11_NIST-hsf-vendor-comparison.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Exploratory HSF statistics notebook\n",
    "\n",
    "Not intended/designed for any presentation/paper/report, this is purely exploratory and stemmed from internal discussions, but keeping it here as a record and to reproduce it.\n",
    "\n",
    "Specifically, Nikola wanted to see some t-tests/ANOVA between vendors using Agah's HSF dashboard data.\n",
    "\n",
    "Mathieu points out that the difficulty in doing \"correct\" statistics stem from some of assumptions of independence, specifically that some sites shared the same phantom, some sites acquired multiple measurements using the same implementation or one group implemented the sequence at multiple sites. Also, GE was only acquired by two submissions (on three scanners), most Philips measurements came from a single submission, etc. To see a diagram of the distribution of datasets, [click here](https://rrsg2020.github.io/paper/_images/figure_3_full.png).\n",
    "\n",
    "So all that to say, take the results below with a grain of salt if you aren't Mathieu/Nikola or haven't discussed it with them.\n",
    "\n",
    "Note: Duplicate T1 maps stemming from the same measurement (i.e. when both magnitude-only and complex T1 mapping data were used) were removed, so only one T1 map per unique measuement exists in the HSF pickled files."
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pathlib import Path\n",
    "import pandas as pd\n",
    "import json\n",
    "import nibabel as nib\n",
    "import numpy as np\n",
    "from scipy import stats\n",
    "import os\n",
    "import shutil\n",
    "\n",
    "from src.database import *\n",
    "from src.nist import get_reference_NIST_values, get_NIST_ids\n",
    "from src.tools import calc_error\n",
    "from src.nist import temperature_correction\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "plt.style.use('custom_matplotlibrc')\n",
    "plt.rcParams[\"figure.figsize\"] = (10,10)\n"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load Agah's pickled data from Juan's R dashboard\n",
    "Note: Agah only emailed me the actual pickled files when generating the paper figures - need to get instruction by him on how to reproduce these. If I were to guess, it would be to go to Juan's dahsboard repo and use [this file](https://github.com/rrsg2020/rrsg2020-stats-dashboard/blob/main/hierarchical_shift_function.R) and \"pickle\" the outputs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "        <script type=\"text/javascript\">\n",
       "        window.PlotlyConfig = {MathJaxConfig: 'local'};\n",
       "        if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}\n",
       "        if (typeof require !== 'undefined') {\n",
       "        require.undef(\"plotly\");\n",
       "        requirejs.config({\n",
       "            paths: {\n",
       "                'plotly': ['https://cdn.plot.ly/plotly-latest.min']\n",
       "            }\n",
       "        });\n",
       "        require(['plotly'], function(Plotly) {\n",
       "            window._Plotly = Plotly;\n",
       "        });\n",
       "        }\n",
       "        </script>\n",
       "        "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from plotly.offline import init_notebook_mode, iplot\n",
    "from plotly.subplots import make_subplots\n",
    "\n",
    "init_notebook_mode(connected=True) \n",
    "\n",
    "def explode_mount_traces(xp,mosaic,rw,cl,shw):\n",
    "    traces = []\n",
    "    for trace in range(len(xp[\"data\"])):\n",
    "        if not shw:\n",
    "            xp[\"data\"][trace]['showlegend'] = False\n",
    "        traces.append(xp[\"data\"][trace])\n",
    "    for trc in traces:\n",
    "        mosaic.append_trace(trc, row=rw, col=cl)\n",
    "    return mosaic    \n",
    "    \n",
    "\n",
    "sites = pd.read_pickle(\"./databases/R_HSF_sites.pkl\")\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Slice tables per sphere\n",
    "sphere_1=sites[sites['sphere']==1]\n",
    "sphere_2=sites[sites['sphere']==2]\n",
    "sphere_3=sites[sites['sphere']==3]\n",
    "sphere_4=sites[sites['sphere']==4]\n",
    "sphere_5=sites[sites['sphere']==5]\n",
    "sphere_6=sites[sites['sphere']==6]"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sphere 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Slice sphere tables per vendor\n",
    "GE_sphere_1=sphere_1[sphere_1['vendor']=='GE']\n",
    "Siemens_sphere_1=sphere_1[sphere_1['vendor']=='Siemens']\n",
    "Philips_sphere_1=sphere_1[sphere_1['vendor']=='Philips']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['4.001 US', '4.002 US', '10.002 US']\n",
      "['1.002 London', '2.001 US', '3.001 Montreal', '8.001 Montreal', '8.002 Montreal', '9.001 Australia', '11.002 US', '12.001 Montreal']\n",
      "['5.002 Montreal', '6.002 Germany', '6.004 Germany', '6.006 Germany', '6.008 Germany', '6.01 Germany', '6.012 Germany', '6.014 Germany', '7.002 Montreal']\n"
     ]
    }
   ],
   "source": [
    "# Get lists of T1 measurements IDs\n",
    "GE_meas_ids = GE_sphere_1['id'].unique()\n",
    "Siemens_meas_ids = Siemens_sphere_1['id'].unique()\n",
    "Philips_meas_ids = Philips_sphere_1['id'].unique()\n",
    "\n",
    "print(GE_meas_ids.tolist())\n",
    "print(Siemens_meas_ids.tolist())\n",
    "print(Philips_meas_ids.tolist())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Collapse HSF percentiles to one mean value per scanner measurement\n",
    "mean_GE_sphere_1 = []\n",
    "mean_Siemens_sphere_1 = []\n",
    "mean_Philips_sphere_1 = []\n",
    "\n",
    "for scanner in GE_meas_ids:\n",
    "    single_scanner=GE_sphere_1[GE_sphere_1['id']==scanner]\n",
    "    mean_GE_sphere_1.append(np.mean(single_scanner['difference'].tolist()))\n",
    "\n",
    "for scanner in Siemens_meas_ids:\n",
    "    single_scanner=Siemens_sphere_1[Siemens_sphere_1['id']==scanner]\n",
    "    mean_Siemens_sphere_1.append(np.mean(single_scanner['difference'].tolist()))\n",
    "\n",
    "for scanner in Philips_meas_ids:\n",
    "    single_scanner=Philips_sphere_1[Philips_sphere_1['id']==scanner]\n",
    "    mean_Philips_sphere_1.append(np.mean(single_scanner['difference'].tolist()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[10.238518518518605, 83.17208415798103, -55.09555640088037]\n",
      "[227.83511670219409, -110.18351375052855, 59.471162057700816, 44.10481481481487, -54.44674775132689, 111.88042296007445, -26.69925925925918, 15.148022848144869]\n",
      "[19.235060350531462, -42.96338415592082, -58.95338415592086, 185.03024589947523, 177.60036952504566, 61.48269563491782, 65.00777777777782, -6.839273177083265, -126.3129629629629]\n"
     ]
    }
   ],
   "source": [
    "# Raw means\n",
    "print(mean_GE_sphere_1)\n",
    "print(mean_Siemens_sphere_1)\n",
    "print(mean_Philips_sphere_1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ttest_indResult(statistic=-0.31044520455573404, pvalue=0.7632913544593019)\n",
      "Ttest_indResult(statistic=-0.268302733047986, pvalue=0.7939197314405384)\n",
      "Ttest_indResult(statistic=0.05946919777831455, pvalue=0.953363465704431)\n",
      "0.9542141203962895\n"
     ]
    }
   ],
   "source": [
    "# Two group t-tests (independent)\n",
    "print(stats.ttest_ind(mean_GE_sphere_1, mean_Siemens_sphere_1))\n",
    "print(stats.ttest_ind(mean_GE_sphere_1, mean_Philips_sphere_1))\n",
    "print(stats.ttest_ind(mean_Siemens_sphere_1, mean_Philips_sphere_1))\n",
    "\n",
    "# One way anova\n",
    "fvalue, pvalue = stats.f_oneway(mean_GE_sphere_1, mean_Siemens_sphere_1, mean_Philips_sphere_1)\n",
    "print(pvalue)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Sphere 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Slice sphere tables per vendor\n",
    "GE_sphere_2=sphere_2[sphere_2['vendor']=='GE']\n",
    "Siemens_sphere_2=sphere_2[sphere_2['vendor']=='Siemens']\n",
    "Philips_sphere_2=sphere_2[sphere_2['vendor']=='Philips']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get lists of T1 measurements IDs\n",
    "GE_meas_ids = GE_sphere_2['id'].unique()\n",
    "Siemens_meas_ids = Siemens_sphere_2['id'].unique()\n",
    "Philips_meas_ids = Philips_sphere_2['id'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Collapse HSF percentiles to one mean value per scanner measurement\n",
    "mean_GE_sphere_2 = []\n",
    "mean_Siemens_sphere_2 = []\n",
    "mean_Philips_sphere_2 = []\n",
    "\n",
    "for scanner in GE_meas_ids:\n",
    "    single_scanner=GE_sphere_2[GE_sphere_2['id']==scanner]\n",
    "    mean_GE_sphere_2.append(np.mean(single_scanner['difference'].tolist()))\n",
    "\n",
    "for scanner in Siemens_meas_ids:\n",
    "    single_scanner=Siemens_sphere_2[Siemens_sphere_2['id']==scanner]\n",
    "    mean_Siemens_sphere_2.append(np.mean(single_scanner['difference'].tolist()))\n",
    "\n",
    "for scanner in Philips_meas_ids:\n",
    "    single_scanner=Philips_sphere_2[Philips_sphere_2['id']==scanner]\n",
    "    mean_Philips_sphere_2.append(np.mean(single_scanner['difference'].tolist()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[16.616666666666738, -38.465651802667125, 18.638834556468485]\n",
      "[72.61598799677729, -31.61428807076292, 26.168048218420736, 62.264814814814855, 37.40294357142999, 70.8715569820551, 27.870000000000044, 22.798254504444557]\n",
      "[12.609779265874472, -10.837616407906227, -28.791690481980343, 107.6630656878293, 144.10031603794988, 104.25717256614183, 74.06851851851856, 61.00545684606472, -15.164814814814767]\n"
     ]
    }
   ],
   "source": [
    "# Raw means\n",
    "print(mean_GE_sphere_2)\n",
    "print(mean_Siemens_sphere_2)\n",
    "print(mean_Philips_sphere_2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ttest_indResult(statistic=-1.624037874934912, pvalue=0.13881477461794206)\n",
      "Ttest_indResult(statistic=-1.3208387844764264, pvalue=0.21598239707799294)\n",
      "Ttest_indResult(statistic=-0.5544632406208166, pvalue=0.587434311465362)\n",
      "0.3273063743297731\n"
     ]
    }
   ],
   "source": [
    "# Two group t-tests (independent)\n",
    "print(stats.ttest_ind(mean_GE_sphere_2, mean_Siemens_sphere_2))\n",
    "print(stats.ttest_ind(mean_GE_sphere_2, mean_Philips_sphere_2))\n",
    "print(stats.ttest_ind(mean_Siemens_sphere_2, mean_Philips_sphere_2))\n",
    "\n",
    "# One way anova\n",
    "fvalue, pvalue = stats.f_oneway(mean_GE_sphere_2, mean_Siemens_sphere_2, mean_Philips_sphere_2)\n",
    "print(pvalue)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sphere 3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Slice sphere tables per vendor\n",
    "GE_sphere_3=sphere_3[sphere_3['vendor']=='GE']\n",
    "Siemens_sphere_3=sphere_3[sphere_3['vendor']=='Siemens']\n",
    "Philips_sphere_3=sphere_3[sphere_3['vendor']=='Philips']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get lists of T1 measurements IDs\n",
    "GE_meas_ids = GE_sphere_3['id'].unique()\n",
    "Siemens_meas_ids = Siemens_sphere_3['id'].unique()\n",
    "Philips_meas_ids = Philips_sphere_3['id'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Collapse HSF percentiles to one mean value per scanner measurement\n",
    "mean_GE_sphere_3 = []\n",
    "mean_Siemens_sphere_3 = []\n",
    "mean_Philips_sphere_3 = []\n",
    "\n",
    "for scanner in GE_meas_ids:\n",
    "    single_scanner=GE_sphere_3[GE_sphere_3['id']==scanner]\n",
    "    mean_GE_sphere_3.append(np.mean(single_scanner['difference'].tolist()))\n",
    "\n",
    "for scanner in Siemens_meas_ids:\n",
    "    single_scanner=Siemens_sphere_3[Siemens_sphere_3['id']==scanner]\n",
    "    mean_Siemens_sphere_3.append(np.mean(single_scanner['difference'].tolist()))\n",
    "\n",
    "for scanner in Philips_meas_ids:\n",
    "    single_scanner=Philips_sphere_3[Philips_sphere_3['id']==scanner]\n",
    "    mean_Philips_sphere_3.append(np.mean(single_scanner['difference'].tolist()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[-160.7303703703703, -151.48780824363428, -32.87711143476776]\n",
      "[-15.075165305018155, -65.09453204151757, -10.873688099371444, -35.865555555555474, -55.99463965608433, -19.225177317708372, -130.45999999999995, -0.6584843977774426]\n",
      "[-36.69957423941829, -55.64417318204325, -66.17380281167286, 2.250230396825133, -0.853037360693722, -29.248589708991858, -31.46925925925921, -34.21404230902777, -81.71740740740736]\n"
     ]
    }
   ],
   "source": [
    "# Raw means\n",
    "print(mean_GE_sphere_3)\n",
    "print(mean_Siemens_sphere_3)\n",
    "print(mean_Philips_sphere_3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ttest_indResult(statistic=-2.159073062254625, pvalue=0.05915666189831824)\n",
      "Ttest_indResult(statistic=-2.8950019211480784, pvalue=0.015969740538737034)\n",
      "Ttest_indResult(statistic=-0.26670020539810346, pvalue=0.7933315531451209)\n",
      "0.029885677216495432\n"
     ]
    }
   ],
   "source": [
    "# Two group t-tests (independent)\n",
    "print(stats.ttest_ind(mean_GE_sphere_3, mean_Siemens_sphere_3))\n",
    "print(stats.ttest_ind(mean_GE_sphere_3, mean_Philips_sphere_3))\n",
    "print(stats.ttest_ind(mean_Siemens_sphere_3, mean_Philips_sphere_3))\n",
    "\n",
    "# One way anova\n",
    "fvalue, pvalue = stats.f_oneway(mean_GE_sphere_3, mean_Siemens_sphere_3, mean_Philips_sphere_3)\n",
    "print(pvalue)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sphere 4"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Slice sphere tables per vendor\n",
    "GE_sphere_4=sphere_4[sphere_4['vendor']=='GE']\n",
    "Siemens_sphere_4=sphere_4[sphere_4['vendor']=='Siemens']\n",
    "Philips_sphere_4=sphere_4[sphere_4['vendor']=='Philips']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get lists of T1 measurements IDs\n",
    "GE_meas_ids = GE_sphere_4['id'].unique()\n",
    "Siemens_meas_ids = Siemens_sphere_4['id'].unique()\n",
    "Philips_meas_ids = Philips_sphere_4['id'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Collapse HSF percentiles to one mean value per scanner measurement\n",
    "\n",
    "mean_GE_sphere_4 = []\n",
    "mean_Siemens_sphere_4 = []\n",
    "mean_Philips_sphere_4 = []\n",
    "\n",
    "for scanner in GE_meas_ids:\n",
    "    single_scanner=GE_sphere_4[GE_sphere_4['id']==scanner]\n",
    "    mean_GE_sphere_4.append(np.mean(single_scanner['difference'].tolist()))\n",
    "\n",
    "for scanner in Siemens_meas_ids:\n",
    "    single_scanner=Siemens_sphere_4[Siemens_sphere_4['id']==scanner]\n",
    "    mean_Siemens_sphere_4.append(np.mean(single_scanner['difference'].tolist()))\n",
    "\n",
    "for scanner in Philips_meas_ids:\n",
    "    single_scanner=Philips_sphere_4[Philips_sphere_4['id']==scanner]\n",
    "    mean_Philips_sphere_4.append(np.mean(single_scanner['difference'].tolist()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[-203.8148148148148, -283.3969468547454, 4.754188611524581]\n",
      "[0.6657211913426662, -20.20899539230465, -1.1346400669639378, 3.7677777777777868, -7.462786322751084, 14.152764924768512, -141.3511111111111, -55.25051672518485]\n",
      "[-39.51672701719609, -18.95590667741361, -20.218128899635822, 19.56848076719546, 21.59307750041343, 9.098638439153357, 4.883703703703708, 4.339554010416704, -25.174074074074056]\n"
     ]
    }
   ],
   "source": [
    "# Raw means\n",
    "print(mean_GE_sphere_4)\n",
    "print(mean_Siemens_sphere_4)\n",
    "print(mean_Philips_sphere_4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ttest_indResult(statistic=-2.3893499826196267, pvalue=0.040600629500444504)\n",
      "Ttest_indResult(statistic=-3.374802761931636, pvalue=0.00706410304484939)\n",
      "Ttest_indResult(statistic=-1.1220881385647434, pvalue=0.27946622432686785)\n",
      "0.005424373669179906\n"
     ]
    }
   ],
   "source": [
    "# Two group t-tests (independent)\n",
    "print(stats.ttest_ind(mean_GE_sphere_4, mean_Siemens_sphere_4))\n",
    "print(stats.ttest_ind(mean_GE_sphere_4, mean_Philips_sphere_4))\n",
    "print(stats.ttest_ind(mean_Siemens_sphere_4, mean_Philips_sphere_4))\n",
    "\n",
    "# One way anova\n",
    "fvalue, pvalue = stats.f_oneway(mean_GE_sphere_4, mean_Siemens_sphere_4, mean_Philips_sphere_4)\n",
    "print(pvalue)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sphere 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Slice sphere tables per vendor\n",
    "GE_sphere_5=sphere_5[sphere_5['vendor']=='GE']\n",
    "Siemens_sphere_5=sphere_5[sphere_5['vendor']=='Siemens']\n",
    "Philips_sphere_5=sphere_5[sphere_5['vendor']=='Philips']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get lists of T1 measurements IDs\n",
    "GE_meas_ids = GE_sphere_5['id'].unique()\n",
    "Siemens_meas_ids = Siemens_sphere_5['id'].unique()\n",
    "Philips_meas_ids = Philips_sphere_5['id'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Collapse HSF percentiles to one mean value per scanner measurement\n",
    "mean_GE_sphere_5 = []\n",
    "mean_Siemens_sphere_5 = []\n",
    "mean_Philips_sphere_5 = []\n",
    "\n",
    "for scanner in GE_meas_ids:\n",
    "    single_scanner=GE_sphere_5[GE_sphere_5['id']==scanner]\n",
    "    mean_GE_sphere_5.append(np.mean(single_scanner['difference'].tolist()))\n",
    "\n",
    "for scanner in Siemens_meas_ids:\n",
    "    single_scanner=Siemens_sphere_5[Siemens_sphere_5['id']==scanner]\n",
    "    mean_Siemens_sphere_5.append(np.mean(single_scanner['difference'].tolist()))\n",
    "\n",
    "for scanner in Philips_meas_ids:\n",
    "    single_scanner=Philips_sphere_5[Philips_sphere_5['id']==scanner]\n",
    "    mean_Philips_sphere_5.append(np.mean(single_scanner['difference'].tolist()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[-59.02962962962964, -82.35586638020831, 21.68946290219911]\n",
      "[4.12777410641203, -7.022923136148841, -15.418364438657383, -0.02518518518519386, -2.2215320370370364, 13.045134551504601, -39.229629629629656, 0.39892904222222747]\n",
      "[-25.84679976851851, -5.731127934027793, -4.757794600694455, 24.160389444444476, 14.11835179108799, 3.822041296296277, -1.0962962962963034, 0.45354595486111293, -13.598518518518524]\n"
     ]
    }
   ],
   "source": [
    "# Raw means\n",
    "print(mean_GE_sphere_5)\n",
    "print(mean_Siemens_sphere_5)\n",
    "print(mean_Philips_sphere_5)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ttest_indResult(statistic=-1.7210122821228842, pvalue=0.1193559769168836)\n",
      "Ttest_indResult(statistic=-2.111181590492552, pvalue=0.06092096388676267)\n",
      "Ttest_indResult(statistic=-0.6583950357083797, pvalue=0.5202626508697117)\n",
      "0.06673723177452698\n"
     ]
    }
   ],
   "source": [
    "# Two group t-tests (independent)\n",
    "print(stats.ttest_ind(mean_GE_sphere_5, mean_Siemens_sphere_5))\n",
    "print(stats.ttest_ind(mean_GE_sphere_5, mean_Philips_sphere_5))\n",
    "print(stats.ttest_ind(mean_Siemens_sphere_5, mean_Philips_sphere_5))\n",
    "\n",
    "# One way anova\n",
    "fvalue, pvalue = stats.f_oneway(mean_GE_sphere_5, mean_Siemens_sphere_5, mean_Philips_sphere_5)\n",
    "print(pvalue)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sphere 6"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Slice sphere tables per vendor\n",
    "GE_sphere_6=sphere_6[sphere_6['vendor']=='GE']\n",
    "Siemens_sphere_6=sphere_6[sphere_6['vendor']=='Siemens']\n",
    "Philips_sphere_6=sphere_6[sphere_6['vendor']=='Philips']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get lists of T1 measurements IDs\n",
    "GE_meas_ids = GE_sphere_6['id'].unique()\n",
    "Siemens_meas_ids = Siemens_sphere_6['id'].unique()\n",
    "Philips_meas_ids = Philips_sphere_6['id'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Collapse HSF percentiles to one mean value per scanner measurement\n",
    "mean_GE_sphere_6 = []\n",
    "mean_Siemens_sphere_6 = []\n",
    "mean_Philips_sphere_6 = []\n",
    "\n",
    "for scanner in GE_meas_ids:\n",
    "    single_scanner=GE_sphere_6[GE_sphere_6['id']==scanner]\n",
    "    mean_GE_sphere_6.append(np.mean(single_scanner['difference'].tolist()))\n",
    "\n",
    "for scanner in Siemens_meas_ids:\n",
    "    single_scanner=Siemens_sphere_6[Siemens_sphere_6['id']==scanner]\n",
    "    mean_Siemens_sphere_6.append(np.mean(single_scanner['difference'].tolist()))\n",
    "\n",
    "for scanner in Philips_meas_ids:\n",
    "    single_scanner=Philips_sphere_6[Philips_sphere_6['id']==scanner]\n",
    "    mean_Philips_sphere_6.append(np.mean(single_scanner['difference'].tolist()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[-28.04037037037034, -84.46218462094905, 9.350071711723432]\n",
      "[3.9484034478009185, -0.09778852618832287, -1.7551220444773636, -31.0922222222222, -32.377728439153316, 10.370886530671308, -16.2622222222222, 2.931727359258888]\n",
      "[-1.773545800264723, 0.9065175553899394, -1.8216305927582286, 19.519385846560713, -22.56585421420249, -28.038657539682138, -31.898148148148135, -28.836768026620334, -34.942222222222206]\n"
     ]
    }
   ],
   "source": [
    "# Raw means\n",
    "print(mean_GE_sphere_6)\n",
    "print(mean_Siemens_sphere_6)\n",
    "print(mean_Philips_sphere_6)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ttest_indResult(statistic=-1.463981423224981, pvalue=0.17723267546515367)\n",
      "Ttest_indResult(statistic=-1.1064207337757475, pvalue=0.29445008840318165)\n",
      "Ttest_indResult(statistic=0.7304361120900091, pvalue=0.4763782200872042)\n",
      "0.27490841588805615\n"
     ]
    }
   ],
   "source": [
    "# Two group t-tests (independent)\n",
    "print(stats.ttest_ind(mean_GE_sphere_6, mean_Siemens_sphere_6))\n",
    "print(stats.ttest_ind(mean_GE_sphere_6, mean_Philips_sphere_6))\n",
    "print(stats.ttest_ind(mean_Siemens_sphere_6, mean_Philips_sphere_6))\n",
    "\n",
    "# One way anova\n",
    "fvalue, pvalue = stats.f_oneway(mean_GE_sphere_6, mean_Siemens_sphere_6, mean_Philips_sphere_6)\n",
    "print(pvalue)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Summary Table\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Summary Table\n",
    "\n",
    "import pandas as pd\n",
    "import dataframe_image as dfi\n",
    "\n",
    "pd.set_option('display.max_rows', None)\n",
    "pd.set_option('display.max_columns', None)\n",
    "pd.set_option('display.width', 1000)\n",
    "pd.set_option('display.colheader_justify', 'center')\n",
    "pd.set_option('display.precision', 2)\n",
    "\n",
    "\n",
    "columns = [\n",
    "    'T-test (GE vs Siemens)',\n",
    "    'T-test (GE vs Philips)',\n",
    "    'T-test (Siemens vs Philips)',\n",
    "    'ANOVA',\n",
    "]\n",
    "\n",
    "col_vals = [\n",
    "    None,\n",
    "    None,\n",
    "    None,\n",
    "    None,\n",
    "]\n",
    "\n",
    "df_setup = {\n",
    "    'Sphere 1': col_vals,\n",
    "    'Sphere 2': col_vals,\n",
    "    'Sphere 3': col_vals,\n",
    "    'Sphere 4': col_vals,\n",
    "    'Sphere 5': col_vals,\n",
    "    'Sphere 6': col_vals,\n",
    "}\n",
    "\n",
    "stats_table = pd.DataFrame.from_dict(df_setup, orient='index', columns=columns)\n",
    "\n",
    "stats_table['T-test (GE vs Siemens)']['Sphere 1'] = stats.ttest_ind(mean_GE_sphere_1, mean_Siemens_sphere_1).pvalue\n",
    "stats_table['T-test (GE vs Siemens)']['Sphere 2'] = stats.ttest_ind(mean_GE_sphere_2, mean_Siemens_sphere_2).pvalue\n",
    "stats_table['T-test (GE vs Siemens)']['Sphere 3'] = stats.ttest_ind(mean_GE_sphere_3, mean_Siemens_sphere_3).pvalue\n",
    "stats_table['T-test (GE vs Siemens)']['Sphere 4'] = stats.ttest_ind(mean_GE_sphere_4, mean_Siemens_sphere_4).pvalue\n",
    "stats_table['T-test (GE vs Siemens)']['Sphere 5'] = stats.ttest_ind(mean_GE_sphere_5, mean_Siemens_sphere_5).pvalue\n",
    "stats_table['T-test (GE vs Siemens)']['Sphere 6'] = stats.ttest_ind(mean_GE_sphere_6, mean_Siemens_sphere_6).pvalue\n",
    "\n",
    "stats_table['T-test (GE vs Philips)']['Sphere 1'] = stats.ttest_ind(mean_GE_sphere_1, mean_Philips_sphere_1).pvalue\n",
    "stats_table['T-test (GE vs Philips)']['Sphere 2'] = stats.ttest_ind(mean_GE_sphere_2, mean_Philips_sphere_2).pvalue\n",
    "stats_table['T-test (GE vs Philips)']['Sphere 3'] = stats.ttest_ind(mean_GE_sphere_3, mean_Philips_sphere_3).pvalue\n",
    "stats_table['T-test (GE vs Philips)']['Sphere 4'] = stats.ttest_ind(mean_GE_sphere_4, mean_Philips_sphere_4).pvalue\n",
    "stats_table['T-test (GE vs Philips)']['Sphere 5'] = stats.ttest_ind(mean_GE_sphere_5, mean_Philips_sphere_5).pvalue\n",
    "stats_table['T-test (GE vs Philips)']['Sphere 6'] = stats.ttest_ind(mean_GE_sphere_6, mean_Philips_sphere_6).pvalue\n",
    "\n",
    "stats_table['T-test (Siemens vs Philips)']['Sphere 1'] = stats.ttest_ind(mean_Siemens_sphere_1, mean_Philips_sphere_1).pvalue\n",
    "stats_table['T-test (Siemens vs Philips)']['Sphere 2'] = stats.ttest_ind(mean_Siemens_sphere_2, mean_Philips_sphere_2).pvalue\n",
    "stats_table['T-test (Siemens vs Philips)']['Sphere 3'] = stats.ttest_ind(mean_Siemens_sphere_3, mean_Philips_sphere_3).pvalue\n",
    "stats_table['T-test (Siemens vs Philips)']['Sphere 4'] = stats.ttest_ind(mean_Siemens_sphere_4, mean_Philips_sphere_4).pvalue\n",
    "stats_table['T-test (Siemens vs Philips)']['Sphere 5'] = stats.ttest_ind(mean_Siemens_sphere_5, mean_Philips_sphere_5).pvalue\n",
    "stats_table['T-test (Siemens vs Philips)']['Sphere 6'] = stats.ttest_ind(mean_Siemens_sphere_6, mean_Philips_sphere_6).pvalue\n",
    "\n",
    "stats_table['ANOVA']['Sphere 1'] = stats.f_oneway(mean_GE_sphere_1, mean_Siemens_sphere_1, mean_Philips_sphere_1).pvalue\n",
    "stats_table['ANOVA']['Sphere 2'] = stats.f_oneway(mean_GE_sphere_2, mean_Siemens_sphere_2, mean_Philips_sphere_2).pvalue\n",
    "stats_table['ANOVA']['Sphere 3'] = stats.f_oneway(mean_GE_sphere_3, mean_Siemens_sphere_3, mean_Philips_sphere_3).pvalue\n",
    "stats_table['ANOVA']['Sphere 4'] = stats.f_oneway(mean_GE_sphere_4, mean_Siemens_sphere_4, mean_Philips_sphere_4).pvalue\n",
    "stats_table['ANOVA']['Sphere 5'] = stats.f_oneway(mean_GE_sphere_5, mean_Siemens_sphere_5, mean_Philips_sphere_5).pvalue\n",
    "stats_table['ANOVA']['Sphere 6'] = stats.f_oneway(mean_GE_sphere_6, mean_Siemens_sphere_6, mean_Philips_sphere_6).pvalue\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "         T-test (GE vs Siemens) T-test (GE vs Philips) T-test (Siemens vs Philips) ANOVA\n",
      "Sphere 1          0.76                   0.79                      0.95             0.95\n",
      "Sphere 2          0.14                   0.22                      0.59             0.33\n",
      "Sphere 3          0.06                   0.02                      0.79             0.03\n",
      "Sphere 4          0.04                   0.01                      0.28             0.01\n",
      "Sphere 5          0.12                   0.06                      0.52             0.07\n",
      "Sphere 6          0.18                   0.29                      0.48             0.27\n"
     ]
    }
   ],
   "source": [
    "print(stats_table)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Export"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "objc[8023]: Class WebSwapCGLLayer is implemented in both /System/Library/Frameworks/WebKit.framework/Versions/A/Frameworks/WebCore.framework/Versions/A/Frameworks/libANGLE-shared.dylib (0x7ffa53ccc550) and /Applications/Google Chrome.app/Contents/Frameworks/Google Chrome Framework.framework/Versions/111.0.5563.110/Libraries/libGLESv2.dylib (0x1121df850). One of the two will be used. Which one is undefined.\n",
      "33498 bytes written to file /var/folders/l1/1xswjffd73l8yp7dd7pq9lyw0000gn/T/tmpq7wzc6a6/temp.png\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "PosixPath('/Users/mathieuboudreau/neuropoly/github/analysis/plots/11_NIST_hsf_vendor_comparison/NIST_hsf_vendor_comparison.png')"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pwd = os.getcwd()\n",
    "folder_path = Path(\"plots/11_NIST_hsf_vendor_comparison/\")\n",
    "dfi.export(stats_table, 'NIST_hsf_vendor_comparison.png')\n",
    "\n",
    "shutil.move('NIST_hsf_vendor_comparison.png', Path(pwd) / folder_path / 'NIST_hsf_vendor_comparison.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from IPython.display import Image\n",
    "Image(filename=Path(pwd) / folder_path / 'NIST_hsf_vendor_comparison.png') "
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "analysis_arm",
   "language": "python",
   "name": "analysis_arm"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.16"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}