction accuracy and result interpretability. Our results illustrate the addedvalue in facilitating flexible integration of prior knowledge beyond sparsity in large-scale model learning problems.Keywords: brain decoding, fMRI classification, prior-informed learning, sparse optimization, spatiotemporal regularization IntroductionRecent years witnessed a surging interest in exploiting sparsity [1-9] to handle the ever-increasing scale and complexity of current medical image analysis problems [10,11]. Oftentimes, one is faced with exceedingly more predictors than samples. Under such ill-conditioned settings, incorporating sparsity into model learning proved to be of enormous benefits. In particular, enforcing sparsity enables model parameters associated with irrelevant predictors to be implicitly removed, i.e. shrunk to exactly zero [1], which reduces overfitting thus enhances model generalizability. Learning parsimonious models by imposing sparsity also simplifies result interpretation [1], which is of utmost importance in most medical studies. Since the advent of the least absolute shrinkage and selection operator (LASSO) regression model [1], where Tibshirani showed that penalizing the l 1 norm induces sparsity in the regression coefficients, numerous powerful variants were subsequently proposed [2][3][4][5][6][7][8][9]. Zou and Hastie, for instance, proposed the elastic net penalty [2], which retains the sparse property of LASSO but additionally encourages correlated

Abstract. We propose a novel method for applying active learning strategies to interactive 3D image segmentation. Active learning has been recently introduced to the field of image segmentation. However, so far discussions have focused on 2D images only. Here, we frame interactive 3D image segmentation as a classification problem and incorporate active learning in order to alleviate the user from choosing where to provide interactive input. Specifically, we evaluate a given segmentation by constructing an "uncertainty field" over the image domain based on boundary, regional, smoothness and entropy terms. We then calculate and highlight the plane of maximal uncertainty in a batch query step. The user can proceed to guide the labeling of the data on the query plane, hence actively providing additional training data where the classifier has the least confidence. We validate our method against random plane selection showing an average DSC improvement of 10% in the first five plane suggestions (batch queries). Furthermore, our user study shows that our method saves the user 64% of their time, on average.

No abstract

BackgroundRecent pathological studies have suggested that thalamic degeneration may represent a site of non-dopaminergic degeneration in Parkinson's Disease (PD). Our objective was to determine if changes in the thalami could be non-invasively detected in structural MRI images obtained from subjects with Parkinson disease (PD), compared to age-matched controls.ResultsNo significant differences in volume were detected in the thalami between eighteen normal subjects and eighteen PD subjects groups. However significant (p < 0.03) shape differences were detected between the Left vs. Right thalami in PD, between the left thalami in PD and controls, and between the right thalami in PD and controls using a recently-developed, spherical harmonic-based representation.ConclusionSystematic changes in thalamic shape can be non-invasively assessed in PD in vivo. Shape changes, in addition to volume changes, may represent a new avenue to assess the progress of neurodegenerative processes. Although not directly discernable at the resolution of standard MRI, previous pathological studies would suggest that the shape changes detected in this study represent degeneration in the centre median-parafascicular (CM-Pf) complex, an area known to represent selective non-dopaminergic degeneration in PD.

Motor symptoms of Parkinson's disease (PD) do not appear until the majority of dopaminergic cells in the substantia nigra pars compacta are lost, suggesting significant redundancy or compensation in the motor systems affected by PD. Using functional magnetic resonance imaging, we examined whether compensation in PD is manifested by changes in amplitude and/or spatial extent of activity within normal networks (active motor reserve) and/or newly recruited regions [novel area recruitment (NAR)]. Ten PD subjects off and on medication and 10 age-matched controls performed a visually guided sinusoidal force task at 0.25, 0.5 and 0.75 Hz. Regression was used to determine the combination of regions where activation amplitude scaled linearly with movement speed in controls. We then determined the activation of PD subjects in this network, as well as the corresponding PD network. To measure the spatial variance of activation, we used an invariant spatial feature approach. Control subjects monotonically increased activity within striato-thalamo-cortical and cerebello-thalamo-cortical regions with increasing movement speed. In PD subjects, the activity of this network at low speeds was similar to that in controls at higher speeds. Additionally, PD subjects off medication demonstrated NARs of the bilateral cerebellum and primary motor cortex, which were incompletely normalized by levodopa. Our results suggest that PD subjects tap into motor reserve, increase the spatial extent of activation and demonstrate NAR to maintain near-normal motor output.