{
 "metadata": {
  "name": "",
  "signature": "sha256:e2b988ec2072f53f76001e59a307580dceb17180cd75dd7ced527dda42dc84ea"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# Quantifying spinal cord regeneration in zebrafish\n",
      "\n",
      "We want to quantify the amount of fluorescent cells in a wounded zebrafish embryo spinal column:\n",
      "\n",
      "<img src=\"images/zebrafish-spinal-cord-color.png\" alt=\"Zebrafish spinal cord\" style=\"width: 300px;\"/>\n",
      "\n",
      "The key steps are:\n",
      "\n",
      "- estimating the position and width of the cord\n",
      "- estimating the average fluorescence along the length of the cord"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%matplotlib inline\n",
      "from matplotlib import pyplot as plt, cm\n",
      "from skimage import io\n",
      "image = io.imread('images/zebrafish-spinal-cord.png')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# SciPy to estimate coordinates\n",
      "\n",
      "First, we get just the top and bottom rows of pixels, and use a 1D gaussian filter to smooth the signal."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from scipy import ndimage as nd\n",
      "top, bottom = image[[0, -1], :]\n",
      "\n",
      "fig, (ax0, ax1) = plt.subplots(nrows=1, ncols=2, figsize=(8, 3))\n",
      "\n",
      "top_smooth = nd.gaussian_filter1d(top, sigma=20)\n",
      "ax0.plot(top, color='blue', lw=2)\n",
      "ax0.plot(top_smooth, color='orange', lw=2)\n",
      "ax0.set_title('top')\n",
      "\n",
      "bottom_smooth = nd.gaussian_filter1d(bottom, sigma=20)\n",
      "ax1.plot(bottom, color='blue', lw=2)\n",
      "ax1.plot(bottom_smooth, color='orange', lw=2)\n",
      "ax1.set_title('bottom')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "With smooth curves, we can get the mode (the position of the center) and width of the signal."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "top_mode = top_smooth.argmax()\n",
      "top_max = top_smooth[top_mode]\n",
      "top_width = (top_smooth > float(top_max) / 2).sum()\n",
      "\n",
      "bottom_mode = bottom_smooth.argmax()\n",
      "bottom_max = bottom_smooth[bottom_mode]\n",
      "bottom_width = (bottom_smooth > float(bottom_max) / 2).sum()\n",
      "\n",
      "width = max(bottom_width, top_width)\n",
      "\n",
      "print(top_mode, top_width, bottom_mode, bottom_width)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# scikit-image to trace the profile\n",
      "\n",
      "Now, use `measure.profile_line` to trace from (0, `top_mode`) to (-1, `bottom_mode`)."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from skimage import measure\n",
      "trace = measure.profile_line(None) # Replace `None` with correct args"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "Finally, plot the trace."
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "plt.plot(trace, color='black', lw=2)\n",
      "plt.xlabel('position along embryo')\n",
      "plt.ylabel('mean fluorescence intensity')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "From this trace, we can compute various summary statistics (e.g. min/max, gap width, slope, etc), and plot these over time as the wound recovers."
     ]
    }
   ],
   "metadata": {}
  }
 ]
}