The reconstruction of cortical surfaces from brain magnetic resonance imaging (MRI) scans is essential for quantitative analyses thickness and sulcal morphology. Although traditional deep learning-based algorithmic pipelines exist this purpose, they have two major drawbacks: lengthy runtimes multiple hours (traditional) or intricate post-processing, such as mesh extraction topology correction (deep learning-based). In work, we address both these issues propose Vox2Cortex, a algorithm that...

Explaining predictions of black-box neural networks is crucial when applied to decision-critical tasks. Thus, attribution maps are commonly used identify important image regions, despite prior work showing that humans prefer explanations based on similar examples. To this end, ProtoPNet learns a set class-representative feature vectors (prototypes) for case-based reasoning. During inference, similarities latent features prototypes linearly classified form and provided explain the similarity....

Predicting future brain states is crucial for understanding healthy aging and neurodegenerative diseases. Longitudinal MRI registration, a cornerstone such analyses, has long been limited by its inability to forecast developments, reliance on extensive, dense longitudinal data, the need balance registration accuracy with temporal smoothness. In this work, we present \emph{TimeFlow}, novel framework that overcomes all these challenges. Leveraging U-Net architecture conditioning inspired...

Abstract Reconstructing the cortex from longitudinal magnetic resonance imaging (MRI) is indispensable for analyzing morphological alterations in human brain. Despite recent advancement of cortical surface reconstruction with deep learning, challenges arising data are still persistent. Especially lack strong spatiotemporal point correspondence between highly convoluted brain surfaces hinders downstream analyses, as local morphology not directly comparable if anatomical location matched...

Despite recent advances in medical image generation, existing methods struggle to produce anatomically plausible 3D structures. In synthetic brain magnetic resonance images (MRIs), characteristic fissures are often missing, and reconstructed cortical surfaces appear scattered rather than densely convoluted. To address this issue, we introduce Cor2Vox, the first diffusion model-based method that translates continuous shape priors MRIs. achieve this, leverage a Brownian bridge process which...

Mesh-based cortical surface reconstruction is essential for neuroimaging, enabling precise measurements of brain morphology such as thickness. Establishing vertex correspondence between individual meshes and group templates allows vertex-level comparisons, but traditional methods require time-consuming post-processing steps to achieve correspondence. While deep learning has improved accuracy in reconstruction, optimizing not been the focus prior work. We introduce Vox2Cortex with...

Abdominal organ segmentation from CT and MRI is an essential prerequisite for surgical planning computer-aided navigation systems. It challenging due to the high variability in shape, size, position of abdominal organs. Three-dimensional numeric representations shapes with point-wise correspondence a template are further important quantitative statistical analyses thereof. Recently, template-based surface extraction methods have shown promising advances direct mesh reconstruction volumetric...

Reconstructing the cortex from longitudinal MRI is indispensable for analyzing morphological changes in human brain. Despite recent disruption of cortical surface reconstruction with deep learning, challenges arising data are still persistent. Especially lack strong spatiotemporal point correspondence hinders downstream analyses due to introduced noise. To address this issue, we present V2C-Long, first dedicated learning-based method MRI. In contrast existing methods, V2C-Long surfaces...

Magnetic resonance imaging (MRI) is critical for diagnosing neurodegenerative diseases, yet accurately assessing mild cortical atrophy remains a challenge due to its subtlety. Automated cortex reconstruction, paired with healthy reference ranges, aids in pinpointing pathological atrophy, their generalization limited by biases from image acquisition and processing. We introduce the concept of stochastic self-reconstruction (SCSR) that creates subject-specific taking MRI-derived thicknesses as...

The reconstruction of cortical surfaces is a prerequisite for quantitative analyses the cerebral cortex in magnetic resonance imaging (MRI). Existing segmentation-based methods separate surface registration from extraction, which computationally inefficient and prone to distortions. We introduce Vox2Cortex-Flow (V2C-Flow), deep mesh-deformation technique that learns deformation field brain template an MRI scan. To this end, we present geometric neural network models deformation-describing...

The field of reinforcement learning offers a large variety concepts and methods to tackle sequential decision-making problems. This has become so that choosing an algorithm for task at hand can be challenging. In this work, we streamline the process reinforcement-learning algorithms action-distribution families. We provide structured overview existing their properties, as well guidelines when choose which methods. An interactive version these is available online https://rl-picker.github.io/.

Meningeal lymphatic vessels (MLVs) are responsible for the drainage of waste products from human brain. An impairment in their functionality has been associated with aging as well brain disorders like multiple sclerosis and Alzheimer's disease. However, MLVs have only recently described first time magnetic resonance imaging (MRI), ramified structure renders manual segmentation particularly difficult. Further, there is no consistent notion appearance, human-annotated MLV structures contain a...

The reconstruction of cortical surfaces from brain magnetic resonance imaging (MRI) scans is essential for quantitative analyses thickness and sulcal morphology. Although traditional deep learning-based algorithmic pipelines exist this purpose, they have two major drawbacks: lengthy runtimes multiple hours (traditional) or intricate post-processing, such as mesh extraction topology correction (deep learning-based). In work, we address both these issues propose Vox2Cortex, a algorithm that...

Abdominal organ segmentation from CT and MRI is an essential prerequisite for surgical planning computer-aided navigation systems. It challenging due to the high variability in shape, size, position of abdominal organs. Three-dimensional numeric representations shapes with point-wise correspondence a template are further important quantitative statistical analyses thereof. Recently, template-based surface extraction methods have shown promising advances direct mesh reconstruction volumetric...

Explaining predictions of black-box neural networks is crucial when applied to decision-critical tasks. Thus, attribution maps are commonly used identify important image regions, despite prior work showing that humans prefer explanations based on similar examples. To this end, ProtoPNet learns a set class-representative feature vectors (prototypes) for case-based reasoning. During inference, similarities latent features prototypes linearly classified form and provided explain the similarity....

The reconstruction of cerebral cortex surfaces from brain MRI scans is instrumental for the analysis morphology and detection cortical thinning in neurodegenerative diseases like Alzheimer's disease (AD). Moreover, a fine-grained atrophy patterns, parcellation into individual regions required. For former task, powerful deep learning approaches, which provide highly accurate tissue boundaries input seconds, have recently been proposed. However, these methods do not come with ability to...