Post-processing Algorithm Research Articles

Post-processing Algorithm of a Novel Five-axis Hybrid Kinematic Machining Unit

Post-processing Algorithm of a Novel Five-axis Hybrid Kinematic Machining Unit

Dynamic Rail Near-Surface Inspection of Multiphysical Coupled Electromagnetic and Thermography Sensing System

The effectiveness of railway fault inspection has remained challenging. Conventional techniques are still functionally limited and unable to meet the increasing demand of railway diagnosis. To mitigate the variety of rail fault detection problems, this article proposes a dynamic railway inspection system based on multiphysical coupled electromagnetic and thermography sensing. It further shows the development and construction of a new inverted L-type magnet yoke abreast with volumetric coil array. The novel structure can not only significantly enhance the sensitivity and detectability of the region of interest (ROI), but also effectively detect the subsurface defects with the compensation of coils array due to the coupled electromagnetic field. Furthermore, the theoretical analysis of the coupled physical fields has been derived and proved to be consistent with the numerical simulation results. A rail test sample with various defects is carried out to verify the feasibility of the proposed system. Additionally, a metric learning post-processing algorithm has been conducted for distilling eddy current signals and thermograms to improve the accuracy of the detection results. On-site experimental and contrast results with various levels of performance validation have demonstrated that the integrated system is well suited for dynamic rail inspection on near-surface cracks at speed of 1 km/h.

Defining a baseline identification of artifacts in confocal laser endomicroscopy in head and neck cancer imaging

PurposeProbe-based confocal laser endomicroscopy (pCLE) allows imaging of the laryngeal mucosa in a thousand-fold magnification. However, the existence of multiple nondiagnostic frames might hinder the performance of automated systems, especially in post-processing and analysis algorithms. This study aims to investigate the feasibility of building a baseline for the identification of diagnostic frames and to evaluate the influence of sequence information in this task. Materials and methodsWe included eight patients with squamous cell carcinoma (SCC) and planned total laryngectomy between October 2020 and February 2021. One hundred frames were randomly selected out of 21,272 pCLE frames of the tumor and healthy mucosa during surgery. Three experts were asked to classify these frames into diagnostic and nondiagnostic ones with different subtypes under two settings, i.e., isolated frames or short video sequences. Inter-rater and intra-rater consistency were analyzed with Cohen's kappa coefficients. ResultsAmong all raters, regarding the classification of diagnostic frames, an average agreement of κ = 0.79 has been achieved. The overall consistency for the subtypes of nondiagnostic frames is κ = 0.75 and κ = 0.57, respectively. ConclusionsThe high level of consistency regarding the identification of diagnostic frames shows the reliability of the classification results and builds the foundation for future automatic algorithms.

Post-acquisition water-signal removal in 3D water-unsuppressed 1 H-MR spectroscopic imaging of the prostate.

To develop a robust processing procedure of raw signals from water-unsuppressed MRSI of the prostate for the mapping of absolute tissue concentrations of metabolites. Water-unsuppressed 3D MRSI data were acquired from a phantom, from healthy volunteers, and a patient with prostate cancer. Signal processing included sequential computation of the modulus of the FID to remove water sidebands, a Hilbert transformation, and k-space Hamming filtering. For the removal of the water signal, we compared Löwner tensor-based blind source separation (BSS) and Hankel Lanczos singular value decomposition techniques. Absolute metabolite levels were quantified with LCModel and the results were statistically analyzed to compare the water removal methods and conventional water-suppressed MRSI. The post-processing algorithms successfully removed the water signal and its sidebands without affecting metabolite signals. The best water removal performance was achieved by Löwner tensor-based BSS. Absolute tissue concentrations of citrate in the peripheral zone derived from water-suppressed and unsuppressed 1 H MRSI were the same and as expected from the known physiology of the healthy prostate. Maps for citrate and choline from water-unsuppressed 3D 1 H-MRSI of the prostate showed expected spatial variations in metabolite levels. We developed a robust relatively simple post-processing method of water-unsuppressed MRSI of the prostate to remove the water signal. Absolute quantification using the water signal, originating from the same location as the metabolite signals, avoids the acquisition of additional reference data.

Motion artifact correction for cone beam CT stroke imaging: a prospective series

BackgroundImaging assessment for acute ischemic stroke (AIS) patients in the angiosuite using cone beam CT (CBCT) has created increased interest since endovascular treatment became the first line therapy for proximal...

A Robust Method: Arbitrary Shape Text Detection Combining Semantic and Position Information

There is a growing interest in scene text detection for arbitrary shapes. The effectiveness of text detection has also evolved from horizontal text detection to the ability to perform text detection in multiple directions and arbitrary shapes. However, scene text detection is still a challenging task due to significant differences in size and aspect ratio and diversity in shape, as well as orientation, coarse annotations, and other factors. Regression-based methods are inspired by object detection and have limitations in fitting the edges of arbitrarily shaped text due to the characteristics of their methods. Segmentation-based methods, on the other hand, perform prediction at the pixel level and thus can fit arbitrarily shaped text better. However, the inaccuracy of image text annotations and the distribution characteristics of text pixels, which contain a large number of background pixels and misclassified pixels, degrades the performance of segmentation-based text detection methods to some extent. Usually, considering whether a pixel belongs to a text region is highly dependent on the strength of the semantic information it has and the position of the pixel in the text area. Based on the above two points, we propose an innovative and robust method for scene text detection combining position and semantic information. First, we add position information to the images using a position encoding module (PosEM) to help the model learn the implicit feature relationships associated with the position. Second, we use the semantic enhancement module (SEM) to enhance the model's focus on the semantic information in the image during feature extraction. Then, to minimize the effect of noise due to inaccurate image text annotations and the distribution characteristics of text pixels, we convert the detection results into a probability map that can more reasonably represent the text distribution. Finally, we reconstruct and filter the text instances using a post-processing algorithm to reduce false positives. The experimental results show that our model improves significantly on the Total-Text, MSRA-TD500, and CTW1500 datasets, outperforming most previous advanced algorithms.

Optical Spectrophotometry as a Promising Method for Quantification and Stage Differentiation of Plasmodium falciparum Parasites.

Malaria is one of the most life-threatening infectious diseases worldwide, claiming half a million lives yearly. Prompt and accurate diagnosis is crucial for disease control and elimination. Currently used diagnostic methods require blood sampling and fail to detect low-level infections. At the symptomatic stage of infection, the parasites feed on red blood cells' (RBCs) hemoglobin, forming inert crystals, the hemozoin, in the process. Thus, along with parasite maturation inside the RBCs, the hemoglobin and hemozoin proportion is inversely related, and they generate specific optical spectra, according to their concentration. Herein, to address the issues of finger prick sampling and the lack of sensitivity of the parasitological test, we explored the optical features of Plasmodium falciparum-infected RBCs through absorbance and reflectance spectrophotometric characterization, aiming for their detection. This is the first work fully characterizing the spectrophotometric properties of P. falciparum-infected RBCs by using only 16 specific wavelengths within the visible optical spectra and two different post-processing algorithms. With such an innovative methodology, low-level infections can be detected and quantified, and early- and late-stage development can be clearly distinguished, not only improving the current detection limits but also proving the successful applicability of spectrophotometry for competitive and accurate malaria diagnosis.

Atrial contractile function and fibrosis in patients with paroxysmal atrial fibrillation in sinus rhythm

The aim of the study was to investigate atrial contractile function in patients with paroxysmal atrial fibrillation (AF) in sinus rhythm using transthoracic echocardiography (EchoCG). Thirty-five patients with paroxysmal AF and arterial hypertension (mean age 62 ± 10 years, 43% male) in sinus rhythm were enrolled in the study. The control group was composed of comparable patients with arterial hypertension without heart rhythm disturbances. EchoCG was performed during sinus rhythm according to an extended protocol, which included the ejection fraction (EF) of the left atrium (LA) and tissue Doppler measurements. Myocardial fibrosis was assessed quantitatively by videodensitometry in intraventricular and intraatrial (IAS) septa using an original image post-processing algorithm. We found a significant decrease in the left atrial contraction function during sinus rhythm in patients with AF when compared to controls. LA EF (34 ± 14 vs. 54 ± 17, p = 0.03) and A' velocity (0.17 ± 0.04 vs. 0.22 ± 0.04, p = 0.008) decreased while A/A' ratio (2.7 ± 0.2 vs. 1.9 ± 0.1, p = 0.006) increased. Peak A velocity was not affected. Videodensitometric analysis revealed a 2.3-fold increase in IAS fibrosis fraction in AF patients compared with controls (p = 0.01). Patients with AF in sinus rhythm have markedly depressed atrial contractile function. Videodensitometry of IAS has the potential to be used as inexpensive method of atrial fibrosis assessment in patients with AF.

Experimental investigation into the effect of main stage swirl on flow and spray characteristics inside a stratified partially premixed combustor

Lean combustion has been applied in gas turbine combustors to meet increasingly stringent regulations for pollutant emissions. As a representative low-emission technology, stratified partially premixed (SPP) combustion is capable of effectively reducing NOx emissions while still providing robust flames. In this work, the flow and spray processes inside an SPP combustor were experimentally investigated to further reveal the role of main stage swirl in flow development and fuel placement. Particle image velocimetry (PIV) and Planar Mie scattering (PMie) techniques were applied to provide the velocity distributions and droplet distributions under isothermal conditions with a relative pressure drop ranging from 1% to 5%. A spray post-processing algorithm was developed to acquire the characteristic parameters of the sprays without clear boundaries. Results show that the strong swirl from the main stage leads to the radial flow of pilot air streams and dominates the formation of central toroidal recirculation zone (CTRZ). With the addition of the main stage swirl, a proper flow structure including jet zones, recirculation zones and shear layers is formed, which promotes the widening of droplet radial distribution and the increase of the spray cone angle. The spray distribution therefore transforms from a bubble shape into a cone shape. The findings and conclusions could promote the understanding of interactions between pilot and main stage swirls for stage combustors and contribute to the optimization of such SPP injectors.

Comparison of Gray-scale Inversion to Improve Detection of Pulmonary Nodules on Chest X-rays Between Radiologists and a Deep Convolutional Neural Network

Detection of pulmonary nodules on chest x-rays is an important task for radiologists. Previous studies have shown improved detection rates using gray-scale inversion. The purpose of our study was to compare the efficacy of gray-scale inversion in improving the detection of pulmonary nodules on chest x-rays for radiologists and machine learning models (ML). We created a mixed dataset consisting of 60, 2-view (posteroanterior view - PA and lateral view) chest x-rays with computed tomography confirmed nodule(s) and 62 normal chest x-rays. Twenty percent of the cases were separated for a testing dataset (24 total images). Data augmentation through mirroring and transfer learning was used for the remaining cases (784 total images) for supervised training of 4 ML models (grayscale PA, grayscale lateral, gray-scale inversion PA, and gray-scale inversion lateral) on Google's cloud-based AutoML platform. Three cardiothoracic radiologists analyzed the complete 2-view dataset (n=120) and, for comparison to the ML, the single-view testing subsets (12 images each). Gray-scale inversion (area under the curve (AUC) 0.80, 95% confidence interval (CI) 0.75-0.85) did not improve diagnostic performance for radiologists compared to grayscale (AUC 0.84, 95% CI 0.79-0.88). Gray-scale inversion also did not improve diagnostic performance for the ML. The ML did demonstrate higher sensitivity and negative predictive value for grayscale PA (72.7% and 75.0%), grayscale lateral (63.6% and 66.6%), and gray-scale inversion lateral views (72.7% and 76.9%), comparing favorably to the radiologists (63.9% and 72.3%, 27.8% and 58.3%, 19.5% and 50.5% respectively). In the limited testing dataset, the ML did demonstrate higher sensitivity and negative predictive value for grayscale PA (72.7% and 75.0%), grayscale lateral (63.6% and 66.6%), and gray-scale inversion lateral views (72.7% and 76.9%), comparing favorably to the radiologists (63.9% and 72.3%, 27.8% and 58.3%, 19.5% and 50.5%, respectively). Further investigation of other post-processing algorithms to improve diagnostic performance of ML is warranted.

An Event Matching Energy Disaggregation Algorithm Using Smart Meter Data

Energy disaggregation algorithms disintegrate aggregate demand into appliance-level demands. Among various energy disaggregation approaches, non-intrusive load monitoring (NILM) algorithms requiring a single sensor have gained much attention in recent years. Various machine learning and optimization-based NILM approaches are available in the literature, but bulk training data and high computational time are their respective drawbacks. Considering these drawbacks, we devised an event matching energy disaggregation algorithm (EMEDA) for NILM of multistate household appliances using smart meter data. Having limited training data, K-means clustering was employed to estimate appliance power states. These power states were accumulated to generate an event database (EVD) containing all combinations of appliance operations in their various states. Prior to matching, the test samples of aggregate demand events were decreased by event-driven data compression for computational effectiveness. The compressed test events were matched in the sorted EVD to assess the contribution of each appliance in the aggregate demand. To counter the effects of transient spikes and/or dips that occurred during the state transition of appliances, a post-processing algorithm was also developed. The proposed approach was validated using the low-rate data of the Reference Energy Disaggregation Dataset (REDD). With better energy disaggregation performance, the proposed EMEDA exhibited reductions of 97.5 and 61.7% in computational time compared with the recent smart event-based optimization and optimization-based load disaggregation approaches, respectively.

Proton beam range verification by means of ionoacoustic measurements at clinically relevant doses using a correlation-based evaluation.

The Bragg peak located at the end of the ion beam range is one of the main advantages of ion beam therapy compared to X-Ray radiotherapy. However, verifying the exact position of the Bragg peak within the patient online is a major challenge. The goal of this work was to achieve submillimeter proton beam range verification for pulsed proton beams of an energy of up to 220 MeV using ionoacoustics for a clinically relevant dose deposition of typically 2 Gy per fraction by i) using optimal proton beam characteristics for ionoacoustic signal generation and ii) improved signal detection by correlating the signal with simulated filter templates. A water tank was irradiated with a preclinical 20 MeV proton beam using different pulse durations ranging from 50 ns up to 1 μs in order to maximise the signal-to-noise ratio (SNR) of ionoacoustic signals. The ionoacoustic signals were measured using a piezo-electric ultrasound transducer in the MHz frequency range. The signals were filtered using a cross correlation-based signal processing algorithm utilizing simulated templates, which enhances the SNR of the recorded signals. The range of the protons is evaluated by extracting the time of flight (ToF) of the ionoacoustic signals and compared to simulations from a Monte Carlo dose engine (FLUKA). Optimised SNR of 28.0 ± 10.6 is obtained at a beam current of 4.5 μA and a pulse duration of 130 ns at a total peak dose deposition of 0.5 Gy. Evaluated ranges coincide with Monte Carlo simulations better than 0.1 mm at an absolute range of 4.21 mm. Higher beam energies require longer proton pulse durations for optimised signal generation. Using the correlation-based post-processing filter a SNR of 17.8 ± 5.5 is obtained for 220 MeV protons at a total peak dose deposition of 1.3 Gy. For this clinically relevant dose deposition and proton beam energy, submillimeter range verification was achieved at an absolute range of 303 mm in water. Optimal proton pulse durations ensure an ideal trade-off between maximising the ionoacoustic amplitude and minimising dose deposition. In combination with a correlation-based post-processing evaluation algorithm, a reasonable SNR can be achieved at low dose levels putting clinical applications for online proton or ion beam range verification into reach.

Towards development of an intelligent failure analysis system based on infrared thermography

Electronic components of which reliability cannot be quantified are unacceptable and potentially hazardous, especially in safety-relevant areas such as driver assistance systems and medical technology where the zero-error principle applies. Reliability as a quality criterion has its origin in production, i.e. process variations have a negative influence on the structural integrity of the contact elements on the packaging and interconnect technology and, thus on the device performance in the field under thermo-mechanical load (temperature changes, vibration, humidity). At present, to ensure reproducibility of the reliability of each component, regular quality tests are often carried out in practice. However, a better and reliable approach will carry out 100 % inline checks for traceability and immediate readjustment. This work is the first step towards developing an intelligent non-destructive inline-capable failure analysis technique using infrared thermography. Good data forms the base on which robust and accurate AI algorithms can be trained and developed. However, the obtained thermographic images need to be processed so that subsurface defects can be detected. In this work, prominent algorithms, namely Pulse Phase Thermography (PPT), Thermographic Signal Reconstruction (TSR), Principal Component Thermography (PCT), Slope and Correlation Coefficient, have been thoroughly discussed and examined on the thermographic sequence from a plexiglass sample. A hybrid algorithm of TSR and PCT has also been suggested with promising results. In the end, potential post-processing algorithms from which the obtained results can be used for training an ML/AI model have been discussed.The major problem associated with the deep learning approach is the lack of data in the training set. The model's performance is dependent on the quality and quantity of the training data. The training data should be a good representation of the real-world scenario, i.e. it should be accurate enough and contain enough images, including exceptions for learning reliable features. However, getting sufficient data is a challenge in the manufacturing industry. However, using the Finite Element Model (FEM) approach for data generation can help overcome the data hurdle. In this paper, we also evaluate the potential application of FEM and the problems one faces when trying to generate a large amount of data for training using FEM.

Hex-Mesh Generation and Processing: A Survey

In this article, we provide a detailed survey of techniques for hexahedral mesh generation. We cover the whole spectrum of alternative approaches to mesh generation, as well as post-processing algorithms for connectivity editing and mesh optimization. For each technique, we highlight capabilities and limitations, also pointing out the associated unsolved challenges. Recent relaxed approaches, aiming to generate not pure-hex but hex-dominant meshes, are also discussed. The required background, pertaining to geometrical as well as combinatorial aspects, is introduced along the way.

Breast cancer detection and classification in mammogram using a three-stage deep learning framework based on PAA algorithm

In recent years, deep learning has been used to develop an automatic breast cancer detection and classification tool to assist doctors. In this paper, we proposed a three-stage deep learning framework based on an anchor-free object detection algorithm, named the Probabilistic Anchor Assignment (PAA) to improve diagnosis performance by automatically detecting breast lesions (i.e., mass and calcification) and further classifying mammograms into benign or malignant. Firstly, a single-stage PAA-based detector roundly finds suspicious breast lesions in mammogram. Secondly, we designed a two-branch ROI detector to further classify and regress these lesions that aim to reduce the number of false positives. Besides, in this stage, we introduced a threshold-adaptive post-processing algorithm with dense breast information. Finally, the benign or malignant lesions would be classified by an ROI classifier which combines local-ROI features and global-image features. In addition, considering the strong correlation between the task of detection head of PAA and the task of whole mammogram classification, we added an image classifier that utilizes the same global-image features to perform image classification. The image classifier and the ROI classifier jointly guide to enhance the feature extraction ability and further improve the performance of classification. We integrated three public datasets of mammograms (CBIS-DDSM, INbreast, MIAS) to train and test our model and compared our framework with recent state-of-the-art methods. The results show that our proposed method can improve the diagnostic efficiency of radiologists by automatically detecting and classifying breast lesions and classifying benign and malignant mammograms.

Enlarging the frequency dynamic range of DMZI sensors based on the switching control feedback system.

The feedback loop in dual Mach-Zehnder interferometer (DMZI) sensors stabilizes the system operating at the quadratic point for the highest sensitivity but requires the minimum measurable vibration frequency out of the feedback bandwidth, resulting in a limited dynamic range. In this paper, we point out that the feedback operation is unnecessary while vibration is occurring and propose a strategy to adaptively enable/disable the feedback phase compensation depending on the vibration state, lowering the minimum measurable vibration frequency tenfold. Moreover, the state variable employed enables direct extraction of vibration-related data, with no need of complicated postprocessing algorithms.

Nurse rostering with fatigue modelling

We use a real Nurse Rostering Problem and a validated model of human sleep to formulate the Nurse Rostering Problem with Fatigue. The fatigue modelling includes individual biologies, thus enabling personalised schedules for every nurse. We create an approximation of the sleep model in the form of a look-up table, enabling its incorporation into nurse rostering. The problem is solved using an algorithm that combines Mixed-Integer Programming and Constraint Programming with a Large Neighbourhood Search. A post-processing algorithm deals with errors, to produce feasible rosters minimising global fatigue. The results demonstrate the realism of protecting nurses from highly fatiguing schedules and ensuring the alertness of staff. We further demonstrate how minimally increased staffing levels enable lower fatigue, and find evidence to suggest biological complementarity among staff can be used to reduce fatigue. We also demonstrate how tailoring shifts to nurses’ biology reduces the overall fatigue of the team, which means managers must grapple with the issue of fairness in rostering.

A deep learning-based post-processing method for automated pulmonary lobe and airway trees segmentation using chest CT images in PET/CT.

BackgroundThe proposed algorithm could support accurate localization of lung disease. To develop and validate an automated deep learning model combined with a post-processing algorithm to segment six pulmonary anatomical regions in chest computed tomography (CT) images acquired during positron emission tomography/computed tomography (PET/CT) scans. The pulmonary regions have five pulmonary lobes and airway trees.MethodsPatients who underwent both PET/CT imaging with an extra chest CT scan were retrospectively enrolled. The pulmonary segmentation of six regions in CT was performed via a convolutional neural network (CNN) of DenseVNet architecture with some post-processing algorithms. Three evaluation metrics were used to assess the performance of this method, which combined deep learning and the post-processing method. The agreement between the combined model and ground truth segmentations in the test set was analyzed.ResultsA total of 640 cases were enrolled. The combined model, which involved deep learning and post-processing methods, had a higher performance than the single deep learning model. In the test set, the all-lobes overall Dice coefficient, Hausdorff distance, and Jaccard coefficient were 0.972, 12.025 mm, and 0.948, respectively. The airway-tree Dice coefficient, Hausdorff distance, and Jaccard coefficient were 0.849, 32.076 mm, and 0.815, respectively. A good agreement was observed between our segmentation in every plot.ConclusionsThe proposed model combining two methods can automatically segment five pulmonary lobes and airway trees on chest CT imaging in PET/CT. The performance of the combined model was higher than the single deep learning model in each region in the test set.

Half-sibling regression meets exoplanet imaging: PSF modeling and subtraction using a flexible, domain knowledge-driven, causal framework

Context. High-contrast imaging of exoplanets hinges on powerful post-processing methods to denoise the data and separate the signal of a companion from its host star, which is typically orders of magnitude brighter. Aims. Existing post-processing algorithms do not use all prior domain knowledge that is available about the problem. We propose a new method that builds on our understanding of the systematic noise and the causal structure of the data-generating process. Methods. Our algorithm is based on a modified version of half-sibling regression (HSR), a flexible denoising framework that combines ideas from the fields of machine learning and causality. We adapted the method to address the specific requirements of high-contrast exoplanet imaging data obtained in pupil tracking mode. The key idea is to estimate the systematic noise in a pixel by regressing the time series of this pixel onto a set of causally independent, signal-free predictor pixels. We use regularized linear models in this work; however, other (nonlinear) models are also possible. In a second step, we demonstrate how the HSR framework allows us to incorporate observing conditions such as wind speed or air temperature as additional predictors. Results. When we applied our method to four data sets from the VLT/NACO instrument, our algorithm provided a better false-positive fraction than a popular baseline method in the field. Additionally, we found that the HSR-based method provides direct and accurate estimates for the contrast of the exoplanets without the need to insert artificial companions for calibration in the data sets. Finally, we present a first piece of evidence that using the observing conditions as additional predictors can improve the results. Conclusions. Our HSR-based method provides an alternative, flexible, and promising approach to the challenge of modeling and subtracting the stellar PSF and systematic noise in exoplanet imaging data.

Reduction of the cardiac pulsation artifact and improvement of lesion conspicuity in flow-compensated diffusion images in the liver-A quantitative evaluation of postprocessing algorithms.

To enhance image quality of flow-compensated diffusion-weighted liver MRI data by increasing the lesion conspicuity and reducing the cardiac pulsation artifact using postprocessing algorithms. Diffusion-weighted image data of 40 patients with liver lesions had been acquired at 1.5T. These data were postprocessed with 5 different algorithms (weighted averaging, p-mean, percentile, outlier exclusion, and exception set). Four image properties of the postprocessed data were evaluated for optimizing the algorithm parameters. These properties were the lesion to tissue contrast-to-noise ratio (CNR), the reduction of the cardiac pulsation artifact, the data consistency, and the vessel darkness. They were combined into a total quality score ( set to 1 for the trace-weighted reference image), which was used to rate the image quality objectively. The weighted averaging algorithm performed best according to the total quality score ( ). The further ranking was outlier exclusion algorithm ( ), p-mean algorithm ( ), percentile algorithm ( ), and exception set algorithm ( ). All optimized algorithms except for the exception set algorithm corrected the pulsation artifact and increased the lesion CNR. Changes in were significant for all optimized algorithms except for the percentile algorithm. Liver ADC was significantly reduced (except for the exception set algorithm), particularly in the left lobe. Postprocessing algorithms should be used for flow-compensated liver DWI. The proposed weighted averaging algorithm seems to be suited best to increase the image quality of artifact-corrupted flow-compensated diffusion-weighted liver data.