Abstract:Hemeproteins
are frequent subjects for ultrafast transient absorption
spectroscopy (TAS) because of biological importance, strong UV–vis
absorption, high photostability, and interesting transient dynamics
that depend on redox, conformation, and ligand binding. TAS on hemeproteins
is usually performed on isolated, purified proteins, though their
response is likely to be different in their native molecular environment,
which involves the formation of protein complexes and supercomplexes.
Recently, we reported a … Show more
“…In addition, PCA/SVD and phase require extra steps to identify endmembers from plots of the projections and then estimate abundance maps. Recently, we have adapted multivariate curve resolution (MCR) 1 for transient absorption analysis, which directly recovers non-orthogonal endmembers and their abundance maps, but in practice we have found it to require an initial estimate of all but one endmember from independent measurements 2,3 .…”
We describe a simple convolutional network for blind unmixing of transient absorption microscopy data along with a model ensembling strategy. Our network is based on an autoencoder previously developed for blind unmixing of hyperspectral satellite images. Its advantages are (a) that it learns to unmix spectra by unsupervised learning, i.e. by learning to reconstruct imaging data, without knowledge of the underlying spectra or their abundances and (b) that the endmember spectra are directly encoded by the output layer’s coefficients. Extensive modifications to both the network architecture and training loss functions were necessary to produce reasonable performance on transient absorption data. We demonstrate results from blind unmixing of transient absorption images of unstained muscle fibers, acquired at 520 nm pump and 620 nm probe, training an ensemble of 500 different networks (i.e. unmixing models), each starting from a different random initialization. Variability among resulting models was analyzed by principal component analysis on the recovered endmembers from all models, deriving from the projections a model probability density function. We found consistent models (predicting similar endmembers and abundance maps) near the most likely model and surrounding high-probability region, with more variability in low-probability regions. Then, a permutation-aligned average of the ensemble produced much better results than an unweighted ensemble average, or simple selection of one model based on maximum likelihood or best fit. We anticipate this approach of parametrizing models and ensembling based on relative probability to have applications in other chemical imaging modalities such as FLIM, Raman microscopy and mass spectroscopic imaging.
“…In addition, PCA/SVD and phase require extra steps to identify endmembers from plots of the projections and then estimate abundance maps. Recently, we have adapted multivariate curve resolution (MCR) 1 for transient absorption analysis, which directly recovers non-orthogonal endmembers and their abundance maps, but in practice we have found it to require an initial estimate of all but one endmember from independent measurements 2,3 .…”
We describe a simple convolutional network for blind unmixing of transient absorption microscopy data along with a model ensembling strategy. Our network is based on an autoencoder previously developed for blind unmixing of hyperspectral satellite images. Its advantages are (a) that it learns to unmix spectra by unsupervised learning, i.e. by learning to reconstruct imaging data, without knowledge of the underlying spectra or their abundances and (b) that the endmember spectra are directly encoded by the output layer’s coefficients. Extensive modifications to both the network architecture and training loss functions were necessary to produce reasonable performance on transient absorption data. We demonstrate results from blind unmixing of transient absorption images of unstained muscle fibers, acquired at 520 nm pump and 620 nm probe, training an ensemble of 500 different networks (i.e. unmixing models), each starting from a different random initialization. Variability among resulting models was analyzed by principal component analysis on the recovered endmembers from all models, deriving from the projections a model probability density function. We found consistent models (predicting similar endmembers and abundance maps) near the most likely model and surrounding high-probability region, with more variability in low-probability regions. Then, a permutation-aligned average of the ensemble produced much better results than an unweighted ensemble average, or simple selection of one model based on maximum likelihood or best fit. We anticipate this approach of parametrizing models and ensembling based on relative probability to have applications in other chemical imaging modalities such as FLIM, Raman microscopy and mass spectroscopic imaging.
“…Whilst providing adequate peak powers to trigger various non-linear processes of interest, these laser sources offer an improved signal-to-noise ratio and fast data acquisition time-scales. This is vital in combating space-charge effects in condensed-phase photoemission processes, improving low-yield coincidence measurements, or in experiments with naturally degradable samples 9 – 17 . Figure 1 OPCPA design.…”
We report on an ultrafast infrared optical parametric chirped-pulse amplifier (OPCPA), pumped by a 200-W thin-disk Yb-based regenerative amplifier at a repetition rate of 100 kHz. The OPCPA is tunable in the spectral range 1.4–3.9 $$\upmu $$
μ
m, generating up to 23 W of < 100-fs signal and 13 W of < 200-fs idler pulses for infrared spectroscopy, with additional spectral filtering capabilities for Raman spectroscopy. The OPCPA can also yield 19 W of 49-fs 1.75-$$\upmu $$
μ
m signal or 5 W of 62-fs 2.8-$$\upmu $$
μ
m idler pulses with active carrier-to-envelope-phase (CEP) stabilisation for high-harmonic generation (HHG). We illustrate the versatility of the laser design, catering to various experimental requirements for probing ultrafast science.
Mitochondrial redox is an important indicator of cell metabolism and
health, with implications in cancer, diabetes, aging,
neurodegenerative diseases, and mitochondrial disease. The most common
method to observe redox of individual cells and mitochondria is
through fluorescence of NADH and FAD+, endogenous cofactors
serve as electron transport inputs to the mitochondrial respiratory
chain. Yet this leaves out redox within the respiratory chain itself.
To a degree, the missing information can be filled in by exogenous
fluorophores, but at the risk of disturbed mitochondrial permeability
and respiration. Here we show that variations in respiratory chain
redox can be detected up by visible-wavelength transient absorption
microscopy (TAM). In TAM, the selection of pump and probe wavelengths
can provide multiphoton imaging contrast between non-fluorescent
molecules. Here, we applied TAM with a pump at 520nm and probe at
450nm, 490nm, and 620nm to elicit redox contrast from mitochondrial
respiratory chain hemeproteins. Experiments were performed with
reduced and oxidized preparations of isolated mitochondria and whole
muscle fibers, using mitochondrial fuels (malate, pyruvate, and
succinate) to set up physiologically relevant oxidation levels. TAM
images of muscle fibers were analyzed with multivariate curve
resolution (MCR), revealing that the response at 620nm probe provides
the best redox contrast and the most consistent response between whole
cells and isolated mitochondria.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
