How auto-differentiation can improve CT workflows: classical algorithms in a modern framework.

Many of the recent successes of deep learning-based approaches have been enabled by a framework of flexible, composable computational blocks with their parameters adjusted through an automatic differentiation mechanism to implement various data processing tasks. In this work, we explore how the same philosophy can be applied to existing "classical" (i.e., non-learning) algorithms, focusing on computed tomography (CT) as application field. We apply four key design principles of this approach for CT workflow design: end-to-end optimization, explicit quality criteria, declarative algorithm construction by building the forward model, and use of existing classical algorithms as computational blocks. Through four case studies, we demonstrate that auto-differentiation is remarkably effective beyond the boundaries of neural-network training, extending to CT workflows containing varied combinations of classical and machine learning algorithms.

Deep-learning-based joint rigid and deformable contour propagation for magnetic resonance imaging-guided prostate radiotherapy.

BACKGROUND: Deep learning-based unsupervised image registration has recently been proposed, promising fast registration. However, it has yet to be adopted in the online adaptive magnetic resonance imaging-guided radiotherapy (MRgRT) workflow. PURPOSE: In this paper, we design an unsupervised, joint rigid, and deformable registration framework for contour propagation in MRgRT of prostate cancer. METHODS: Three-dimensional pelvic T2-weighted MRIs of 143 prostate cancer patients undergoing radiotherapy were collected and divided into 110, 13, and 20 patients for training, validation, and testing. We designed a framework using convolutional neural networks (CNNs) for rigid and deformable registration. We selected the deformable registration network architecture among U-Net, MS-D Net, and LapIRN and optimized the training strategy (end-to-end vs. sequential). The framework was compared against an iterative baseline registration. We evaluated registration accuracy (the Dice and Hausdorff distance of the prostate and bladder contours), structural similarity index, and folding percentage to compare the methods. We also evaluated the framework's robustness to rigid and elastic deformations and bias field perturbations. RESULTS: The end-to-end trained framework comprising LapIRN for the deformable component achieved the best median (interquartile range) prostate and bladder Dice of 0.89 (0.85-0.91) and 0.86 (0.80-0.91), respectively. This accuracy was comparable to the iterative baseline registration: prostate and bladder Dice of 0.91 (0.88-0.93) and 0.86 (0.80-0.92). The best models complete rigid and deformable registration in 0.002 (0.0005) and 0.74 (0.43) s (Nvidia Tesla V100-PCIe 32 GB GPU), respectively. We found that the models are robust to translations up to 52 mm, rotations up to 15 ∘ $^\circ$ , elastic deformations up to 40 mm, and bias fields. CONCLUSIONS: Our proposed unsupervised, deep learning-based registration framework can perform rigid and deformable registration in less than a second with contour propagation accuracy comparable with iterative registration.

Foam-like phantoms for comparing tomography algorithms.

Tomographic algorithms are often compared by evaluating them on certain benchmark datasets. For fair comparison, these datasets should ideally (i) be challenging to reconstruct, (ii) be representative of typical tomographic experiments, (iii) be flexible to allow for different acquisition modes, and (iv) include enough samples to allow for comparison of data-driven algorithms. Current approaches often satisfy only some of these requirements, but not all. For example, real-world datasets are typically challenging and representative of a category of experimental examples, but are restricted to the acquisition mode that was used in the experiment and are often limited in the number of samples. Mathematical phantoms are often flexible and can sometimes produce enough samples for data-driven approaches, but can be relatively easy to reconstruct and are often not representative of typical scanned objects. In this paper, we present a family of foam-like mathematical phantoms that aims to satisfy all four requirements simultaneously. The phantoms consist of foam-like structures with more than 100000 features, making them challenging to reconstruct and representative of common tomography samples. Because the phantoms are computer-generated, varying acquisition modes and experimental conditions can be simulated. An effectively unlimited number of random variations of the phantoms can be generated, making them suitable for data-driven approaches. We give a formal mathematical definition of the foam-like phantoms, and explain how they can be generated and used in virtual tomographic experiments in a computationally efficient way. In addition, several 4D extensions of the 3D phantoms are given, enabling comparisons of algorithms for dynamic tomography. Finally, example phantoms and tomographic datasets are given, showing that the phantoms can be effectively used to make fair and informative comparisons between tomography algorithms.

Deep learning based classification of dynamic processes in time-resolved X-ray tomographic microscopy.

Time-resolved X-ray tomographic microscopy is an invaluable technique to investigate dynamic processes in 3D for extended time periods. Because of the limited signal-to-noise ratio caused by the short exposure times and sparse angular sampling frequency, obtaining quantitative information through post-processing remains challenging and requires intensive manual labor. This severely limits the accessible experimental parameter space and so, prevents fully exploiting the capabilities of the dedicated time-resolved X-ray tomographic stations. Though automatic approaches, often exploiting iterative reconstruction methods, are currently being developed, the required computational costs typically remain high. Here, we propose a highly efficient reconstruction and classification pipeline (SIRT-FBP-MS-D-DIFF) that combines an algebraic filter approximation and machine learning to significantly reduce the computational time. The dynamic features are reconstructed by standard filtered back-projection with an algebraic filter to approximate iterative reconstruction quality in a computationally efficient manner. The raw reconstructions are post-processed with a trained convolutional neural network to extract the dynamic features from the low signal-to-noise ratio reconstructions in a fully automatic manner. The capabilities of the proposed pipeline are demonstrated on three different dynamic fuel cell datasets, one exploited for training and two for testing without network retraining. The proposed approach enables automatic processing of several hundreds of datasets in a single day on a single GPU node readily available at most institutions, so extending the possibilities in future dynamic X-ray tomographic investigations.

Tomosipo: fast, flexible, and convenient 3D tomography for complex scanning geometries in Python.

Tomography is a powerful tool for reconstructing the interior of an object from a series of projection images. Typically, the source and detector traverse a standard path (e.g., circular, helical). Recently, various techniques have emerged that use more complex acquisition geometries. Current software packages require significant handwork, or lack the flexibility to handle such geometries. Therefore, software is needed that can concisely represent, visualize, and compute reconstructions of complex acquisition geometries. We present tomosipo, a Python package that provides these capabilities in a concise and intuitive way. Case studies demonstrate the power and flexibility of tomosipo.

Deep denoising for multi-dimensional synchrotron X-ray tomography without high-quality reference data.

Synchrotron X-ray tomography enables the examination of the internal structure of materials at submicron spatial resolution and subsecond temporal resolution. Unavoidable experimental constraints can impose dose and time limits on the measurements, introducing noise in the reconstructed images. Convolutional neural networks (CNNs) have emerged as a powerful tool to remove noise from reconstructed images. However, their training typically requires collecting a dataset of paired noisy and high-quality measurements, which is a major obstacle to their use in practice. To circumvent this problem, methods for CNN-based denoising have recently been proposed that require no separate training data beyond the already available noisy reconstructions. Among these, the Noise2Inverse method is specifically designed for tomography and related inverse problems. To date, applications of Noise2Inverse have only taken into account 2D spatial information. In this paper, we expand the application of Noise2Inverse in space, time, and spectrum-like domains. This development enhances applications to static and dynamic micro-tomography as well as X-ray diffraction tomography. Results on real-world datasets establish that Noise2Inverse is capable of accurate denoising and enables a substantial reduction in acquisition time while maintaining image quality.

Fluctuations of a membrane nanotube covered with an actin sleeve.

Many biological functions rely on the reshaping of cell membranes, in particular into nanotubes, which are covered in vivo by dynamic actin networks. Nanotubes are subject to thermal fluctuations, but the effect of these on cell functions is unknown. Here, we form nanotubes from liposomes using an optically trapped bead adhering to the liposome membrane. From the power spectral density of this bead, we study the nanotube fluctuations in the range of membrane tensions measured in vivo. We show that an actin sleeve covering the nanotube damps its high-frequency fluctuations because of the network viscoelasticity. Our work paves the way for further studies of the effect of nanotube fluctuations on cellular functions.

Female-Produced Sex Pheromone of Tetrastichus planipennisi, a Parasitoid Introduced for Biological Control of the Invasive Emerald Ash Borer, Agrilus planipennis.

The Asian eulophid wasp Tetrastichus planipennisi is being released in North America as a biocontrol agent for the emerald ash borer (Agrilus planipennis), a very destructive invasive buprestid beetle that is devastating ash trees (Fraxinus spp.). We identified, synthesized, and tested a female-produced sex pheromone for the wasp. The key component eliciting behavioral responses from male wasps in flight tunnel bioassays was identified as (6S,10S)-(2E,4E,8E)-4,6,8,10-tetramethyltrideca-2,4,8-triene. Female specificity was demonstrated by gas chromatographic (GC) comparison of male and female volatile emissions and whole body extracts. The identification was aided by coupled gas chromatography/mass spectrometry analysis, microchemical reactions, NMR, GC analyses with a chiral stationary phase column, and matching GC retention times and mass spectra with those of synthetic standards. The tetramethyl-triene hydrocarbon was synthesized as a mixture of two enantiomeric pairs of diastereomers, and as the pure insect-produced stereoisomer. In flight-tunnel bioassays, males responded to both the natural pheromone and the chiral synthetic material by upwind flight and landing on the source. In contrast, the mixture of four stereoisomers was not attractive, indicating that one or more of the "unnatural" stereoisomers antagonized attraction. Field trials, using yellow pan traps baited with natural pheromone, captured significantly more male wasps than control traps over a four week trial. The identified pheromone could increase the efficiency and specificity of the current detection methods for Tetrastichus planipennisi and aid in the determination of parasitoid establishment at release sites.

Actin modulates shape and mechanics of tubular membranes.

The actin cytoskeleton shapes cells and also organizes internal membranous compartments. In particular, it interacts with membranes for intracellular transport of material in mammalian cells, yeast, or plant cells. Tubular membrane intermediates, pulled along microtubule tracks, are formed during this process and destabilize into vesicles. While the role of actin in tubule destabilization through scission is suggested, literature also provides examples of actin-mediated stabilization of membranous structures. To directly address this apparent contradiction, we mimic the geometry of tubular intermediates with preformed membrane tubes. The growth of an actin sleeve at the tube surface is monitored spatiotemporally. Depending on network cohesiveness, actin is able to entirely stabilize or locally maintain membrane tubes under pulling. On a single tube, thicker portions correlate with the presence of actin. These structures relax over several minutes and may provide enough time and curvature geometries for other proteins to act on tube stability.

Report on the CCT Supplementary comparison S2 on thermal conductivity measurements of insulating materials by guarded hot plate.

Seven National Metrology Institutes (NMIs) from France, United States, United Kingdom, Russia, Mexico, China and Germany participated in an inter-laboratory comparison on thermal conductivity measurements by the Guarded Hot Plate method. This action was part of a series of supplementary inter-laboratory comparisons (including infrared spectral emittance and thermal diffusivity) sponsored by the Consultative Committee on Thermometry (CCT) Task Group on Thermophysical Quantities (TG-ThQ). The objective of this collaborative work was to strengthen the consistency of thermal conductivity measurements carried out worldwide on low conductive materials. Measurements were conducted successively by all participants on the same sets of specimens of insulating materials (mineral wool and expanded polystyrene) at temperatures ranging from 10 °C to 40 °C, according to the International Standard ISO 8302. This protocol aimed to minimize issues of material variability by circulating the same pairs of specimens among the laboratories following the strict format of a round-robin test program. More than 120 data points (combinations of material, thickness and temperature) were compared. 92 % of the data points were in agreement, with differences to weighted mean values less than the expanded uncertainties calculated from the individual NMI uncertainties and uncertainties related to the comparison process.

Absence of high-risk human papilloma virus in p16 positive inverted sinonasal papilloma.

OBJECTIVES: Sinonasal inverted papilloma (SIP) is a relatively rare disease, and its etiology is not understood. It is characterized by locally aggressive growth and a strong tendency to recur despite its benign histology. AIMS: The aim of this study was to identify the presence of human papilloma virus (HPV) and its surrogate marker p16 in SIP tissue samples from a regional cohort. MATERIAL AND METHODS: Subjects were identified from our regional center cohort of 88 SIP patients treated between 1984-2014. From these subjects, 54 were included in this study. Of these, 53 biopsies were analyzed with PCR, and 54 samples were immunohistochemically stained for p16. DNA was extracted from histopathologically verified SIP. Genotype screening for 13 high risk-, 5 oncogenic and 6 low risk HPV types was performed using the PapilloCheck® HPV-screening test. RESULTS: HPV analysis was successful for 38 of 53 samples. Of the 38 successfully analyzed samples, only 2 samples were positive for HPV 11. Notably, p16 was present in the epithelia in all samples, and in the papilloma lesions in 37 samples. CONCLUSION: Since only 2 out of 38 SIPs were positive for HPV (type 11), and at the same time p16 was positive in epithelia in all samples and in 37 of 38 papilloma lesions of the samples, it is concluded that p16 cannot be used as a surrogate marker for high-risk HPV-infection in SIP. We are currently planning a prospective, multicenter study in order to increase the study power and in order to be able to better evaluate the clinical implications of HPV-and p16 in SIP.

Segmentation of dental cone-beam CT scans affected by metal artifacts using a mixed-scale dense convolutional neural network.

PURPOSE: In order to attain anatomical models, surgical guides and implants for computer-assisted surgery, accurate segmentation of bony structures in cone-beam computed tomography (CBCT) scans is required. However, this image segmentation step is often impeded by metal artifacts. Therefore, this study aimed to develop a mixed-scale dense convolutional neural network (MS-D network) for bone segmentation in CBCT scans affected by metal artifacts. METHOD: Training data were acquired from 20 dental CBCT scans affected by metal artifacts. An experienced medical engineer segmented the bony structures in all CBCT scans using global thresholding and manually removed all remaining noise and metal artifacts. The resulting gold standard segmentations were used to train an MS-D network comprising 100 convolutional layers using far fewer trainable parameters than alternative convolutional neural network (CNN) architectures. The bone segmentation performance of the MS-D network was evaluated using a leave-2-out scheme and compared with a clinical snake evolution algorithm and two state-of-the-art CNN architectures (U-Net and ResNet). All segmented CBCT scans were subsequently converted into standard tessellation language (STL) models and geometrically compared with the gold standard. RESULTS: CBCT scans segmented using the MS-D network, U-Net, ResNet and the snake evolution algorithm demonstrated mean Dice similarity coefficients of 0.87 ± 0.06, 0.87 ± 0.07, 0.86 ± 0.05, and 0.78 ± 0.07, respectively. The STL models acquired using the MS-D network, U-Net, ResNet and the snake evolution algorithm demonstrated mean absolute deviations of 0.44 mm ± 0.13 mm, 0.43 mm ± 0.16 mm, 0.40 mm ± 0.12 mm and 0.57 mm ± 0.22 mm, respectively. In contrast to the MS-D network, the ResNet introduced wave-like artifacts in the STL models, whereas the U-Net incorrectly labeled background voxels as bone around the vertebrae in 4 of the 9 CBCT scans containing vertebrae. CONCLUSION: The MS-D network was able to accurately segment bony structures in CBCT scans affected by metal artifacts.

Sustainable treatment systems for removal of pharmaceutical residues and other priority persistent substances.

Pharmaceutical residues and other emerging substances commonly summarised as micropollutants pass through wastewater treatment plants (WWTPs) and end up in the receiving waters and sludge. Many studies have investigated the removal efficiency of various techniques but a holistic evaluation of various relevant treatment alternatives regarding both the removal efficiency for various micropollutants, investment and operating costs, environmental impacts and future comprehensiveness is still lacking. This paper provides the results from a large 3-year project about the evaluation of sustainable treatment systems for removal of various micropollutants or disruptive effects at Swedish WWTPs and their environmental, economic and future sustainability. The presented results are based on our own pilot tests and related assessment and modelling efforts and provide a holistic view on advanced treatment of wastewater for removal of micropollutants.

Pheromones of three ambrosia beetles in the Euwallacea fornicatus species complex: ratios and preferences.

Three cryptic species in the Euwallacea fornicatus species complex were reared in laboratory colonies and investigated for the presence of pheromones. Collections of volatiles from combinations of diet, fungus, beetles, and galleries from polyphagous shot hole borer (Euwallacea sp. #1) revealed the presence of 2-heneicosanone and 2-tricosanone only in the presence of beetles, regardless of sex. Subsequent examination of volatiles from the other two species, tea shot hole borer (Euwallacea sp. #2) and Kuroshio shot hole borer (Euwallacea sp. #5), revealed these two ketones were present in all three species but in different ratios. In dual choice olfactometer behavioral bioassays, mature mated females were strongly attracted to a synthetic binary blend of ketones matching their own natural ratios. However, females in each species were repelled by ketone blends in ratios corresponding to the other two species. Males of each species responded similarly to females when presented with ratios matching their own or the other two species. The presence of these compounds in the three beetle species, in ratios unique to each species, and their strong species-specific attraction and repellency, suggests they are pheromones. The ecological function of these pheromones is discussed. In addition to the pheromones, the previously known attractant (1S,4R)-p-menth-2-en-1-ol (also known as quercivorol) was discovered in the presence of the fungal symbionts, but not in association with the beetles. Quercivorol was tested in a dual-choice olfactometer and was strongly attractive to all three species. This evidence suggests quercivorol functions as a kairomone for members of the E. fornicatus species complex, likely produced by the symbiotic fungi.

Polyols, not sugars, determine the structural diversity of anti-streptococcal liamocins produced by Aureobasidium pullulans strain NRRL 50380.

Liamocins are polyol lipids produced by the fungus Aureobasidium pullulans, and have selective antibacterial activity against Streptococcus species. Liamocins produced by A. pullulans strain NRRL 50380 on sucrose medium have a d-mannitol head group ester-linked to 3,5-dihydroxydecanoate acyl chains, three or four of which are joined together by 1,5-polyester bonds (liamocins Man-A1 and Man-B1), and similar 3'-O-acetylated analogs (Man-A2 and Man-B2). However, other types of liamocins are produced depending on the choice of strain and growth conditions. In the current study, growth on different polyols, but not sugars, resulted in considerable structural variation, including liamocins with d-galactitol (dulcitol), d-sorbitol (glucitol), d- and l-arabitol, d-xylitol, l-threitol and glycerol head groups. The head groups of liamocins produced on arabitol were shown to be entirely composed of d-arabitol. These liamocin variants were structurally characterized by NMR and MS, and tested for antibacterial activity. The new liamocin variants also had selective activity against Streptococcus. Liamocin structural variants are novel antibacterials against Streptococcus sp. that merit further investigation.

Fluctuations of a membrane nanotube revealed by high-resolution force measurements.

Pulling membrane nanotubes from liposomes presents a powerful method to gain access to membrane mechanics. Here we extend classical optical tweezers studies to infer membrane nanotube dynamics with high spatial and temporal resolution. We first validate our force measurement setup by accurately measuring the bending modulus of EPC membrane in tube pulling experiments. Then we record the position signal of a trapped bead when it is connected, or not, to a tube. We derive the fluctuation spectrum of these signals and find that the presence of a membrane nanotube induces higher fluctuations, especially at low frequencies (10-1000 Hz). We analyse these spectra by taking into account the peristaltic modes of nanotube fluctuations. This analysis provides a new experimental framework for a quantitative study of the fluctuations of nanotubular membrane structures that are present in living cells, and now classically used for in vitro biomimetic approaches.

A Culture Method for Darkling Beetles, Blapstinus spp. (Coleoptera:Tenebrionidae).

Darkling beetles, Blapstinus spp., have become a serious pest of Cucurbitaceae crops, especially in California. A culture method was sought to provide large numbers (>500) of adult beetles of known age and sex that could be used for laboratory testing when needed. A method previously developed for Alphitobius diaperinus (Panzer) using a diet of ground chick feed, with apple slices as a moisture source, was modified for use with Blapstinus spp. and then compared with the same method substituting apple slices with zucchini as the moisture source. Rearing boxes set up with apple slices produced significantly more pupae and adults than boxes containing zucchini slices. However, using either zucchini or apples as a moisture source yielded over the target of 500 adults per rearing box. A previous method designed to sex A. diaperinus based on the presence (♀) or absence (♂) of second valvifers in the pupal stage also proved to be effective for sexing the Blapstinus spp.

Can Coffee Chemical Compounds and Insecticidal Plants Be Harnessed for Control of Major Coffee Pests?

Pests and pathogens threaten coffee production worldwide and are difficult to control using conventional methods, such as insecticides. We review the literature on the chemistry of coffee, concentrating on compounds most commonly reported from Coffea arabica and Coffea canephora. Differences in chemistry can distinguish coffee species and varieties, and plants grown under different biogeographic conditions exhibit different chemotypes. A number of chemical groups, such as alkaloids and caffeoylquinic acids, are known to be insecticidal, but most studies have investigated their effects on coffee quality and flavor. More research is required to bridge this gap in knowledge, so that coffee can be bred to be more resistant to pests. Furthermore, we report on some pesticidal plants that have been used for control of coffee pests. Locally sourced pesticidal plants have been underutilized and offer a sustainable alternative to conventional insecticides and could be used to augment breeding for resilience of coffee plants.

Prolonged outbreak of adenovirus A31 in allogeneic stem cell transplant recipients.

BACKGROUND: An outbreak of human adenovirus (HAdV) A31 occurred from December 2011 to March 2012 at the Center for Allogeneic Stem Cell Transplantation (CAST), Karolinska University Hospital in Sweden. We analyzed the outbreak, the routes of transmission, and report the medical consequences. METHODS: The medical records of all patients admitted to CAST during the outbreak period were studied. Phylogenetic analysis of the patient HAdV strains was performed by sequencing the hexon gene and the more variable E3 gene. RESULTS: We identified 9 cases of HAdV A31. Hygiene measures were implemented, but transmission continued for 2 months. All 9 patients had been admitted to the ward, but 2 had no connection in time to other known HAdV A31 cases. DNA sequencing of the patient strains strongly suggested nosocomial transmission. Transplantation was postponed and then cancelled in 1 patient, and 5 patients were treated with cidofovir because of high levels of viremia. In 7 patients, concomitant graft-versus-host disease (GVHD) grade II-V complicated the clinical picture, as it was difficult to distinguish symptoms of GVHD from those of HAdV infection. CONCLUSION: An outbreak of HAdV in HSCT recipients can be difficult to control. Although none of the patients had severe disease, the medical consequences were significant. It is possible that unidentified cases with mild symptoms may have caused continuous transmission at the unit. Regular testing of all patients several weeks beyond the last case identified may be an important measure to control transmission.

Development and Evaluation of a Loop-Mediated Isothermal Amplification Assay for the Detection of Adenovirus 40 and 41.

Human adenoviruses (hAdVs) of subgroup F (enteric serotypes 40 and 41) display characteristic gut tropism, in vivo, fastidious growth characteristics in cell culture, and are estimated to be associated with 5-20% worldwide of acute gastroenteritis cases among infants and young children. Adequate hAdV gastroenteritis case management requires laboratory-based diagnosis. The present study aimed to the development and evaluation of a simple and cost-effective, one-step, single-tube adenovirus type 40/41 specific loop-mediated isothermal amplification (LAMP) assay for the detection of hAdV40/41 DNA in environmental and/or clinical samples, since no LAMP assay has previously been reported for the detection of these virus types. The assay targeted the hexon gene and had the advantages of being rapid, simple, specific, and sensitive. Results could be obtained within 60 min, under isothermal conditions at 69 °C. The detection limits for hAdV genomes were between 50 and 100 copies/reaction for hAdV40 and hAdV41, and no cross-reactions with other selected viruses, were found. The assay was evaluated with clinical as well as environmental samples. The developed assay is expected to provide a potential molecular tool in obtaining greater knowledge of the hAdV40/41 importance in the epidemiology and clinical manifestations of gastroenteritis.