Limits...
Puzzle Imaging: Using Large-Scale Dimensionality Reduction Algorithms for Localization.

Glaser JI, Zamft BM, Church GM, Kording KP - PLoS ONE (2015)

Bottom Line: This technique takes many spatially disordered samples, and then pieces them back together using local properties embedded within the sample.We demonstrate the theoretical capabilities of puzzle imaging in three biological scenarios, showing that (1) relatively precise 3-dimensional brain imaging is possible; (2) the physical structure of a neural network can often be recovered based only on the neural connectivity matrix; and (3) a chemical map could be reproduced using bacteria with chemosensitive DNA and conjugative transfer.The ability to reconstruct scrambled images promises to enable imaging based on DNA sequencing of homogenized tissue samples.

View Article: PubMed Central - PubMed

Affiliation: Department of Physical Medicine and Rehabilitation, Northwestern University and Rehabilitation Institute of Chicago, Chicago, Illinois, United States of America.

ABSTRACT
Current high-resolution imaging techniques require an intact sample that preserves spatial relationships. We here present a novel approach, "puzzle imaging," that allows imaging a spatially scrambled sample. This technique takes many spatially disordered samples, and then pieces them back together using local properties embedded within the sample. We show that puzzle imaging can efficiently produce high-resolution images using dimensionality reduction algorithms. We demonstrate the theoretical capabilities of puzzle imaging in three biological scenarios, showing that (1) relatively precise 3-dimensional brain imaging is possible; (2) the physical structure of a neural network can often be recovered based only on the neural connectivity matrix; and (3) a chemical map could be reproduced using bacteria with chemosensitive DNA and conjugative transfer. The ability to reconstruct scrambled images promises to enable imaging based on DNA sequencing of homogenized tissue samples.

No MeSH data available.


Neural Connectomics Puzzling Performance.(A) On the left, an example reconstruction of neural locations based on a simulation of connections of pyramidal cells in layer 3 of cortex. Colors are based on initial locations: those with a larger initial x location are more red, while those with a larger initial y location are more blue. The distance errors are calculated following scaling and rotating the reconstructed volume to match the original volume. In the middle, a 2-dimensional slice through reconstructed volume. On the right, one metric for the accuracy of reconstruction is shown by plotting the reconstructed distances between all points against their true distances for the reconstruction in this panel. The mean plus/minus the standard deviation (shaded) is shown. A perfect reconstruction would be a straight line, corresponding to an R value of 1. (B) Same as panel A, except based on an example simulation of pyramidal cells in both layers 2 and 3 of cortex. (C) Boxplots of R values for layer 3 simulations, and layer 2/3 simulations. The 5 lines (from bottom to top) correspond to 5%, 25%, 50%, 75%, and 95% quantiles of R values across simulations. (D) Boxplots (as in panel C) of mean errors across simulations. (E) The probability of connection as a function of distance between pyramidal cells, which is used in the simulations of the other panels [17]. (F) Using the parameters of the connectivity probability distribution of layer 3, the baseline connection probability (the probability of connection at a distance of 0) of the connectivity distribution is changed. R values and mean errors are shown as a function of this baseline probability. Error bars represent the standard deviation across simulations in this and the next panel. (G) Using the parameters of the connectivity probability distribution of layer 3, the standard deviation of the connectivity distribution is changed. R values and mean errors are shown as a function of the standard deviation.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4507868&req=5

pone.0131593.g005: Neural Connectomics Puzzling Performance.(A) On the left, an example reconstruction of neural locations based on a simulation of connections of pyramidal cells in layer 3 of cortex. Colors are based on initial locations: those with a larger initial x location are more red, while those with a larger initial y location are more blue. The distance errors are calculated following scaling and rotating the reconstructed volume to match the original volume. In the middle, a 2-dimensional slice through reconstructed volume. On the right, one metric for the accuracy of reconstruction is shown by plotting the reconstructed distances between all points against their true distances for the reconstruction in this panel. The mean plus/minus the standard deviation (shaded) is shown. A perfect reconstruction would be a straight line, corresponding to an R value of 1. (B) Same as panel A, except based on an example simulation of pyramidal cells in both layers 2 and 3 of cortex. (C) Boxplots of R values for layer 3 simulations, and layer 2/3 simulations. The 5 lines (from bottom to top) correspond to 5%, 25%, 50%, 75%, and 95% quantiles of R values across simulations. (D) Boxplots (as in panel C) of mean errors across simulations. (E) The probability of connection as a function of distance between pyramidal cells, which is used in the simulations of the other panels [17]. (F) Using the parameters of the connectivity probability distribution of layer 3, the baseline connection probability (the probability of connection at a distance of 0) of the connectivity distribution is changed. R values and mean errors are shown as a function of this baseline probability. Error bars represent the standard deviation across simulations in this and the next panel. (G) Using the parameters of the connectivity probability distribution of layer 3, the standard deviation of the connectivity distribution is changed. R values and mean errors are shown as a function of the standard deviation.

Mentions: Performance. We tested the ability of connectomics puzzling to determine the locations of a simulated network of neurons. Hellwig [17] described the probability of connections as a function of distance between pyramidal cells in layers 2 and 3 of the rat visual cortex. We simulated 8000 neurons in a 400 μm edge-length cube (so they were on average spaced ∼ 20 μm apart from each other in a given direction). Connections between neurons were randomly determined based on their distance using the previously mentioned probability function (Fig 5E).


Puzzle Imaging: Using Large-Scale Dimensionality Reduction Algorithms for Localization.

Glaser JI, Zamft BM, Church GM, Kording KP - PLoS ONE (2015)

Neural Connectomics Puzzling Performance.(A) On the left, an example reconstruction of neural locations based on a simulation of connections of pyramidal cells in layer 3 of cortex. Colors are based on initial locations: those with a larger initial x location are more red, while those with a larger initial y location are more blue. The distance errors are calculated following scaling and rotating the reconstructed volume to match the original volume. In the middle, a 2-dimensional slice through reconstructed volume. On the right, one metric for the accuracy of reconstruction is shown by plotting the reconstructed distances between all points against their true distances for the reconstruction in this panel. The mean plus/minus the standard deviation (shaded) is shown. A perfect reconstruction would be a straight line, corresponding to an R value of 1. (B) Same as panel A, except based on an example simulation of pyramidal cells in both layers 2 and 3 of cortex. (C) Boxplots of R values for layer 3 simulations, and layer 2/3 simulations. The 5 lines (from bottom to top) correspond to 5%, 25%, 50%, 75%, and 95% quantiles of R values across simulations. (D) Boxplots (as in panel C) of mean errors across simulations. (E) The probability of connection as a function of distance between pyramidal cells, which is used in the simulations of the other panels [17]. (F) Using the parameters of the connectivity probability distribution of layer 3, the baseline connection probability (the probability of connection at a distance of 0) of the connectivity distribution is changed. R values and mean errors are shown as a function of this baseline probability. Error bars represent the standard deviation across simulations in this and the next panel. (G) Using the parameters of the connectivity probability distribution of layer 3, the standard deviation of the connectivity distribution is changed. R values and mean errors are shown as a function of the standard deviation.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4507868&req=5

pone.0131593.g005: Neural Connectomics Puzzling Performance.(A) On the left, an example reconstruction of neural locations based on a simulation of connections of pyramidal cells in layer 3 of cortex. Colors are based on initial locations: those with a larger initial x location are more red, while those with a larger initial y location are more blue. The distance errors are calculated following scaling and rotating the reconstructed volume to match the original volume. In the middle, a 2-dimensional slice through reconstructed volume. On the right, one metric for the accuracy of reconstruction is shown by plotting the reconstructed distances between all points against their true distances for the reconstruction in this panel. The mean plus/minus the standard deviation (shaded) is shown. A perfect reconstruction would be a straight line, corresponding to an R value of 1. (B) Same as panel A, except based on an example simulation of pyramidal cells in both layers 2 and 3 of cortex. (C) Boxplots of R values for layer 3 simulations, and layer 2/3 simulations. The 5 lines (from bottom to top) correspond to 5%, 25%, 50%, 75%, and 95% quantiles of R values across simulations. (D) Boxplots (as in panel C) of mean errors across simulations. (E) The probability of connection as a function of distance between pyramidal cells, which is used in the simulations of the other panels [17]. (F) Using the parameters of the connectivity probability distribution of layer 3, the baseline connection probability (the probability of connection at a distance of 0) of the connectivity distribution is changed. R values and mean errors are shown as a function of this baseline probability. Error bars represent the standard deviation across simulations in this and the next panel. (G) Using the parameters of the connectivity probability distribution of layer 3, the standard deviation of the connectivity distribution is changed. R values and mean errors are shown as a function of the standard deviation.
Mentions: Performance. We tested the ability of connectomics puzzling to determine the locations of a simulated network of neurons. Hellwig [17] described the probability of connections as a function of distance between pyramidal cells in layers 2 and 3 of the rat visual cortex. We simulated 8000 neurons in a 400 μm edge-length cube (so they were on average spaced ∼ 20 μm apart from each other in a given direction). Connections between neurons were randomly determined based on their distance using the previously mentioned probability function (Fig 5E).

Bottom Line: This technique takes many spatially disordered samples, and then pieces them back together using local properties embedded within the sample.We demonstrate the theoretical capabilities of puzzle imaging in three biological scenarios, showing that (1) relatively precise 3-dimensional brain imaging is possible; (2) the physical structure of a neural network can often be recovered based only on the neural connectivity matrix; and (3) a chemical map could be reproduced using bacteria with chemosensitive DNA and conjugative transfer.The ability to reconstruct scrambled images promises to enable imaging based on DNA sequencing of homogenized tissue samples.

View Article: PubMed Central - PubMed

Affiliation: Department of Physical Medicine and Rehabilitation, Northwestern University and Rehabilitation Institute of Chicago, Chicago, Illinois, United States of America.

ABSTRACT
Current high-resolution imaging techniques require an intact sample that preserves spatial relationships. We here present a novel approach, "puzzle imaging," that allows imaging a spatially scrambled sample. This technique takes many spatially disordered samples, and then pieces them back together using local properties embedded within the sample. We show that puzzle imaging can efficiently produce high-resolution images using dimensionality reduction algorithms. We demonstrate the theoretical capabilities of puzzle imaging in three biological scenarios, showing that (1) relatively precise 3-dimensional brain imaging is possible; (2) the physical structure of a neural network can often be recovered based only on the neural connectivity matrix; and (3) a chemical map could be reproduced using bacteria with chemosensitive DNA and conjugative transfer. The ability to reconstruct scrambled images promises to enable imaging based on DNA sequencing of homogenized tissue samples.

No MeSH data available.