jackman/methodsPaper
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

rev abstr

Browse Source
This commit is contained in:
ackman678
2022-04-28 15:07:03 -07:00
parent 2a17a47c09
commit 87f9c920b2
2 changed files with 54 additions and 32 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
comments.md
View File

@@ -5,6 +5,8 @@
* Q: Should a measure of information rate per recording be reported?
- i.e. the data flow amounts in and out of optimized vs non-optimized imaging sessions
<!-- figure 2D: What is the value of the mean neural component curve when it levels off? Maybe around 1000 s or 16.6 min? -->
<!-- * Q: What is the dynamic range of our cortical calcium signal before and after filtration? (e.g. raw data vs rICA data) -->
<!-- * Q: How many bits of the cameras ADC are needed to encode our calcium signal variance? What about for calcium signal variance+hemodynamic variance+vascular artifact variance all together on the same monochromatic channel? -->
@@ -99,6 +101,17 @@
## Results
### ICA separates signal sources from high resolution data
### High resolution spontaneous activity improves noise separation and increasing data length results in a stable number of signal components
### Spatiotemporal metrics can be derived from each component to assess the classification of each signal source
### GCaMP mice have strong distinct globular domains that cover the entire cortical surface
### Spatial metrics best separate neural components from artifacts
### Machine learning performs as well as human classification
### Global mean needs a high-pass filter to account for removed artifacts before re-addition
### Domain maps optimize time course extraction from underlying data
### Animal specific domain maps can be regionalized based on reference maps and domain features
* [x] Check when Fig. S6 is referred to in text
* [ ] Check the um/px value for the lateral spatial resolution
* [ ] Check colormap dots overlay in figureS7
@@ -110,10 +123,23 @@
* [ ] Check phrase 'detected regions' as in 'We additionally quantified whether detected regions...'
* [ ] Fix Figure S6 "Nueral"
* [ ] rw "considered collecting spatial samples higher than our current resolution"
## Methods
### Mice
### Surgical procedure
### Recording calcium dynamics
### ICA decomposition and saving
### Dynamic Thresholding
### ICA and Data processing
### Metric generation and classification of Neural Independent Components
### Map creation and comparisons
### Compression and filtering residuals
### Domain residuals and domain signal analyses
### Statistical significance
* [ ] rw 'components that are unsorted and often flipped'
* [ ] rw 'mean time series is pre-subtracted from the array before SVD'SVD used before defined in next paragraph. The mean signal effect is removed with the pre-whitening/sphereing step of SVD

60
main.md
View File

@@ -20,13 +20,27 @@ Sydney C. Weiser¹•, Brian R. Mullen¹•, Desiderio Ascencio², & James B. Ac
²Department of Psychology, University of California Santa Cruz, Santa Cruz, CA, USA
## Abstract
Demixing neural signals from artifact signals in videos of neuronal calcium flux across the cerebral hemispheres could reveal key functional features of cortical organization. Here we demonstrate that the multichannel blind source deconvolution algorithm, independent component analysis, can optimally recover neural signal content in CMOS sensor imaging of pan-neuronal cortical calcium dynamics at a sampling of 1.5M cortical pixels per 0.1 s for 17 min. We characterize each component using a set of spatial and temporal metrics and build a random forest classifier that separates neural activity and artifact components automatically with human performance. We show how this data produces a functional tesselation of the neocortical sheet, providing a map of 230±14 domains from which extracted time courses maximally represent the underlying signal in each recording. This workflow of data-driven video decomposition and machine classification of signal sources will aid high quality mapping of complex cerebral dynamics.
Demixing neural signals from artifact signals in videos of neuronal calcium flux across the cerebral hemispheres could help reveal functional features of cortical organization. Here we demonstrate that the general solution to multichannel source signal separation, independent component analysis, can optimally recover neural signal content in recordings of neuronal cortical calcium dynamics captured at a rate of 1.5×10⁶ pixels per one-hundred millisecond frame for seventeen minutes. We show that a set of spatial and temporal metrics can be used to build a random forest classifier which separates neural activity and artifact components automatically at human performance. We show how this data produces a functional segmentation of the neocortical sheet, providing a map of 230±14 domains from which extracted time courses maximally represent the underlying signal in each recording. This workflow of data-driven video decomposition and machine classification of signal sources will aid high quality mapping of complex cerebral dynamics.
<!-- figure 2D: What is the value of the mean neural component curve when the derivative approaches zero? Around 900 s or 15 min?
<!--
In order to understand how information flows across cerebral networks we need to be able to record neuronal activity across the cortical hemispheres from awake behaving animals. One method recently developed in mice uses calcium imaging data to access information flow.
Functional imaging of neuronal activity is important for understanding cerebral cortical dynamics.
Imaging neuronal calcium dynamics across the hemispheres is an important method for investigating the functional organization of cortical networks.
-->
<!--
@@ -37,12 +51,6 @@ This workflow of data-driven video decomposition and machine classification of s
Methods for analyzing cortical calcium imaging videos are underdeveloped and progress has been hindered by the challenges of live recordings, such as intermixing of artifact sources with neural signals. Here, we build a calcium flux video processing pipeline that uncovers the set of signals which optimally represent the underlying structure in observed neocortical imaging data.
We demonstrate the critical importance of choosing sufficient spatial and temporal sampling parameters for optimizing extraction of source signals and gaining the maximum information from a given recording, capturing and processing x samples
However, there are challenges with analyzing cortical calcium imaging videos, including optical artifacts and litte report on the samplif lack of u
(how to deal with undesired signal sources | optical artifacts) and (little work on optimizing the baseline sampling parameters that give best extraction of neuronal calcium source signals) | limited references) for time series extraction.
Here we present a data-driven workflow that isolates artifacts from calcium activity patterns, and segments independent functional units across the cortical surface using Independent Component Analysis (ICA).
ICA utilizes the statistical interdependence of pixel activation to completely unmix signals from background noise, given sufficient spatial and temporal samples.
@@ -52,41 +60,29 @@ We characterize each component using a set of extracted spatial and temporal met
We demonstrate that the performance of the machine classifier matches human identification of signal components in novel data sets.
We also analyze and compare control data recorded in a glial cell reporter and non-fluorescent mouse lines that validates human and machine identification of functional component class.
These improved techniques for data pre-processing, spatial segmentation, and time series extraction result in optimal signals for further analysis and model development.
-->
<!-- These improved techniques for data pre-processing, spatial segmentation, and time series extraction result in optimal signals for further analysis and model development. -->
<!--
In order to understand how information flows across cerebral networks we need to be able to record neuronal activity across the cortical hemispheres from awake behaving animals.
One method recently developed in mice uses calcium imaging data to access information flow.
Functional imaging of neuronal activity is important for understanding cerebral cortical dynamics.
Imaging neuronal calcium dynamics across the hemispheres is an important method for investigating the functional organization of cortical networks.
functional organization
investigate
important method
calcium dynamics
cortical networks
-->
## Introduction
<!-- Capturing maps of functional tissue organization is a key task -->
Optical techniques have long been used to monitor the functional dynamics in sets of neuronal elements ranging from isolated invertebrate nerve fibers[^Cohen:1968][^Salzberg:1977] to entire regions of mammalian visual cortex in vivo[^Grinvald:1986][^Grinvald:2004][^Ackman:2012]. Imaging calcium flux with calcium sensors[^Tsien:1989][^Chen:2013a] allows neural activity monitoring across the entire neocortex with high enough spatiotemporal resolution to identify sub-areal networks of the neocortex[^Ackman2014c][^Vanni2014]. These techniques have the potential to map supracellular group function at unprecedented resolution and scale across the neocortical sheet in awake behaving mice; however identifying neural signals from calcium imaging sessions is challenging due to numerous confounding signal sources.
Optical techniques have long been used to monitor the functional dynamics in sets of neuronal elements ranging from isolated invertebrate nerve fibers[^Cohen:1968][^Salzberg:1977] to entire regions of mammalian visual cortex in vivo[^Grinvald:1986][^Grinvald:2004][^Ackman:2012]. Imaging of calcium flux with calcium sensors[^Tsien:1989][^Chen:2013a] allows for transcranial neural activity monitoring across the cortical surface of mouse with high enough spatiotemporal resolution to identify sub-areal networks of the neocortex[^Vanni2014][^Ackman2014c]. These techniques have the potential to map supracellular group function at unprecedented resolution and scale across the neocortical sheet in awake behaving mice; however identifying neural signals from calcium imaging sessions is challenging due to numerous confounding signal sources.
<!-- Understanding cerebral dynamics at multiple scales is important for exploring how environmental and genetic influences give rise to altered neural connectivity patterns linked to behavioral phenotypes [^Ma:2016][^Kozberg:2016] -->
Wide-field imaging of neuronal calcium flux offers mesoscale observation of cortical neural dynamics and allows for viewing the supracellular group organization between microscale (cell) and macroscale (tissue lobe) investigations; however, it is affected by issues common to optical imaging recordings. Body or facial movements can create large fluctuations in autofluorescence of the brain and blood vessels, which produce significant artifacts in the data. Vascular artifacts are commonly seen due to vasodynamics and the resulting changes in blood flow to meet the energy demands of surrounding tissue. Fluid exchange between vascular and neural tissue causes cortical hemodynamics, resulting in region specific changes of optical properties among cerebral lobes[^Ma:2016]. Further, though the skull is fixed to a specific location during the experiment, slight brain movements occur within the cranium, thereby influencing the recordings. Any optical property differences that originate from the experimental preparation may be highlighted in the dataset as signal due to changes in tissue contrast.
Wide-field cortical calcium imaging provides a unique combination of spatially and temporally resolved dynamics across the cortical surface, with scale ranging from complex activation patterns in high-order circuits, to discrete activations hundreds of micrometers in diameter, to whole cortical lobe activity patterns[^Vanni2014][^Ackman2014c]. However, it is affected by issues common to optical imaging recordings. Body or facial movements can create large fluctuations in autofluorescence of the brain and blood vessels, which produce significant artifacts in the data. Vascular artifacts are commonly seen due to vasodynamics and the resulting changes in blood flow to meet the energy demands of surrounding tissue. Fluid exchange between vascular and neural tissue causes cortical hemodynamics, resulting in region specific changes of optical properties among cerebral lobes[^Ma:2016]. Further, though the skull is fixed to a specific location during the experiment, slight brain movements occur within the cranium, thereby influencing the recordings. Any optical property differences that originate from the experimental preparation may be highlighted in the dataset as signal due to changes in tissue contrast.
Eigendecompositions can be used to identify and filter components of signal[^Kozberg:2016][^Patel2015][^Pnevmatikakis2016], and present a flexible method of filtering that is not hardware dependent, and can be applied to any video dataset regardless of the recording hardware or parameters. Independent Component Analysis (ICA)[^Hyvarinen:2000] has been previously applied to fMRI and EEG data with varying success; for example, identifying both intrinsic connectivity networks rather than individual areas, and artifacts that represent large-scale effects rather than spatially localized effects[^Mckeown1998][^Pruim2015][^Parkes2018][^Beckmann2004]. We hypothesize that this is due to the lower density of spatial sampling in fMRI and EEG data. Wide-field calcium imaging provides a unique combination of spatially and temporally resolved dynamics across the cortical surface, with scale ranging from complex activation patterns in high-order circuits, to discrete activations hundreds of micrometers in diameter, to whole cortical lobe activity patterns[^Ackman2014c][^Vanni2014]. Researchers have recorded wide field calcium dynamics at frame rates ranging from 5-100Hz[^Ackman2014c][^Murphy2016][^Valley2020]. In addition, spatial resolution varies between different researchers’ setups, but is typically in the range of 256x256 to 512x512 pixels (0.06 to 0.2 megapixels) for the entire cortical surface, and is often further spatially reduced for processing[^Ackman2014c][^Murphy2016][^Allen2017]. Selection of resolution is often dependent on the video observer’s perceived quality of the data or available computational resources, rather than a quantified comparison of signal content.
Researchers have recorded wide field calcium dynamics at frame rates ranging from 5-100Hz[^Ackman2014c][^Murphy2016][^Valley2020]. In addition, spatial resolution varies between different researchers’ setups, but is typically in the range of 256x256 to 512x512 pixels (0.06 to 0.2 megapixels) for the entire cortical surface, and is often further spatially reduced for processing[^Ackman2014c][^Murphy2016][^Allen2017]. Selection of resolution is often dependent on the video observer’s perceived quality of the data or available computational resources, rather than a quantified comparison of signal content.
It is common to use sensory stimulation to identify specific regions in the neocortex and align a reference map based on the location of these defined regions[^Allen2017][^Vanni2017][^Clancy2019]. Even if these maps are reliable for locating primary sensory areas, they often lack specificity for higher order areas, or even completely lack sub-regional divisions. This is especially true in areas with a high degree of interconnectedness and overlapping functionality, such as motor cortex[^Mountcastle:1997]. Moreover, there is evidence that the shape and location of higher order regions can vary from subject to subject[^Zhuang:2017][^Glasser2016]. Improper map alignment or misinformed regional boundaries can lead to a loss in dynamic range between signals across a regional border. Thus, to extract the most information from a recorded dataset, the level of parcellation must reflect the quality and sources present within the data. Thus, a flexible data-driven method is necessary and must also respect functional boundaries of the cortex and be sensitive to age, genotype and individual variation.
It is common to use sensory stimulation to identify specific regions in the neocortex and align a reference map based on the location of these defined regions[^Allen2017][^Vanni2017][^Clancy2019]. Even if these maps are reliable for locating primary sensory areas, they often lack specificity for higher order areas, or even completely lack sub-regional divisions. This is especially true in areas with a high degree of interconnectedness and overlapping functionality, such as motor cortex[^Mountcastle:1997]. Moreover, there is evidence that the shape and location of higher order regions can vary from subject to subject[^Zhuang:2017][^Glasser2016]. Improper map alignment or misinformed regional boundaries can lead to a loss in dynamic range between signals across a regional border. Thus, to extract the most information from a recorded dataset, the level of parcellation must reflect the quality and sources present within the data. Thus, a flexible data-driven method is necessary and must also respect functional boundaries of the cortex and be sensitive to age, genotype and individual variation.
Eigendecompositions can be used to identify and filter components of signal[^Kozberg:2016][^Patel2015][^Pnevmatikakis2016], and present a flexible method of filtering that is not hardware dependent, and can be applied to any video dataset regardless of the recording hardware or parameters. Independent Component Analysis (ICA)[^Hyvarinen:2000] has been previously applied to fMRI and EEG data with varying success; for example, identifying both intrinsic connectivity networks rather than individual areas, and artifacts that represent large-scale effects rather than spatially localized effects[^Mckeown1998][^Pruim2015][^Parkes2018][^Beckmann2004]. We hypothesize that this is due to the lower density of spatial sampling in fMRI and EEG data.
Here we present an ICA-based workflow that isolates and filters artifacts from calcium imaging videos, with principled exploration of each component to identify each signal source necessary to reduce the contamination resulting from these physiological dynamics. Independent component analysis (ICA) is a nonparametric unsupervised machine learning technique that can identify each signal source in densely sampled (5.5 million pixels per frame) calcium imaging videos based on their spatially co-activating pixels and temporal properties. The global mean time course was initially subtracted and stored, thereby allowing ICA to decompose each signal distinct from global effects. The decomposition results in hundreds of neural source components per hemisphere that are distinctly de-mixed from artifact source signals. Our concurrent analysis of control wide-field imaging data corroborates the identification of artifact signal sources and gives insight into the structure of neuronal calcium dynamics across neocortex.
@@ -95,7 +91,7 @@ Further, we explore the resolution-dependent effect of signal extraction on ICA
## Results
To record neural activity patterns in the cortex of awake behaving adult mice, we transcranially imaged fluorescence from a mouse that has the genetically encoded calcium indicator, GCaMP6s, expressed in all neurons under the control of the Snap25 promoter[^Madisen:2015]. We expose and illuminate the cranium with blue wavelength light and capture emitted green light with a sCMOS camera at high spatial resolution (2160x2560 pixels, 5.5 megapixels; ∼6.9 µm/pixel). To observe the spatiotemporal properties of the recorded activity patterns, we crop the video to only neural tissue, and compare the change in fluorescence over the mean fluorescence: ∆F/F over time (Fig. 1A-B). In order to identify eigenvectors associated with artifacts and hemodynamic responses, similar data was recorded and processed in three sets of age matched control mice: cx3cr1 GFP (microglia; mGFP), adhl1 GFP (astrocyte; aGFP), and the non-transgenic C57/black 6 (Bl6) mice.
To record neural activity patterns in the cortex of awake behaving mice, we transcranially imaged fluorescence from a mouse that has the genetically encoded calcium indicator, GCaMP6s, expressed in all neurons under the control of the Snap25 promoter[^Madisen:2015]. We expose and illuminate the cranium with blue wavelength light and capture emitted green light with a sCMOS camera at high spatial resolution (2160x2560 pixels, 5.5 megapixels; ∼6.9 µm/pixel). To observe the spatiotemporal properties of the recorded activity patterns, we crop the video to only neural tissue, and compare the change in fluorescence over the mean fluorescence: ∆F/F over time (Fig. 1A-B). In order to identify eigenvectors associated with artifacts and hemodynamic responses, similar data was recorded and processed in three sets of age matched control mice: cx3cr1 GFP (microglia; mGFP), adhl1 GFP (astrocyte; aGFP), and the non-transgenic C57/black 6 (Bl6) mice.
@@ -157,7 +153,7 @@ To test how ICA component separation is affected by spatiotemporal resolution an
</figcaption></figure>
Increasing the sampling rate above 10Hz showed little to no effect on the peak to peak distance (Fig. 2C; ∆p−p < 0.01), and a slight decrease in the autocorrelation of the primary peak (∆p1 = 0.03). However temporal downsampling below 10Hz resulted in a shifting of the signal and noise peaks (∆p1 = 0.06), and a reduction in the peak to peak distance (∆p−p = 0.02). This result agrees with previous analyses that found 10Hz to be the maximal sampling frequency required for measuring population calcium dynamics 20. Together, these findings suggest that the separation quality of our captured dynamics are highly sensitive to spatial resolution, and not as sensitive to temporal resolution. We considered collecting spatial samples higher than our current resolution of ∼6.9 µm/px, but computing decompositions on datasets this large would push the limits of available computing.
Increasing the sampling rate above 10Hz showed little to no effect on the peak to peak distance (Fig. 2C; ∆p−p < 0.01), and a slight decrease in the autocorrelation of the primary peak (∆p1 = 0.03). However temporal downsampling below 10Hz resulted in a shifting of the signal and noise peaks (∆p1 = 0.06), and a reduction in the peak to peak distance (∆p−p = 0.02). This result agrees with previous analyses that found 10Hz to be the maximal sampling frequency required for measuring population calcium dynamics[^Vanni2017]. Together, these findings suggest that the separation quality of our captured dynamics are highly sensitive to spatial resolution, and not as sensitive to temporal resolution. We considered collecting spatial samples higher than our current resolution of ∼6.9 µm/px, but computing decompositions on datasets this large would push the limits of available computing.
To determine the ideal duration of video collected, we calculated the number of significant signal and noise components for various video durations (Fig. 2D). We found that for ICA decompositions on activity patterns from a P21 mouse, the number of signal and artifact components leveled off after 20 minutes. Population analyses showed that this number was highly similar among P21 mice (n signal components: 244±25.7; n artifact components: 87.2±20.7; N=3 mice, 2 subsequent recordings each).
@@ -345,9 +341,9 @@ We additionally quantified whether detected regions were similar across map comp
## Discussion
Wide-field calcium imaging has grown in popularity in the last decade due to advances in genetically encoded calcium indicators, however, the methods used to isolate neural signal sources are underdeveloped[^Saxena2020]. Here we use an ICA-based algorithm that overcomes many of these limitationsEigendecomposition algorithms have been essential to understand signals across neuroscience. Another recent eigendecomposition pipeline has been developed to explore the functional activities across wide-field imaging of the cortex, but is limited by the use of a reference map and was not able to separate artifact signals from neural activations[^Saxena2020]. The methods presented here are able to achieve similar expository results with artifact removal, allowing researchers to explore datasets of any age, treatment, genotype, or strain that would be impeded by the use of a reference map.
Wide-field calcium imaging is increasingly used for mapping neural tissues, due to advances in genetic calcium indicators, transgenic animals, image sensors, and computational processing. However, the methods used to isolate neural signal sources are underdeveloped[^Saxena2020]. Here we use an ICA-based algorithm that overcomes many of these limitations. Eigendecomposition algorithms have been essential to understand signals across neuroscience. Another recent eigendecomposition pipeline has been developed to explore the functional activities across wide-field imaging of the cortex, but is limited by the use of a reference map and was not able to separate artifact signals from neural activations[^Saxena2020]. The methods presented here are able to achieve similar expository results with artifact removal, allowing researchers to explore datasets of any age, treatment, genotype, or strain that would be impeded by the use of a reference map.
High resolution imaging of mesoscale cortical calcium dynamics combined with data-driven decomposition using ICA results in an optimized extraction of neural source signals. We have built de novo anour own Independent Component Analysis (ICA) based pipeline to that can not onlynot only achieve (isolate?, identify? sub-regional neural components, but also can to show utilization of components to quantify the impact on data quality based on recording parameters, to improve data quality through removal of artifacts, and to build domain maps based on the limitations of the fluorescent signal sources. We demonstrate that these methods provide precise isolation and filtration of video artifacts due to movement, optical deformations, or blood vessel dynamics while recovering cortical source signals with minimal alteration. Our lab and another have successfully implemented an ICA-based filtration to isolate the neural signal from artifacts[^Lu2021][^West2021]. This approach can either be used alone, or in conjunction with techniques to correct calcium dynamics from tissue hemodynamics[^Ma:2016][^Valley2020].
Wide-field imaging of neuronal calcium flux offers mesoscale observation of cortical neural dynamics and allows for viewing the supracellular group organization between microscale (cell) and macroscale (tissue lobe) investigations. High resolution imaging of mesoscale cortical calcium dynamics combined with data-driven decomposition using ICA results in an optimized extraction of neural source signals. We have built de novo anour own Independent Component Analysis (ICA) based pipeline to that can not onlynot only achieve (isolate?, identify? sub-regional neural components, but also can to show utilization of components to quantify the impact on data quality based on recording parameters, to improve data quality through removal of artifacts, and to build domain maps based on the limitations of the fluorescent signal sources. We demonstrate that these methods provide precise isolation and filtration of video artifacts due to movement, optical deformations, or blood vessel dynamics while recovering cortical source signals with minimal alteration. Our lab and another have successfully implemented an ICA-based filtration to isolate the neural signal from artifacts[^Lu2021][^West2021]. This approach can either be used alone, or in conjunction with techniques to correct calcium dynamics from tissue hemodynamics[^Ma:2016][^Valley2020].
Signal separation from mesoscale calcium dynamics recorded across the cortical surface was the most complete at the highest spatial resolution tested (pixel size of 6.9 µm/px). Our recordings consisted of large sets of densely sampled image frames having at least 12 bits of dynamic range across pixel intensities. Temporal resolution had less of an effect on ICA signal separation; we found that a 10Hz sampling rate was sufficient for spatial segregation. These metrics for signal quality are automatically generated by our algorithm, and can be used to optimize signal collection on any given experimental setup. The number of components identified is highly stable after recording sufficient duration of dynamics, and provides a metric for spatial complexity of neural signals across the neocortex. Compared with the high density optical recordings we used here, other neurophysiological techniques remain limited in the number of available spatial samples; as such, the effect of signal recording resolution on ICA decomposition of neural signal sources had not previously been reported.