arXiv subjects

Andreas Maier

Publications and source records attributed to Andreas Maier.

At least 397 records · Page 22Linked to original sources

arXiv2020

Multi-modal Deep Guided Filtering for Comprehensible Medical Image Processing

Deep learning-based image processing is capable of creating highly appealing results. However, it is still widely considered as a "blackbox" transformation. In medical imaging, this lack of comprehensibility of the results is a sensitive issue. The integration of known operators into the deep learning environment has proven to be advantageous for the comprehensibility and reliability of the computations. Consequently, we propose the use of the locally linear guided filter in combination with a learned guidance map for general purpose medical image processing. The output images are only processed by the guided filter while the guidance map can be trained to be task-optimal in an end-to-end fashion. We investigate the performance based on two popular tasks: image super resolution and denoising. The evaluation is conducted based on pairs of multi-modal magnetic resonance imaging and cross-modal computed tomography and magnetic resonance imaging datasets. For both tasks, the proposed approach is on par with state-of-the-art approaches. Additionally, we can show that the input image's content is almost unchanged after the processing which is not the case for conventional deep learning approaches. On top, the proposed pipeline offers increased robustness against degraded input as well as adversarial attacks.

eess.IV↗

arXiv2020

Data Consistent CT Reconstruction from Insufficient Data with Learned Prior Images

Image reconstruction from insufficient data is common in computed tomography (CT), e.g., image reconstruction from truncated data, limited-angle data and sparse-view data. Deep learning has achieved impressive results in this field. However, the robustness of deep learning methods is still a concern for clinical applications due to the following two challenges: a) With limited access to sufficient training data, a learned deep learning model may not generalize well to unseen data; b) Deep learning models are sensitive to noise. Therefore, the quality of images processed by neural networks only may be inadequate. In this work, we investigate the robustness of deep learning in CT image reconstruction by showing false negative and false positive lesion cases. Since learning-based images with incorrect structures are likely not consistent with measured projection data, we propose a data consistent reconstruction (DCR) method to improve their image quality, which combines the advantages of compressed sensing and deep learning: First, a prior image is generated by deep learning. Afterwards, unmeasured projection data are inpainted by forward projection of the prior image. Finally, iterative reconstruction with reweighted total variation regularization is applied, integrating data consistency for measured data and learned prior information for missing data. The efficacy of the proposed method is demonstrated in cone-beam CT with truncated data, limited-angle data and sparse-view data, respectively. For example, for truncated data, DCR achieves a mean root-mean-square error of 24 HU and a mean structure similarity index of 0.999 inside the field-of-view for different patients in the noisy case, while the state-of-the-art U-Net method achieves 55 HU and 0.995 respectively for these two metrics.

eess.IV↗

arXiv2020

A Positive Resampler for Monte Carlo Events with Negative Weights

We propose the Positive Resampler to solve the problem associated with event samples from state-of-the-art predictions for scattering processes at hadron colliders typically involving a sizeable number of events contributing with negative weight. The proposed method guarantees positive weights for all physical distributions, and a correct description of all observables. A desirable side product of the method is the possibility to reduce the size of event samples produced by General Purpose Event Generators, thus lowering the resource demands for subsequent computing-intensive event processing steps. We demonstrate the viability and efficiency of our approach by considering its application to a next-to-leading order + parton shower merged prediction for the production of a $W$ boson in association with multiple jets.

hep-ph↗

arXiv2020

A Global Benchmark of Algorithms for Segmenting Late Gadolinium-Enhanced Cardiac Magnetic Resonance Imaging

Segmentation of cardiac images, particularly late gadolinium-enhanced magnetic resonance imaging (LGE-MRI) widely used for visualizing diseased cardiac structures, is a crucial first step for clinical diagnosis and treatment. However, direct segmentation of LGE-MRIs is challenging due to its attenuated contrast. Since most clinical studies have relied on manual and labor-intensive approaches, automatic methods are of high interest, particularly optimized machine learning approaches. To address this, we organized the "2018 Left Atrium Segmentation Challenge" using 154 3D LGE-MRIs, currently the world's largest cardiac LGE-MRI dataset, and associated labels of the left atrium segmented by three medical experts, ultimately attracting the participation of 27 international teams. In this paper, extensive analysis of the submitted algorithms using technical and biological metrics was performed by undergoing subgroup analysis and conducting hyper-parameter analysis, offering an overall picture of the major design choices of convolutional neural networks (CNNs) and practical considerations for achieving state-of-the-art left atrium segmentation. Results show the top method achieved a dice score of 93.2% and a mean surface to a surface distance of 0.7 mm, significantly outperforming prior state-of-the-art. Particularly, our analysis demonstrated that double, sequentially used CNNs, in which a first CNN is used for automatic region-of-interest localization and a subsequent CNN is used for refined regional segmentation, achieved far superior results than traditional methods and pipelines containing single CNNs. This large-scale benchmarking study makes a significant step towards much-improved segmentation methods for cardiac LGE-MRIs, and will serve as an important benchmark for evaluating and comparing the future works in the field.

cs.CV↗

arXiv2020

Deep Scatter Splines: Learning-Based Medical X-ray Scatter Estimation Using B-splines

The idea of replacing hardware by software to compensate for scattered radiation in flat-panel X-ray imaging is well established in the literature. Recently, deep-learningbased image translation approaches, most notably the U-Net, have emerged for scatter estimation. These yield considerable improvements over model-based methods. Such networks, however, involve potential drawbacks that need to be considered. First, they are trained in a data-driven fashion without making use of prior knowledge and X-ray physics. Second, due to their high parameter complexity, the validity of deep neural networks is difficult to assess. To circumvent these issues, we introduce here a surrogate function to model X-ray scatter distributions that can be expressed by few parameters. We could show empirically that cubic B-splines are well-suited to model X-ray scatter in the diagnostic energy regime. Based on these findings, we propose a lean convolutional encoder architecture that extracts local scatter characteristics from X-ray projection images. These characteristics are embedded into a global context using a constrained weighting matrix yielding spline coefficients that model the scatter distribution. In a first simulation study with 17 thorax data sets, we could show that our method and the U-Net-based state of the art reach about the same accuracy. However, we could achieve these comparable outcomes with orders of magnitude fewer parameters while ensuring that not high-frequency information gets manipulated.

physics.med-ph↗

arXiv2020

Effects of Tissue Material Properties on X-Ray Image, Scatter and Patient Dose Determined using Monte Carlo Simulations

With increasing patient and staff X-ray radiation awareness, many efforts have been made to develop accurate patient dose estimation methods. To date, Monte Carlo (MC) simulations are considered golden standard to simulate the interaction of X-ray radiation with matter. However, sensitivity of MC simulation results to variations in the experimental or clinical setup of image guided interventional procedures are only limited studied. In particular, the impact of patient material compositions is poorly investigated. This is mainly due to the fact, that these methods are commonly validated in phantom studies utilizing a single anthropomorphic phantom. In this study, we therefore investigate the impact of patient material parameters mapping on the outcome of MC X-ray dose simulations. A computation phantom geometry is constructed and three different commonly used material composition mappings are applied. We used the MC toolkit Geant4 to simulate X-ray radiation in an interventional setup and compared the differences in dose deposition, scatter distributions and resulting X-ray images. The evaluation shows a discrepancy between different material composition mapping up to 20 % concerning directly irradiated organs. These results highlight the need for standardization of material composition mapping for MC simulations in a clinical setup.

physics.med-ph↗

arXiv2020

Tenfold your photons -- a physically-sound approach to filtering-based variance reduction of Monte-Carlo-simulated dose distributions

X-ray dose constantly gains interest in the interventional suite. With dose being generally difficult to monitor reliably, fast computational methods are desirable. A major drawback of the gold standard based on Monte Carlo (MC) methods is its computational complexity. Besides common variance reduction techniques, filter approaches are often applied to achieve conclusive results within a fraction of time. Inspired by these methods, we propose a novel approach. We down-sample the target volume based on the fraction of mass, simulate the imaging situation, and then revert the down-sampling. To this end, the dose is weighted by the mass energy absorption, up-sampled, and distributed using a guided filter. Eventually, the weighting is inverted resulting in accurate high resolution dose distributions. The approach has the potential to considerably speed-up MC simulations since less photons and boundary checks are necessary. First experiments substantiate these assumptions. We achieve a median accuracy of 96.7 % to 97.4 % of the dose estimation with the proposed method and a down-sampling factor of 8 and 4, respectively. While maintaining a high accuracy, the proposed method provides for a tenfold speed-up. The overall findings suggest the conclusion that the proposed method has the potential to allow for further efficiency.

physics.med-ph↗

arXiv2020

Fully-automatic CT data preparation for interventional X-ray skin dose simulation

Recently, deep learning (DL) found its way to interventional X-ray skin dose estimation. While its performance was found to be acceptable, even more accurate results could be achieved if more data sets were available for training. One possibility is to turn to computed tomography (CT) data sets. Typically, computed tomography (CT) scans can be mapped to tissue labels and mass densities to obtain training data. However, care has to be taken to make sure that the different clinical settings are properly accounted for. First, the interventional environment is characterized by wide variety of table setups that are significantly different from the typical patient tables used in conventional CT. This cannot be ignored, since tables play a crucial role in sound skin dose estimation in an interventional setup, e. g., when the X-ray source is directly underneath a patient (posterior-anterior view). Second, due to interpolation errors, most CT scans do not facilitate a clean segmentation of the skin border. As a solution to these problems, we applied connected component labeling (CCL) and Canny edge detection to (a) robustly separate the patient from the table and (b) to identify the outermost skin layer. Our results show that these extensions enable fully-automatic, generalized pre-processing of CT scans for further simulation of both skin dose and corresponding X-ray projections.

physics.med-ph↗

arXiv2020

Pitfalls in interventional X-ray organ dose assessment -- combined experimental and computational phantom study: application to prostatic artery embolization

With X-ray radiation protection and dose management constantly gaining interest in interventional radiology, novel procedures often undergo prospective dose studies using anthropomorphic phantoms to determine expected dose values. Due to inherent uncertainties, these dose values might not allow for patient-specific risk assessment. Therefore, first the aim of this study is to quantify the influence of these parameters on local X-ray dose to evaluate their relevance in the assessment of patient-specific organ doses. Second, this knowledge further enables validating a simulation approach, which allows employing physiological material models and patient-specific geometries. Phantom dosimetry experiments using MOSFET dosimeters were conducted reproducing imaging scenarios in prostatic arterial embolization (PAE). Associated organ-equivalent dose of prostate, bladder, colon and skin was determined. Dose deviation induced by possible small displacements of the patient was reproduced by moving the X-ray source. Dose deviation induced by material differences was investigated by analyzing two different commonly used phantoms. We reconstructed the experiments using Monte Carlo (MC) simulations, a reference male geometry, and different material properties. Overall, MC simulated organ dose values are in accordance with the measured ones for the majority of cases. Marginal displacements of X-ray source relative to the phantoms lead to deviations of 6% to 135% in organ dose values, while skin dose remains relatively constant. Regarding the impact of phantom material composition, underestimation of internal organ dose values by 12% to 20% is prevalent in all simulated phantoms. Skin dose, however, can be estimated with low deviation of 1% to 8% at least for two materials. Prospective reference dose studies might not extend to precise patient-specific dose assessment.

physics.med-ph↗

arXiv2020

Projection Inpainting Using Partial Convolution for Metal Artifact Reduction

In computer tomography, due to the presence of metal implants in the patient body, reconstructed images will suffer from metal artifacts. In order to reduce metal artifacts, metals are typically removed in projection images. Therefore, the metal corrupted projection areas need to be inpainted. For deep learning inpainting methods, convolutional neural networks (CNNs) are widely used, for example, the U-Net. However, such CNNs use convolutional filter responses on both valid and corrupted pixel values, resulting in unsatisfactory image quality. In this work, partial convolution is applied for projection inpainting, which only relies on valid pixels values. The U-Net with partial convolution and conventional convolution are compared for metal artifact reduction. Our experiments demonstrate that the U-Net with partial convolution is able to inpaint the metal corrupted areas better than that with conventional convolution.

cs.CV↗

arXiv2020

Are fast labeling methods reliable? A case study of computer-aided expert annotations on microscopy slides

Deep-learning-based pipelines have shown the potential to revolutionalize microscopy image diagnostics by providing visual augmentations to a trained pathology expert. However, to match human performance, the methods rely on the availability of vast amounts of high-quality labeled data, which poses a significant challenge. To circumvent this, augmented labeling methods, also known as expert-algorithm-collaboration, have recently become popular. However, potential biases introduced by this operation mode and their effects for training neuronal networks are not entirely understood. This work aims to shed light on some of the effects by providing a case study for three pathologically relevant diagnostic settings. Ten trained pathology experts performed a labeling tasks first without and later with computer-generated augmentation. To investigate different biasing effects, we intentionally introduced errors to the augmentation. Furthermore, we developed a novel loss function which incorporates the experts' annotation consensus in the training of a deep learning classifier. In total, the pathology experts annotated 26,015 cells on 1,200 images in this novel annotation study. Backed by this extensive data set, we found that the consensus of multiple experts and the deep learning classifier accuracy, was significantly increased in the computer-aided setting, versus the unaided annotation. However, a significant percentage of the deliberately introduced false labels was not identified by the experts. Additionally, we showed that our loss function profited from multiple experts and outperformed conventional loss functions. At the same time, systematic errors did not lead to a deterioration of the trained classifier accuracy. Furthermore, a classifier trained with annotations from a single expert with computer-aided support can outperform the combined annotations from up to nine experts.

cs.HC↗

arXiv2020

Recognizing Characters in Art History Using Deep Learning

In the field of Art History, images of artworks and their contexts are core to understanding the underlying semantic information. However, the highly complex and sophisticated representation of these artworks makes it difficult, even for the experts, to analyze the scene. From the computer vision perspective, the task of analyzing such artworks can be divided into sub-problems by taking a bottom-up approach. In this paper, we focus on the problem of recognizing the characters in Art History. From the iconography of $Annunciation$ $of$ $the$ $Lord$ (Figure 1), we consider the representation of the main protagonists, $Mary$ and $Gabriel$, across different artworks and styles. We investigate and present the findings of training a character classifier on features extracted from their face images. The limitations of this method, and the inherent ambiguity in the representation of $Gabriel$, motivated us to consider their bodies (a bigger context) to analyze in order to recognize the characters. Convolutional Neural Networks (CNN) trained on the bodies of $Mary$ and $Gabriel$ are able to learn person related features and ultimately improve the performance of character recognition. We introduce a new technique that generates more data with similar styles, effectively creating data in the similar domain. We present experiments and analysis on three different models and show that the model trained on domain related data gives the best performance for recognizing character. Additionally, we analyze the localized image regions for the network predictions. Code is open-sourced and available at https://github.com/prathmeshrmadhu/recognize_characters_art_history and the link to the published peer-reviewed article is https://dl.acm.org/citation.cfm?id=3357242.

cs.CV↗

arXiv2020

Will we ever have Conscious Machines?

The question of whether artificial beings or machines could become self-aware or consciousness has been a philosophical question for centuries. The main problem is that self-awareness cannot be observed from an outside perspective and the distinction of whether something is really self-aware or merely a clever program that pretends to do so cannot be answered without access to accurate knowledge about the mechanism's inner workings. We review the current state-of-the-art regarding these developments and investigate common machine learning approaches with respect to their potential ability to become self-aware. We realise that many important algorithmic steps towards machines with a core consciousness have already been devised. For human-level intelligence, however, many additional techniques have to be discovered.

cs.AI↗

arXiv2020

The Effect of Data Augmentation on Classification of Atrial Fibrillation in Short Single-Lead ECG Signals Using Deep Neural Networks

Cardiovascular diseases are the most common cause of mortality worldwide. Detection of atrial fibrillation (AF) in the asymptomatic stage can help prevent strokes. It also improves clinical decision making through the delivery of suitable treatment such as, anticoagulant therapy, in a timely manner. The clinical significance of such early detection of AF in electrocardiogram (ECG) signals has inspired numerous studies in recent years, of which many aim to solve this task by leveraging machine learning algorithms. ECG datasets containing AF samples, however, usually suffer from severe class imbalance, which if unaccounted for, affects the performance of classification algorithms. Data augmentation is a popular solution to tackle this problem. In this study, we investigate the impact of various data augmentation algorithms, e.g., oversampling, Gaussian Mixture Models (GMMs) and Generative Adversarial Networks (GANs), on solving the class imbalance problem. These algorithms are quantitatively and qualitatively evaluated, compared and discussed in detail. The results show that deep learning-based AF signal classification methods benefit more from data augmentation using GANs and GMMs, than oversampling. Furthermore, the GAN results in circa $3\%$ better AF classification accuracy in average while performing comparably to the GMM in terms of f1-score.

eess.SP↗

arXiv2020

Weakly Supervised Segmentation of Cracks on Solar Cells using Normalized Lp Norm

Photovoltaic is one of the most important renewable energy sources for dealing with world-wide steadily increasing energy consumption. This raises the demand for fast and scalable automatic quality management during production and operation. However, the detection and segmentation of cracks on electroluminescence (EL) images of mono- or polycrystalline solar modules is a challenging task. In this work, we propose a weakly supervised learning strategy that only uses image-level annotations to obtain a method that is capable of segmenting cracks on EL images of solar cells. We use a modified ResNet-50 to derive a segmentation from network activation maps. We use defect classification as a surrogate task to train the network. To this end, we apply normalized Lp normalization to aggregate the activation maps into single scores for classification. In addition, we provide a study how different parameterizations of the normalized Lp layer affect the segmentation performance. This approach shows promising results for the given task. However, we think that the method has the potential to solve other weakly supervised segmentation problems as well.

cs.CV↗

arXiv2020

CLCNet: Deep learning-based Noise Reduction for Hearing Aids using Complex Linear Coding

Noise reduction is an important part of modern hearing aids and is included in most commercially available devices. Deep learning-based state-of-the-art algorithms, however, either do not consider real-time and frequency resolution constrains or result in poor quality under very noisy conditions. To improve monaural speech enhancement in noisy environments, we propose CLCNet, a framework based on complex valued linear coding. First, we define complex linear coding (CLC) motivated by linear predictive coding (LPC) that is applied in the complex frequency domain. Second, we propose a framework that incorporates complex spectrogram input and coefficient output. Third, we define a parametric normalization for complex valued spectrograms that complies with low-latency and on-line processing. Our CLCNet was evaluated on a mixture of the EUROM database and a real-world noise dataset recorded with hearing aids and compared to traditional real-valued Wiener-Filter gains.

eess.AS↗

arXiv2020

An Investigation of Feature-based Nonrigid Image Registration using Gaussian Process

For a wide range of clinical applications, such as adaptive treatment planning or intraoperative image update, feature-based deformable registration (FDR) approaches are widely employed because of their simplicity and low computational complexity. FDR algorithms estimate a dense displacement field by interpolating a sparse field, which is given by the established correspondence between selected features. In this paper, we consider the deformation field as a Gaussian Process (GP), whereas the selected features are regarded as prior information on the valid deformations. Using GP, we are able to estimate the both dense displacement field and a corresponding uncertainty map at once. Furthermore, we evaluated the performance of different hyperparameter settings for squared exponential kernels with synthetic, phantom and clinical data respectively. The quantitative comparison shows, GP-based interpolation has performance on par with state-of-the-art B-spline interpolation. The greatest clinical benefit of GP-based interpolation is that it gives a reliable estimate of the mathematical uncertainty of the calculated dense displacement map.

cs.CV↗

arXiv2020

Limited Angle Tomography for Transmission X-Ray Microscopy Using Deep Learning

In transmission X-ray microscopy (TXM) systems, the rotation of a scanned sample might be restricted to a limited angular range to avoid collision to other system parts or high attenuation at certain tilting angles. Image reconstruction from such limited angle data suffers from artifacts due to missing data. In this work, deep learning is applied to limited angle reconstruction in TXMs for the first time. With the challenge to obtain sufficient real data for training, training a deep neural network from synthetic data is investigated. Particularly, the U-Net, the state-of-the-art neural network in biomedical imaging, is trained from synthetic ellipsoid data and multi-category data to reduce artifacts in filtered back-projection (FBP) reconstruction images. The proposed method is evaluated on synthetic data and real scanned chlorella data in $100^\circ$ limited angle tomography. For synthetic test data, the U-Net significantly reduces root-mean-square error (RMSE) from $2.55 \times 10^{-3}$ μm$^{-1}$ in the FBP reconstruction to $1.21 \times 10^{-3}$ μm$^{-1}$ in the U-Net reconstruction, and also improves structural similarity (SSIM) index from 0.625 to 0.920. With penalized weighted least square denoising of measured projections, the RMSE and SSIM are further improved to $1.16 \times 10^{-3}$ μm$^{-1}$ and 0.932, respectively. For real test data, the proposed method remarkably improves the 3-D visualization of the subcellular structures in the chlorella cell, which indicates its important value for nano-scale imaging in biology, nanoscience and materials science.

eess.IV↗