diff --git a/manuscript/.gitignore b/manuscript/.gitignore
index d26d511e8211d9b88b7be4fb3bcea56fae68d49c..7412014a6324100ee2e628b9cb9b9134a2de0e73 100644
--- a/manuscript/.gitignore
+++ b/manuscript/.gitignore
@@ -7,3 +7,5 @@ archive*/*
*.gz
*.log
*.out
+*.blg
+*.bbl
diff --git a/manuscript/RSF.bbl b/manuscript/RSF.bbl
deleted file mode 100644
index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..0000000000000000000000000000000000000000
diff --git a/manuscript/RSF.pdf b/manuscript/RSF.pdf
index 6f0fb54fb6b8a5420361b042f51330909d3b5e1a..cc15f3237739144d243c67569d15bc076b3c1751 100644
Binary files a/manuscript/RSF.pdf and b/manuscript/RSF.pdf differ
diff --git a/manuscript/RSF.tex b/manuscript/RSF.tex
index 6590dfe44e28cd8cc5d1a9db0b0d1315a543cba3..181c4e351c9dacba6acacb61c765388b4eb79788 100644
--- a/manuscript/RSF.tex
+++ b/manuscript/RSF.tex
@@ -17,7 +17,7 @@
\usepackage{subcaption}
\usepackage[]{soul}
\usepackage[bibstyle=nature,citestyle=nature]{biblatex}
-% \addbibresource{ssfClustering.bib}
+\addbibresource{rsf.bib}
\hypersetup{
colorlinks=true,
linkcolor={blue},
@@ -83,40 +83,34 @@
\end{list}
-\section{The retinal structure function}
-
-We build on these technologies:
-
-0. \href{https://pure.uva.nl/ws/files/4346395/68110\_297537.pdf}{Normalized information distance (NID)}.
+\section{Introduction}
-and
+\subsection{Why metric embedding?}
+tl;dr RKHS make subsequent optimization learning easier to implement and more likely to generalize better [primer].
-\href{https://doi.org/10.1109/tpami.2023.3264690}{Kolmogorov structure functions}
-1. Semi-supervised spectral learning as in \href{https://dl.acm.org/doi/10.5555/1630659.1630742}{Kamvar et al., 2003} [Cohen, et al. 2008]. Key advantage of RKHS [RKHS primer].
+\subsection{Optical Coherence Tomography of the Retina}
+[Ronald?] -- ~ 4 paragraphs summarize impact of retinal OCT (Glaucoma, occulomics, etc.), background on VFMD, background on metadata, imaging physics.
+Key recent deep learning papers [Schuman et al.]. Other learning approaches [NNMF et al.].
-2. \href{https://doi.org/10.1093/bioinformatics/btz523}{Hydra Image Processor (HIP)} : The HIP provides 3-D multi-scale structure enhancing filters applied here for retinal OCT stacks []. The HIP allows an efficient plate-enhancement based on 3-D LoG. The HIP also gives us a 3-D non-local means denoising filter. See figure 1. Preliminary experiments with a HIP plate-enhancing non-local means via anisotropic kernel is a promising area for future study. See figure 2.
+Key observation -- inherent anisotropy of imaging implements Frangi-like plate filter along the high resolution axis.
-observation: Inherent anisotropy of retinal OCT implements optically a plate-enhancing filter, as in \href{https://dspace.library.uu.nl/bitstream/1874/377/18/c2.pdf}{Frangi }.
+\subsection{Kolmogorov complexity and the normalized information distance (NID)}
+\subsection{Semi-supervised spectral learning}
+\subsection{Kolmogorov structure function vs. multi-scale vessel ehancing image filters}
+$H_{Kolmogorov}$ [Vereschegin and Vitanyi] \emph{vs.} $H_{Frangi}$
-3. Free lossless image format (FLIF) 3.3D lossless image compression []. A maniac entropy encoder represents spatial voxel patterns using decision trees. Images that are more similar provide more improvement when compressed together than would images that are less similar. After we encode \emph{a priori} shape knowledge via the structure function radii, the FLIF compression approximates relative Kolmogorov complexity between image pairs to measure spatiotemporal patterns of change.
+\section{The retinal structure function}
-\section{Using $NCD(OCT_1,OCT_2)$ to predict $|VFMD_1 -VFMD_2|$}
-key results mostly done.
+\section{Validating the NCD vs. prediction against functional changes of the visual field }
-open source code in progress (?).
+Idea : $NCD(OCT_1,OCT_2)$ to predict $|VFMD_1 -VFMD_2|$
-open source image data? what can we release? maybe a small sample, e.g. $\sim 100$ stacks?
+open source code in progress (?). open source image data? what can we release? maybe a small sample, e.g. $\sim 100$ stacks?
Deep learning state of the art in estimating VFMD from retinal OCT achieves RMSE $\lessapprox 2.5$. We pose the question differently, looking instead to predict
$|\Delta VFMD|$ given the normalized compression distance between image pairs. Training a single-layer single-node regression network with sigmoid activation yields cross-validation RMSE $\sim 0.8$. A second single-layer 10-node regression network with input including one of the two VFMD values (from either image date) achieves cross-validation RMSE $\sim 0.7$.
-\section{Why metric embedding?}
-tl;dr RKHS make subsequent optimization learning easier to implement and more likely to generalize better [primer].
-
-
-\section{Retinal OCT and VFMD}
-related literature -- background on metadata, imaging physics. Key recent deep learning papers [Schuman et al.]. Other learning approaches [NNMF].
\section{Future work}
@@ -126,14 +120,28 @@ Clustering by compression: e.g. M vs. F, OD vs. OS, zip code?
RKHS combinations: macula + ONH simultaneously, multi-resolution by channel vs. dimensionality reduction, metadata w/ bzip,...?
+\section{Acknowledgements}
+
+
+\section{Software and data availability}
+All of the software tools used are available free and open source, see \url{https://git-bioimage.coe.drexel.edu/opensource/rsf}.
+The image data together with segmentation and tracking results can be viewed interactively at \url{https://leverjs.net/ssfCluster}. The LEVERSC 4-D WEBGL viewer \cite[]{Cohen2022} renders 3-D kymographs and images, and the web API also supports downloading metadata and results directly.
+\label{Sect.Software}
+
+\printbibliography
+
+\newpage
\begin{figure*}[!ht]
{\includegraphics[page=1,width=1.0\textwidth]{figureX.pdf}}
\centering
- \caption[]{\textbf{The retinal structure function (RSF) computes at each image voxel how ‘plate-like’ is the 3-D intensity configuration of the surrounding voxels}. The input image (a, single slice shown) is denoised using non-local means (c,d), and then processed with a Laplacian of Gaussian (LoG) convolution kernel (b, single slice shown). The LoG positive response (red) represents dark structure against a bright background, the negative response (green) represents bright structure against a dark background. The 2-D regional maxima weighted by the LoG, with the interior of each dot scaled by filter response, are shown for illustration (e). The 3-D regional maxima weighted by the LoG (f), are the representation used as input to the normalized compression distance (NCD). Top row (a-d) shows optic nerve OCT using LoG of size $[0.5,10,10]$ voxels, the bottom row (e-h, same captions as a-d) is macula OCT with LoG of size $[0.35,7,7]$ voxels. The goal of the RSF is to extract a local measure of the voxel structure so the NCD best captures patterns of similarity between image pairs. See also supplementary movie (\href[]{https://bioimage.coe.drexel.edu/media/andy/regional_max_pt5_20.mp4}{1}) and
- (\href[]{https://bioimage.coe.drexel.edu/media/andy/rsf_pt5_pt35_P10053_Macular Cube 200x200_6-17-2019_13-16-17_OD_sn1677_cube_raw.mp4}{2}).
+ \caption[]{\textbf{The retinal structure function (RSF) computes at each image voxel how ‘plate-like’ is the 3-D intensity configuration of the surrounding voxels}. The input image (a, single slice shown) is denoised using non-local means (c,d), and then processed with a Laplacian of Gaussian (LoG) convolution kernel (b, single slice shown). The LoG positive response (red) represents dark structure against a bright background, the negative response (green) represents bright structure against a dark background. The 2-D regional maxima weighted by the LoG, with the interior of each dot scaled by filter response, are shown for illustration (e). The 3-D regional maxima weighted by the LoG (f), are the representation used as input to the normalized compression distance (NCD). Top row (a-d) shows optic nerve OCT using LoG of size $[0.5,10,10]$ voxels, the bottom row (e-h, same captions as a-d) is macula OCT with LoG of size $[0.35,7,7]$ voxels. The goal of the RSF is to extract a local measure of the voxel structure so the NCD best captures patterns of similarity between image pairs. See also
+ \href{https://bioimage.coe.drexel.edu/media/andy/regional_max_pt5_20.mp4}{\textbf{ONH movie}} and
+ \href{https://bioimage.coe.drexel.edu/media/andy/rsf_pt5_pt35_P10053_Macular\%20Cube\%20200x200_6-17-2019_13-16-17_OD_sn1677_cube_raw.mp4}{\textbf{macula movie}}. TODO brighten macula movie...
+
+ }
- }
+
\label{fig:figure1}
\end{figure*}
diff --git a/manuscript/rsf.bib b/manuscript/rsf.bib
new file mode 100644
index 0000000000000000000000000000000000000000..81ccdc002d1309c6fcc76ca27adfbc6a0e6fb42a
--- /dev/null
+++ b/manuscript/rsf.bib
@@ -0,0 +1,240 @@
+@article{Cohen2023,
+ title = {The cluster structure function},
+ author = {Andrew R. Cohen and Paul M. Vitányi},
+ journal = {{IEEE} Transactions on Pattern Analysis and Machine Intelligence},
+ publisher = {Institute of Electrical and Electronics Engineers ({IEEE})},
+ year = {2023},
+ pages = {1-13},
+ doi = {10.1109/tpami.2023.3264690}
+}
+@article{Cohen2023,
+ author = {Maska and M. and Ulman and V. and Delgado-Rodriguez and P. and Gomez-de-Mariscal and E. and Necasova and T. and Pena, Guerrero and A., F. and Ren and I., T. and Meyerowitz and M., E. and Scherr and T. and Loffler and K. and Mikut and R. and Guo and T. and Wang and Y. and Allebach and P., J. and Bao and R. and Al-Shakarji and M., N. and Rahmon and G. and Toubal and E., I. and Palaniappan and K. and Lux and F. and Matula and P. and Sugawara and K. and Magnusson and G., K. E. and Aho and L. and Cohen and R., A. and Arbelle and A. and Ben-Haim and T. and Raviv and R., T. and Isensee and F. and Jager and F., P. and Maier-Hein and H., K. and Zhu and Y. and Ederra and C. and Urbiola and A. and Meijering and E. and Cunha and A. and Munoz-Barrutia and A. and Kozubek and M. and Ortiz-de-Solorzano and C.},
+ title = {The Cell Tracking Challenge: 10 years of objective benchmarking},
+ journal = {Nat Methods},
+ ISSN = {1548-7105 (Electronic)
+1548-7091 (Linking)},
+ DOI = {10.1038/s41592-023-01879-y},
+ url = {https://www.ncbi.nlm.nih.gov/pubmed/37202537},
+ year = {2023},
+ type = {Journal Article}
+}
+@article{Cohen2022,
+ author = {Winter and Mark and Cohen and R., Andrew},
+ title = {LEVERSC: Cross-Platform Scriptable Multichannel 3-D Visualization for Fluorescence Microscopy Images},
+ journal = {Frontiers in Bioinformatics},
+ volume = {2},
+ ISSN = {2673-7647},
+ DOI = {10.3389/fbinf.2022.740078},
+ url = {https://www.frontiersin.org/article/10.3389/fbinf.2022.740078},
+ year = {2022},
+ type = {Journal Article}
+}
+@article{Cohen2019,
+ author = {Wait and E. and Winter and M. and Cohen and R., A.},
+ title = {Hydra image processor: 5-D GPU image analysis library with MATLAB and python wrappers},
+ journal = {Bioinformatics},
+ volume = {35},
+ number = {24},
+ pages = {5393-5395},
+ ISSN = {1367-4811 (Electronic)
+1367-4803 (Linking)},
+ DOI = {10.1093/bioinformatics/btz523},
+ url = {http://www.ncbi.nlm.nih.gov/pubmed/31240306},
+ year = {2019},
+ type = {Journal Article}
+}
+@article{Cohen2018,
+ author = {Winter, M. and Mankowski, W. and Wait, E. and Hoz, E. Cardenas De La and Aguinaldo, A. and Cohen, and A. R.},
+ title = {Separating Touching Cells using Pixel Replicated Elliptical Shape Models},
+ journal = {IEEE Trans Medical Imaging},
+ volume = {38},
+ number = {4},
+ pages = {883-893},
+ year = {2018},
+ type = {Journal Article}
+}
+@article{Cohen2017,
+ author = {Apostolopoulou and Maria and Kiehl and R., Thomas and Winter and Mark and Hoz, Cardenas De La and Edgar and Boles and C., Nathan and Bjornsson and S., Christopher and Zuloaga and L., Kristen and Goderie and K., Susan and Wang and Yue and Cohen and R., Andrew and Temple and Sally},
+ title = {Non-monotonic Changes in Progenitor Cell Behavior and Gene Expression during Aging of the Adult V-SVZ Neural Stem Cell Niche},
+ journal = {Stem Cell Reports},
+ volume = {9},
+ number = {6},
+ pages = {1931-1947},
+ ISSN = {2213-6711},
+ DOI = {https://doi.org/10.1016/j.stemcr.2017.10.005},
+ url = {http://www.sciencedirect.com/science/article/pii/S2213671117304605},
+ year = {2017},
+ type = {Journal Article}
+}
+@article{Cohen2015_NCDM,
+ author = {Cohen, A. R. and Vitanyi, P. M. B.},
+ title = {Normalized Compression Distance of Multisets with Applications},
+ journal = {Pattern Analysis and Machine Intelligence, IEEE Transactions on},
+ volume = {37},
+ number = {8},
+ pages = {1602-1614},
+ ISSN = {0162-8828},
+ DOI = {10.1109/TPAMI.2014.2375175},
+ year = {2015},
+ type = {Journal Article}
+}
+@article{Cohen2015_StemCell,
+ author = {Winter and Mark R and Liu and Mo and Monteleone and David and Melunis and Justin and Hershberg and Uri and Goderie and Susan K and Temple and Sally and Cohen, A. R.},
+ title = {Computational Image Analysis Reveals Intrinsic Multigenerational Differences between Anterior and Posterior Cerebral Cortex Neural Progenitor Cells},
+ journal = {Stem Cell Reports},
+ volume = {5},
+ number = {4},
+ pages = {609-620},
+ ISSN = {2213-6711},
+ url = {http://www.sciencedirect.com/science/article/pii/S2213671115002362},
+ year = {2015},
+ type = {Journal Article}
+}
+@article{Cohen2014,
+ author = {Chenouard and N. and Smal and I. and Chaumont and F. and Maška and M. and Sbalzarini and F., I. and Gong and Y. and Cardinale and J. and Carthel and C. and Coraluppi and S. and Winter and M. and Cohen and R., A. and Godinez and J., W. and Rohr and K. and Kalaidzidis and Y. and Liang and L. and Duncan and J. and Shen and H. and Xu and Y. and Magnusson and G., K. E. and Jaldén and J. and Blau and M., H. and Paul-Gilloteaux and P. and Roudot and P. and Kervrann and C. and Waharte and F. and Tinevez and Y., J. -. and Shorte and L., S. and Willemse and J. and Celler and K. and Wezel and P., G. and Dan and W., H. -. and Tsai and S., Y. -. and Solórzano, Ortiz de and C. and Olivo-Marin and C., J. -. and Meijering and E.},
+ title = {Objective comparison of particle tracking methods},
+ journal = {Nat. Methods},
+ volume = {11},
+ DOI = {10.1038/nmeth.2808},
+ url = {http://dx.doi.org/10.1038/nmeth.2808},
+ year = {2014},
+ type = {Journal Article}
+}
+@article{Cohen2010,
+ author = {Cohen, A. R. and Gomes and Francisco and Roysam and Badrinath and Cayouette, Michel},
+ title = {Computational prediction of neural progenitor cell fates},
+ journal = {Nat Methods},
+ volume = {7},
+ number = {3},
+ pages = {213 - 218},
+ year = {2010},
+ type = {Journal Article}
+}
+@article{Cohen2009,
+ title = {Automatic Summarization of Changes in Biological Image Sequences Using Algorithmic Information Theory},
+ author = {Andrew R. Cohen and Christopher S. Bjornsson and Sally Temple and Gary Banker and Badrinath Roysam},
+ journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
+ publisher = {Institute of Electrical and Electronics Engineers (IEEE)},
+ year = {2009},
+ volume = {31},
+ number = {8},
+ pages = {1386-1403},
+ doi = {10.1109/tpami.2008.162}
+}
+
+
+@inproceedings{Keogh2004,
+ title = {Towards parameter-free data mining},
+ author = {Eamonn Keogh and Stefano Lonardi and Chotirat Ann Ratanamahatana},
+ booktitle = {Proceedings of the tenth ACM SIGKDD international conference on Knowledge discovery and data mining},
+ publisher = {ACM},
+ year = {2004},
+ series = {KDD04},
+ doi = {10.1145/1014052.1014077}
+}
+
+@article{Regot2014,
+ title = {High-Sensitivity Measurements of Multiple Kinase Activities in Live Single Cells},
+ author = {Sergi Regot and Jacob J. Hughey and Bryce T. Bajar and Silvia Carrasco and Markus W. Covert},
+ journal = {Cell},
+ publisher = {Elsevier {BV}},
+ year = {2014},
+ volume = {157},
+ number = {7},
+ pages = {1724-1734},
+ doi = {10.1016/j.cell.2014.04.039}
+}
+
+@book{Theodoridis2009,
+ title = {Pattern Recognition, Fourth Edition},
+ author = {Sergios Theodoridis and Konstantinos Koutroumbas},
+ publisher = {Elsevier},
+ year = {2009},
+ doi = {10.1016/b978-1-59749-272-0.x0001-2}
+}
+
+@article{Vitanyi2001,
+ title = {Algorithmic statistics},
+ author = {Peter Gacs and John Tromp and Paul M. Vitányi},
+ journal = {{IEEE} Transactions on Information Theory},
+ publisher = {Institute of Electrical and Electronics Engineers ({IEEE})},
+ year = {2001},
+ volume = {47},
+ number = {6},
+ pages = {2443-2463},
+ doi = {10.1109/18.945257},
+}
+@article{Vitanyi2004,
+ title = {Kolmogorov's Structure Functions and Model Selection},
+ author = {Nikolai Vereshchagin and Paul M. Vitányi},
+ journal = {{IEEE} Transactions on Information Theory},
+ publisher = {Institute of Electrical and Electronics Engineers ({IEEE})},
+ year = {2004},
+ volume = {50},
+ number = {12},
+ pages = {3265-3290},
+ doi = {10.1109/tit.2004.838346},
+}
+@article{Vitanyi2005,
+ author = {Rudi Cilibrasi and Paul M. Vitányi},
+ title = {Clustering by Compression},
+ journal = {{IEEE} Transactions on Information Theory},
+ publisher = {Institute of Electrical and Electronics Engineers ({IEEE})},
+ year = {2005},
+ volume = {51},
+ number = {4},
+ pages = {1523-1545},
+ doi = {10.1109/tit.2005.844059}
+}
+@article{Vitanyi2006,
+ title = {Meaningful Information},
+ author = {Paul Vitányi},
+ journal = {{IEEE} Transactions on Information Theory},
+ publisher = {Institute of Electrical and Electronics Engineers ({IEEE})},
+ year = {2006},
+ volume = {52},
+ number = {10},
+ pages = {4617-4626},
+ doi = {10.1109/tit.2006.881729}
+}
+@book{Vitanyi2008,
+ title = {An Introduction to Kolmogorov Complexity and Its Applications},
+ author = {Ming Li and Paul Vitányi},
+ publisher = {Springer New York},
+ year = {2008},
+ doi = {10.1007/978-0-387-49820-1}
+}
+
+@book{DIP_MATLAB,
+ author = {Gonzales, R and Woods, R and Eddins, S},
+ title = {Digital Image Processing Using MATLAB},
+ publisher = {Gatesmark Publishing},
+ address = {Knoxville, TN},
+ year = {2009},
+ type = {Book}
+}
+
+@article{Kamvar,
+ author = {Kamvar, S. D. and Klein, D. and Manning, C. D.},
+ title = {Spectral learning},
+ journal = {International Joint Conference of Artificial Intelligence},
+ pages = {561-566},
+ year = {2003},
+ type = {Journal Article}
+}
+
+@article{Jordan,
+ author = {Ng, Andrew Y. and Jordan, Michael and Weiss, Yair},
+ title = {On Spectral Clustering: Analysis and an algorithm},
+ journal = {Advances in Neural Information Processing Systems 14},
+ year = {2002},
+ type = {Journal Article}
+}
+
+@inproceedings{FLIF,
+ author = {Wuille, Jon Sneyers and Pieter},
+ title = {FLIF: Free lossless image format based on MANIAC compression},
+ booktitle = {2016 IEEE International Conference on Image Processing (ICIP)},
+ type = {Conference Proceedings}
+}
+