Search

Generative molecular design (GMD) is an increasingly popular strategy for drug discovery, using machine learning models to propose, evaluate and optimize chemical structures against a set of target design criteria. We present the ATOM-GMD platform, a scalable multiprocessing framework to optimize many parameters simultaneously over large populations of proposed molecules. ATOM-GMD uses a junction tree variational autoencoder mapping structures to latent vectors, along with a genetic algorithm operating on latent vector elements, to search a diverse molecular space for compounds that meet the design criteria. We used the ATOM-GMD framework in a lead optimization case study to develop potent and selective histamine H1 receptor antagonists. We synthesized 103 of the top scoring compounds and measured their properties experimentally. Six of the tested compounds bind H1 withKi’s between 10 and 100 nM and are at least 100-fold selective relative to muscarinic M2 receptors, validating the effectiveness of our GMD approach.

Multistatic ground-penetrating radar (GPR) signals can be imaged tomographically to produce three-dimensional distributions of image intensities. In the absence of objects of interest, these intensities can be considered to be estimates of clutter. These clutter intensities spatially vary over several orders of magnitude, and vary across different arrays, which makes direct comparison of these raw intensities difficult. However, by gathering statistics on these intensities and their spatial variation, a variety of metrics can be determined. In this study, the clutter distribution is found to fit better to a two-parameter Weibull distribution than Gaussian or lognormal distributions. Based upon the spatial variation of the two Weibull parameters, scale and shape, more information may be gleaned from these data. How well the GPR array is illuminating various parts of the ground, in depth and cross-track, may be determined from the spatial variation of the Weibull scale parameter, which may in turn be used to estimate an effective attenuation coefficient in the soil. The transition in depth from clutter-limited to noise-limited conditions (which is one possible definition of GPR penetration depth) can be estimated from the spatial variation of the Weibull shape parameter. Finally, the underlying clutter distributions also provide an opportunity to standardize image intensities to determine when a statistically significant deviation from background (clutter) has occurred, which is convenient for buried threat detection algorithm development which needs to be robust across multiple different arrays., 12 pages, 11 figures

New ultrasound data, obtained with a circular experimental scanner, are compared with data obtained with standard X-ray CT. Ultrasound data obtained by scanning fixed breast tissue were used to generate images of sound speed and reflectivity. The ultrasound images exhibit approximately 1 mm resolution and about 20 dB of dynamic range. All data were obtained in a circular geometry. X-ray CT scans were used to generate X-ray images corresponding to the same 'slices' obtained with the ultrasound scanner. The good match of sensitivity, resolution and angular coverage between the ultrasound and X-ray data makes possible a direct comparison of the three types of images. We present the results of such a comparison for an excised breast fixed in formalin. The results are presented visually using various types of data fusion. A general correspondence between the sound speed, reflectivity and X-ray morphologies is found. The degree to which data fusion can help characterize tissue is assessed by examining the quantitative correlations between the ultrasound and X-ray images.

In contrast to standard reflection ultrasound (US), transmission US holds the promise of more thorough tissue characterization by generating quantitative acoustic parameters. We compare results from a conventional US scanner with data acquired using an experimental circular scanner operating at frequencies of 0.3 - 1.5 MHz. Data were obtained on phantoms and a normal, formalin-fixed, excised breast. Both reflection and transmission-based algorithms were used to generate images of reflectivity, sound speed and attenuation.. Images of the phantoms demonstrate the ability to detect sub-mm features and quantify acoustic properties such as sound speed and attenuation. The human breast specimen showed full field evaluation, improved penetration and tissue definition. Comparison with conventional US indicates the potential for better margin definition and acoustic characterization of masses, particularly in the complex scattering environments of human breast tissue. The use of morphology, in the context of reflectivity, sound speed and attenuation, for characterizing tissue, is discussed.

Extremely high quality data was acquired using an experimental ultrasound scanner developed at Lawrence Livermore National Laboratory using a 2D ring geometry with up to 720 transmitter/receiver transducer positions. This unique geometry allows reflection and transmission modes and transmission imaging and quantification of a 3D volume using 2D slice data. Standard image reconstruction methods were applied to the data including straight-ray filtered back projection, reflection tomography, and diffraction tomography. Newer approaches were also tested such as full wave, full wave adjoint method, bent-ray filtered back projection, and full-aperture tomography. A variety of data sets were collected including a formalin-fixed human breast tissue sample, a commercial ultrasound complex breast phantom, and cylindrical objects with and without inclusions. The resulting reconstruction quality of the images ranges from poor to excellent. The method and results of this study are described including like-data reconstructions produced by different algorithms with side-by-side image comparisons. Comparisons to medical B-scan and x-ray CT scan images are also shown. Reconstruction methods with respect to image quality using resolution, noise, and quantitative accuracy, and computational efficiency metrics will also be discussed.

Current ground penetrating radars (GPR) have been tested for land mine detection, but they have generally been costly and have poor performance. Comprehensive modeling and experimentation must be done to predict the electromagnetic (EM) signatures of mines to access the effect of clutter on the EM signature of the mine, and to understand the merit and limitations of using radar for various mine detection scenarios. This modeling can provide a basis for advanced radar design and detection techniques leading to superior performance. Lawrence Livermore National Laboratory (LLNL) has developed a radar technology that when combined with comprehensive modeling and detection methodologies could be the basis of an advanced mine detection system. Micropower Impulse Radar (MIR) technology exhibits a combination of properties, including wideband operation, extremely low power consumption, extremely small size and low cost, array configurability, and noise encoded pulse generation. LLNL is in the process of developing an 'optimal' processing algorithm to use with the MIR sensor. In this paper, we use classical numerical models to obtain the signature of mine-like targets and examine the effect of surface roughness on the reconstructed signals. These results are then qualitatively compared to experimental data.

Corrosion of rebar in concrete bridges causes subsurface cracks and is a major cause of structural degradation that necessitates repair or replacement. Early detection of corrosion effects can limit the location and extent of necessary repairs, while providing long-term information about the infrastructure status. Most current detection methods, however, are destructive of the road surface and require closing or restricting traffic while the tests are performed. A ground-penetrating radar imaging system has been designed and developed that will perform the nondestructive evaluation of road-bed cracking at traffic speeds; i.e., without the need to restrict traffic flow. The first-generation system (called the HERMES bridge inspector), consists of an offset-linear array of 64 impulse radar transceivers and associated electronics housed in a trailer. Computers in the trailer and in the towing vehicle control the data acquisition, processing, and display. Cross-road resolution is three centimeters at up to 30 cm in depth, while down-road resolution depends on speed; 3 cm below 20 mph up to 8 cm at 50 mph. A two-meter- wide path is inspected on each pass over the roadway. This paper, describes the design of this system, shows preliminary results, and lays out its deployment schedule.

Lawrence Livermore National Laboratory (LLNL)l is developing a prototype imaging radar for inspecting steel reinforced concrete bridge decks. The system is designed to acquire Synthetic Aperture Radar (SAR) data and provide high-resolution images of internal structure, flaws, and defects enabling bridge inspectors to nondestructively evaluate and characterize bridge deck condition. Concrete delamination resulting from corrosion of steel reinforcing bars (rebars) is an important structural defect that the system is designed to detect. The prototype system uses arrays of compact, low-cost Micropower Impulse Radar (MIR) modules, supported by appropriate data acquisition and storage subsystems, to generate and collect the radar data, and unique imaging codes to reconstruct images of bridge deck internals. In this paper, we provide an overview of the prototype system concept, discuss its expected performance, and present recent experimental results showing the capability of this approach to detect thin delamination simulations embedded in concrete.

In this paper we present results from a three-dimensional image reconstruction algorithm for impulse radar operating in monostatic pulse-echo mode. The application of interest to us is the nondestructive evaluation of civil structures such as bridge decks. We use a multi-frequency diffraction tomography imaging technique in which coherent backward propagations of the received reflected wavefield form a spatial image of the scattering interfaces within the region of interest. This imaging technique provides high-resolution range and azimuthal visualization of the subsurface region. We incorporate the ability to image in planarly layered conductive media and apply the algorithm to experimental data from an offset radar system in which the radar antenna is not directly coupled to the surface of the region. We present a rendering in three-dimensions of the resulting image data which provides high-detail visualization.

This paper discusses a three-dimensional synthetic aperture imaging technique based on time- domain focusing of pulse-echo radar data. We describe the basic image formation process, important data processing issues, and compensation for planar variations in the media. We present a high-resolution volumetric image reconstruction of a concrete test slab and show that we are able to identify steel reinforcing bars in the image. We conclude with a brief comparison of this imaging method with a technique based on diffraction tomography.© (1994) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

A method and system for examining tissue are provided in which the tissue is maintained in a position so that it may be insonified with a plurality of pulsed spherical or cylindrical acoustic waves. The insonifying acoustic waves are scattered by the tissue so that scattered acoustic radiation including a mix of reflected and transmitted acoustic waves is received. A representation of a portion of the tissue is then derived from the received scattered acoustic radiation.