Diffuse optical imaging is an effective technique for noninvasive functional brain imaging. However, the measurements respond to systemic hemodynamic fluctuations caused by the cardiac cycle, respiration, and blood pressure, which may obscure or overwhelm the desired stimulus-evoked response. Previous work on this problem employed temporal filtering, estimation of systemic effects from background pixels, or modeling of interference signals with predefined basis functions, with some success. However, weak signals are still lost in the interference, and other complementary methods are desirable. We use the spatial behavior of measured baseline signals to identify the interference subspaces. We then project signals components in this subspace out of the stimulation data. In doing so, we assume that systemic interference components will be more global spatially, with higher energy, than the stimulus-evoked signals of interest. Thus, the eigenvectors corresponding to the largest eigenvalues of an appropriate correlation matrix form the basis for an interference subspace. By projecting the data onto the orthogonal nullspace of these eigenvectors, we can obtain more localized response, as reflected in improved contrast-to-noise ratio and correlation coefficient maps.

The recent advances in deep neural networks have convincingly demonstrated high capability in learning vision models on large datasets. Nevertheless, collecting expert labeled datasets especially with pixel-level annotations is an extremely expensive process. An appealing alternative is to render synthetic data (e.g., computer games) and generate ground truth automatically. However, simply applying the models learnt on synthetic images may lead to high generalization error on real images due to domain shift. In this paper, we facilitate this issue from the perspectives of both visual appearance-level and representation-level domain adaptation. The former adapts source-domain images to appear as if drawn from the "style" in the target domain and the latter attempts to learn domain-invariant representations. Specifically, we present Fully Convolutional Adaptation Networks (FCAN), a novel deep architecture for semantic segmentation which combines Appearance Adaptation Networks (AAN) and Representation Adaptation Networks (RAN). AAN learns a transformation from one domain to the other in the pixel space and RAN is optimized in an adversarial learning manner to maximally fool the domain discriminator with the learnt source and target representations. Extensive experiments are conducted on the transfer from GTA5 (game videos) to Cityscapes (urban street scenes) on semantic segmentation and our proposal achieves superior results when comparing to state-of-theart unsupervised adaptation techniques. More remarkably, we obtain a new record: mIoU of 47.5% on BDDS (drivecam videos) in an unsupervised setting.

Digital tomosynthesis mammography (DTM) is a promising new modality for breast cancer detection. In DTM, projection-view images are acquired at a limited number of angles over a limited angular range and the imaged volume is reconstructed from the two-dimensional projections, thus providing three-dimensional structural information of the breast tissue. In this work, we investigated three representative reconstruction methods for this limited-angle cone-beam tomographic problem, including the backprojection (BP) method, the simultaneous algebraic reconstruction technique (SART) and the maximum likelihood method with the convex algorithm (ML-convex). The SART and ML-convex methods were both initialized with BP results to achieve efficient reconstruction. A second generation GE prototype tomosynthesis mammography system with a stationary digital detector was used for image acquisition. Projection-view images were acquired from 21 angles in 3 degrees increments over a +/- 30 degrees angular range. We used an American College of Radiology phantom and designed three additional phantoms to evaluate the image quality and reconstruction artifacts. In addition to visual comparison of the reconstructed images of different phantom sets, we employed the contrast-to-noise ratio (CNR), a line profile of features, an artifact spread function (ASF), a relative noise power spectrum (NPS), and a line object spread function (LOSF) to quantitatively evaluate the reconstruction results. It was found that for the phantoms with homogeneous background, the BP method resulted in less noisy tomosynthesized images and higher CNR values for masses than the SART and ML-convex methods. However, the two iterative methods provided greater contrast enhancement for both masses and calcification, sharper LOSF, and reduced interplane blurring and artifacts with better ASF behaviors for masses. For a contrast-detail phantom with heterogeneous tissue-mimicking background, the BP method had strong blurring artifacts along the x-ray source motion direction that obscured the contrast-detail objects, while the other two methods can remove the superimposed breast structures and significantly improve object conspicuity. With a properly selected relaxation parameter, the SART method with one iteration can provide tomosynthesized images comparable to those obtained from the ML-convex method with seven iterations, when BP results were used as initialization for both methods.

No abstract

Purpose: The purpose of this study is to investigate the feasibility of increasing the system spatial resolution and scanning speed of Hologic Selenia Dimensions digital breast tomosynthesis (DBT) scanner by replacing the rotating mammography x-ray tube with a specially designed carbon nanotube (CNT) x-ray source array, which generates all the projection images needed for tomosynthesis reconstruction by electronically activating individual x-ray sources without any mechanical motion. The stationary digital breast tomosynthesis (s-DBT) design aims to (i) increase the system spatial resolution by eliminating image blurring due to x-ray tube motion and (ii) reduce the scanning time. Low spatial resolution and long scanning time are the two main technical limitations of current DBT technology. Methods: A CNT x-ray source array was designed and evaluated against a set of targeted system performance parameters. Simulations were performed to determine the maximum anode heat load at the desired focal spot size and to design the electron focusing optics. Field emission current from CNT cathode was measured for an extended period of time to determine the stable life time of CNT cathode for an expected clinical operation scenario. The source array was manufactured, tested, and integrated with a Selenia scanner. An electronic control unit was developed to interface the source array with the detection system and to scan and regulate x-ray beams. The performance of the s-DBT system was evaluated using physical phantoms. Results: The spatially distributed CNT x-ray source array comprised 31 individually addressable x-ray sources covering a 30 angular span with 1 pitch and an isotropic focal spot size of 0.6 mm at full width at half-maximum. Stable operation at 28 kV(peak) anode voltage and 38 mA tube current was demonstrated with extended lifetime and good source-to-source consistency. For the standard imaging protocol of 15 views over 14, 100 mAs dose, and 2 Â 2 detector binning, the projection resolution along the scanning direction increased of 1.08. The improvement is more pronounced for faster scanning speeds, wider angular coverage, and smaller detector pixel sizes. The scanning speed depends on the detector, the number of views, and the imaging dose. With 240 ms detector readout time, the s-DBT system scanning time is 6.3 s for a 15-view, 100 mAs scan regardless of the angular coverage. The scanning speed can be reduced to less than 4 s when detectors become faster. Initial phantom studies showed good quality reconstructed images. Conclusions: A prototype s-DBT scanner has been developed and evaluated by retrofitting the Selenia rotating gantry DBT scanner with a spatially distributed CNT x-ray source array. Preliminary results show that it improves system spatial resolution substantially by eliminating image blur due to x-ray focal spot motion. The scanner speed of s-DBT system is independent of angular coverage and can be increased with faster detector without image degration. The accelerated lifetime measurement demo...

Pyrene derivatives can absorb onto the surface of carbon nanotubes and graphite particles through pi-pi interactions to functionalize these inorganic building blocks with organic surface moieties. Using single molecule force spectroscopy, we have demonstrated the first direct measurement of the interaction between pyrene and a graphite surface. In particular, we have connected a pyrene molecule onto an AFM tip via a flexible poly(ethylene glycol) (PEG) chain to ensure the formation of a molecular bridge. The pi-pi interaction between pyrene and graphite is thus indicated to be approximately 55 pN with no hysteresis between the desorption and adhesion forces.

In this paper, we propose a Customizable Architecture Search (CAS) approach to automatically generate a network architecture for semantic image segmentation. The generated network consists of a sequence of stacked computation cells. A computation cell is represented as a directed acyclic graph, in which each node is a hidden representation (i.e., feature map) and each edge is associated with an operation (e.g., convolution and pooling), which transforms data to a new layer. During the training, the CAS algorithm explores the search space for an optimized computation cell to build a network. The cells of the same type share one architecture but with different weights. In real applications, however, an optimization may need to be conducted under some constraints such as GPU time and model size. To this end, a cost corresponding to the constraint will be assigned to each operation. When an operation is selected during the search, its associated cost will be added to the objective. As a result, our CAS is able to search an optimized architecture with customized constraints. The approach has been thoroughly evaluated on Cityscapes and CamVid datasets, and demonstrates superior performance over several stateof-the-art techniques. More remarkably, our CAS achieves 72.3% mIoU on the Cityscapes dataset with speed of 108 FPS on an Nvidia TitanXp GPU.

A methodology was developed to determine the mass concentrations of cellulase and cells applicable to studies of microbial cellulose utilization in systems for which a substantial fraction of cellulase is cell-associated. Antibodies raised against a 14-amino acid synthetic peptide with sequence taken from the cohesin domain of the scaffoldin protein of Clostridium thermocellum ATCC 27405 were used to develop an indirect ELISA protocol. Six cellulase calibration standards were prepared using affinity digestion (Morag, E.; Bayer, E. A.; Lamed, R. Enzyme Microb. Technol. 1992, 14, 289-292.). These included supernatant and pellet samples from an Avicelgrown culture with fractional cellulose conversion (X) = 0.98, as well as supernatant, pellet, cell-associated, and cellulose-associated samples from an Avicel-grown culture with X = 0.8. All six standards displayed a very similar absorbance versus concentration relationship when subjected to ELISA, essentially identical SDS-PAGE banding patterns, and similar cellulase specific activity in relation to both other purified cellulase preparations and crude samples. Coefficients of variation for cellulase concentration measurements were 5.2% for supernatant samples and 5.9% for pellet samples. The ELISA method was applied to batch cultures of C. thermocellum grown on Avicel. Cell concentration was calculated from the pellet protein concentration and the cell protein fraction of a cellobiose-grown control. Two alternative methods appeared to overpredict the cell concentration and were not capable of quantifying cells as distinct from cellulase. Cellulase protein production by Avicel-grown batch cultures represented approximately 20% of cell mass exclusive of cellulase. It is concluded that the reported protocols establish a reasonable methodological basis for quantitative determination of the mass concentration of cellulase protein produced by C. thermocellum and for calculation of cell mass concentration as distinct from cellulase concentration.