Search

Whereas traditional histology and light microscopy require multiple steps of formalin fixation, paraffin embedding, and sectioning to generate images for pathologic diagnosis, Microscopy using Ultraviolet Surface Excitation (MUSE) operates through UV excitation on the cut surface of tissue, generating images of high resolution without the need to fix or section tissue and allowing for potential use for downstream molecular tests. Here, we present the first study of the use and suitability of MUSE microscopy for neuropathological samples. MUSE images were generated from surgical biopsy samples of primary and metastatic brain tumor biopsy samples (n = 27), and blinded assessments of diagnoses, tumor grades, and cellular features were compared to corresponding hematoxylin and eosin (H&E) images. A set of MUSE-treated samples subsequently underwent exome and targeted sequencing, and quality metrics were compared to those from fresh frozen specimens. Diagnostic accuracy was relatively high, and DNA and RNA integrity appeared to be preserved for this cohort. This suggests that MUSE may be a reliable method of generating high-quality diagnostic-grade histologic images for neuropathology on a rapid and sample-sparing basis and for subsequent molecular analysis of DNA and RNA., Competing Interests: M.L. was employed by the Foundation Medicine, Inc., R.L. is a co-founder of start-up MUSE Microscopy Inc. (now part of SmartPath Dx), and R.L., F.F. and A.B. are co-founders of start-up Histolix Inc. MUSE Microscopy Inc. and Histolix Inc both had no role in the design of the study, in the collection, analyses, or interpretation of data, in the writing of the manuscript, or in the decision to publish the results. The other authors declare no conflicts of interest.

Infrasonic waves have been observed to propagate to regional (greater than 15 km) distances through the troposphere. Infrasound propagation in the geometric acoustics approximation has shown that realistic terrain can scatter acoustic energy from tropospheric ducts; however, ray methods cannot intrinsically capture finite-frequency behavior such as diffraction. A two-dimensional finite-difference time-domain (FDTD) method has been developed to solve linearized equations for infrasound propagation with realistic terrain. Acoustic wave propagation over 100 km with both flat terrain and a Gaussian hill was first simulated in order to compare finite-frequency propagation with ray predictions. The effects of realistic terrain and atmospheres on infrasound signals from a 2012 surface explosion at the Utah Testing and Training Range are then investigated. Propagation through the troposphere is suggested by array processing results, but eigenrays are not predicted due to weak to nonexistent ducting conditions. FDTD modeling suggests that the inclusion of terrain and finite frequency effects helps explain much of the observed signal in a realistic scenario. These results suggest that geometric acoustics may underestimate propagation through the troposphere, and that recorded waveforms at regional distances may be noticeably affected by terrain., (© 2024 Acoustical Society of America.)

We develop a deep learning-based infrasonic detection and categorization methodology that uses convolutional neural networks with self-attention layers to identify stationary and non-stationary signals in infrasound array processing results. Using features extracted from the coherence and direction-of-arrival information from beamforming at different infrasound arrays, our model more reliably detects signals compared with raw waveform data. Using three infrasound stations maintained as part of the International Monitoring System, we construct an analyst-reviewed data set for model training and evaluation. We construct models using a 4-category framework, a generalized noise vs non-noise detection scheme, and a signal-of-interest (SOI) categorization framework that merges short duration stationary and non-stationary categories into a single SOI category. We evaluate these models using a combination of k-fold cross-validation, comparison with an existing "state-of-the-art" detector, and a transportability analysis. Although results are mixed in distinguishing stationary and non-stationary short duration signals, f-scores for the noise vs non-noise and SOI analyses are consistently above 0.96, implying that deep learning-based infrasonic categorization is a highly accurate means of identifying signals-of-interest in infrasonic data records.