RESUMO
Modern microscopy techniques can be used to investigate soft nano-objects at the nanometer scale. However, time-consuming microscopy measurements combined with low numbers of observable polydisperse objects often limit the statistics. We propose a method for identifying the most representative objects from their respective point clouds. These point cloud data are obtained, for example, through the localization of single emitters in super-resolution fluorescence microscopy. External stimuli, such as temperature, can cause changes in the shape and properties of adaptive objects. Due to the demanding and time-consuming nature of super-resolution microscopy experiments, only a limited number of temperature steps can be performed. Therefore, we propose a deep generative model that learns the underlying point distribution of temperature-dependent microgels, enabling the reliable generation of unlimited samples with an arbitrary number of localizations. Our method greatly cuts down the data collection effort across diverse experimental conditions, proving invaluable for soft condensed matter studies.
RESUMO
The remarkable performance of the Transformer architecture in natural language processing has recently also triggered broad interest in Computer Vision. Among other merits, Transformers are witnessed as capable of learning long-range dependencies and spatial correlations, which is a clear advantage over convolutional neural networks (CNNs), which have been the de facto standard in Computer Vision problems so far. Thus, Transformers have become an integral part of modern medical image analysis. In this review, we provide an encyclopedic review of the applications of Transformers in medical imaging. Specifically, we present a systematic and thorough review of relevant recent Transformer literature for different medical image analysis tasks, including classification, segmentation, detection, registration, synthesis, and clinical report generation. For each of these applications, we investigate the novelty, strengths and weaknesses of the different proposed strategies and develop taxonomies highlighting key properties and contributions. Further, if applicable, we outline current benchmarks on different datasets. Finally, we summarize key challenges and discuss different future research directions. In addition, we have provided cited papers with their corresponding implementations in https://github.com/mindflow-institue/Awesome-Transformer.
Assuntos
Benchmarking , Aprendizagem , Humanos , Redes Neurais de ComputaçãoRESUMO
Foundational models, pretrained on a large scale, have demonstrated substantial success across non-medical domains. However, training these models typically requires large, comprehensive datasets, which contrasts with the smaller and more specialized datasets common in biomedical imaging. Here we propose a multi-task learning strategy that decouples the number of training tasks from memory requirements. We trained a universal biomedical pretrained model (UMedPT) on a multi-task database including tomographic, microscopic and X-ray images, with various labeling strategies such as classification, segmentation and object detection. The UMedPT foundational model outperformed ImageNet pretraining and previous state-of-the-art models. For classification tasks related to the pretraining database, it maintained its performance with only 1% of the original training data and without fine-tuning. For out-of-domain tasks it required only 50% of the original training data. In an external independent validation, imaging features extracted using UMedPT proved to set a new standard for cross-center transferability.
Assuntos
Processamento de Imagem Assistida por Computador , Humanos , Processamento de Imagem Assistida por Computador/métodos , Bases de Dados Factuais , Diagnóstico por Imagem/métodos , Algoritmos , Aprendizado de MáquinaRESUMO
Ultra high frequency (UHF) ultrasound enables the visualization of very small structures that cannot be detected by conventional ultrasound. The utilization of UHF imaging as a new imaging technique for the 3D-in-vivo chorioallantoic membrane (CAM) model can facilitate new insights into tissue perfusion and survival. Therefore, human renal cystic tissue was grafted onto the CAM and examined using UHF ultrasound imaging. Due to the unprecedented resolution of UHF ultrasound, it was possible to visualize microvessels, their development, and the formation of anastomoses. This enabled the observation of anastomoses between human and chicken vessels only 12 h after transplantation. These observations were validated by 3D reconstructions from a light sheet microscopy image stack, indocyanine green angiography, and histological analysis. Contrary to the assumption that the nutrient supply of the human cystic tissue and the gas exchange happens through diffusion from CAM vessels, this study shows that the vasculature of the human cystic tissue is directly connected to the blood vessels of the CAM and perfusion is established within a short period. Therefore, this in-vivo model combined with UHF imaging appears to be the ideal platform for studying the effects of intravenously applied therapeutics to inhibit renal cyst growth.
Assuntos
Membrana Corioalantoide , Rim Policístico Autossômico Dominante , Ultrassonografia , Animais , Membrana Corioalantoide/irrigação sanguínea , Membrana Corioalantoide/diagnóstico por imagem , Humanos , Rim Policístico Autossômico Dominante/diagnóstico por imagem , Ultrassonografia/métodos , Galinhas , Rim/diagnóstico por imagem , Rim/irrigação sanguínea , Imageamento Tridimensional/métodosRESUMO
While neurosurgical interventions are frequently used in laboratory mice, refinement efforts to optimize analgesic management based on multimodal approaches appear to be rather limited. Therefore, we compared the efficacy and tolerability of combinations of the non-steroidal anti-inflammatory drug carprofen, a sustained-release formulation of the opioid buprenorphine, and the local anesthetic bupivacaine with carprofen monotherapy. Female and male C57BL/6J mice were subjected to isoflurane anesthesia and an intracranial electrode implant procedure. Given the multidimensional nature of postsurgical pain and distress, various physiological, behavioral, and biochemical parameters were applied for their assessment. The analysis revealed alterations in Neuro scores, home cage locomotion, body weight, nest building, mouse grimace scales, and fecal corticosterone metabolites. A composite measure scheme allowed the allocation of individual mice to severity classes. The comparison between groups failed to indicate the superiority of multimodal regimens over high-dose NSAID monotherapy. In conclusion, our findings confirmed the informative value of various parameters for assessment of pain and distress following neurosurgical procedures in mice. While all drug regimens were well tolerated in control mice, our data suggest that the total drug load should be carefully considered for perioperative management. Future studies would be of interest to assess potential synergies of drug combinations with lower doses of carprofen.
Assuntos
Anti-Inflamatórios não Esteroides , Camundongos Endogâmicos C57BL , Procedimentos Neurocirúrgicos , Manejo da Dor , Dor Pós-Operatória , Animais , Anti-Inflamatórios não Esteroides/administração & dosagem , Anti-Inflamatórios não Esteroides/uso terapêutico , Camundongos , Masculino , Manejo da Dor/métodos , Feminino , Dor Pós-Operatória/tratamento farmacológico , Procedimentos Neurocirúrgicos/efeitos adversos , Carbazóis/administração & dosagem , Analgesia/métodos , Bupivacaína/administração & dosagem , Buprenorfina/administração & dosagem , Analgésicos Opioides/administração & dosagem , Quimioterapia CombinadaRESUMO
The physiological performance of any sensory organ is determined by its anatomy and physical properties. Consequently, complex sensory structures with elaborate features have evolved to optimize stimulus detection. Understanding these structures and their physical nature forms the basis for mechanistic insights into sensory function. Despite its crucial role as a sensor for pheromones and other behaviorally instructive chemical cues, the vomeronasal organ (VNO) remains a poorly characterized mammalian sensory structure. Fundamental principles of its physico-mechanical function, including basic aspects of stimulus sampling, remain poorly explored. Here, we revisit the classical vasomotor pump hypothesis of vomeronasal stimulus uptake. Using advanced anatomical, histological, and physiological methods, we demonstrate that large parts of the lateral mouse VNO are composed of smooth muscle. Vomeronasal smooth muscle tissue comprises two subsets of fibers with distinct topography, structure, excitation-contraction coupling, and, ultimately, contractile properties. Specifically, contractions of a large population of noradrenaline-sensitive cells mediate both transverse and longitudinal lumen expansion, whereas cholinergic stimulation targets an adluminal group of smooth muscle fibers. The latter run parallel to the VNO's rostro-caudal axis and are ideally situated to mediate antagonistic longitudinal constriction of the lumen. This newly discovered arrangement implies a novel mode of function. Single-cell transcriptomics and pharmacological profiling reveal the receptor subtypes involved. Finally, 2D/3D tomography provides non-invasive insight into the intact VNO's anatomy and mechanics, enables measurement of luminal fluid volume, and allows an assessment of relative volume change upon noradrenergic stimulation. Together, we propose a revised conceptual framework for mouse vomeronasal pumping and, thus, stimulus sampling.
Assuntos
Órgão Vomeronasal , Camundongos , Animais , Órgão Vomeronasal/fisiologia , Mamíferos , Feromônios/fisiologiaRESUMO
Accurate medical image segmentation is of utmost importance for enabling automated clinical decision procedures. However, prevailing supervised deep learning approaches for medical image segmentation encounter significant challenges due to their heavy dependence on extensive labeled training data. To tackle this issue, we propose a novel self-supervised algorithm, S3-Net, which integrates a robust framework based on the proposed Inception Large Kernel Attention (I-LKA) modules. This architectural enhancement makes it possible to comprehensively capture contextual information while preserving local intricacies, thereby enabling precise semantic segmentation. Furthermore, considering that lesions in medical images often exhibit deformations, we leverage deformable convolution as an integral component to effectively capture and delineate lesion deformations for superior object boundary definition. Additionally, our self-supervised strategy emphasizes the acquisition of invariance to affine transformations, which is commonly encountered in medical scenarios. This emphasis on robustness with respect to geometric distortions significantly enhances the model's ability to accurately model and handle such distortions. To enforce spatial consistency and promote the grouping of spatially connected image pixels with similar feature representations, we introduce a spatial consistency loss term. This aids the network in effectively capturing the relationships among neighboring pixels and enhancing the overall segmentation quality. The S3-Net approach iteratively learns pixel-level feature representations for image content clustering in an end-to-end manner. Our experimental results on skin lesion and lung organ segmentation tasks show the superior performance of our method compared to the SOTA approaches.
RESUMO
Vision Transformer (ViT) models have demonstrated a breakthrough in a wide range of computer vision tasks. However, compared to the Convolutional Neural Network (CNN) models, it has been observed that the ViT models struggle to capture high-frequency components of images, which can limit their ability to detect local textures and edge information. As abnormalities in human tissue, such as tumors and lesions, may greatly vary in structure, texture, and shape, high-frequency information such as texture is crucial for effective semantic segmentation tasks. To address this limitation in ViT models, we propose a new technique, Laplacian-Former, that enhances the self-attention map by adaptively re-calibrating the frequency information in a Laplacian pyramid. More specifically, our proposed method utilizes a dual attention mechanism via efficient attention and frequency attention while the efficient attention mechanism reduces the complexity of self-attention to linear while producing the same output, selectively intensifying the contribution of shape and texture features. Furthermore, we introduce a novel efficient enhancement multi-scale bridge that effectively transfers spatial information from the encoder to the decoder while preserving the fundamental features. We demonstrate the efficacy of Laplacian-former on multi-organ and skin lesion segmentation tasks with +1.87% and +0.76% dice scores compared to SOTA approaches, respectively. Our implementation is publically available at GitHub.