super_voxel.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import random\n",
    "import nibabel as nib\n",
    "import matplotlib.pyplot as plt\n",
    "import math\n",
    "from skimage.segmentation import slic\n",
    "from skimage.segmentation import mark_boundaries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def max_min_normalize(img):\n",
    "    img = np.asarray(img)\n",
    "    return (img-np.min(img)) / (np.max(img)-np.min(img))\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "img = nib.load('/home/tony/Documents/IBSR_preprocessed_2d/IBSR_01/image_80.nii.gz').get_data().astype(np.float32)\n",
    "#img = nib.load('/home/tony/Documents/IBSR_preprocessed_2d/IBSR_01/image_80.nii.gz').get_data().astype(np.int64)\n",
    "label = nib.load('/home/tony/Documents/IBSR_preprocessed_2d/IBSR_01/label_80.nii.gz').get_data().astype(np.int64)\n",
    "#img = (max_min_normalize(img)*255).astype(int)\n",
    "#img = max_min_normalize(img)\n",
    "img = np.expand_dims(img, axis=2)\n",
    "img = np.concatenate((img, img, img), 2)\n",
    "segments = slic(img.astype('double'), n_segments=2000, compactness=30)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7fba9d991da0>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 581,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1587"
      ]
     },
     "execution_count": 581,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "segments.max()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 582,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f910b7f9240>"
      ]
     },
     "execution_count": 582,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(mark_boundaries(img,segments))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 583,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f910b75a8d0>"
      ]
     },
     "execution_count": 583,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQYAAAD8CAYAAACVSwr3AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjAsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+17YcXAAAR0ElEQVR4nO3df8id5X3H8fdnGiO1/soSJY1h2pmN2cFSDTbiKB1iq6GQFGzRP2q2CU9xEVro/kjbPxYYhW6slhU24RGlcbRasa3mj2yrDQUZGGuUNP7IrE9tVp8mJGaVVFZIo/3uj3Mfe3Ku8+M+59w/n+fzgodzniv3Oed77pz7c67run88igjMzHr9Xt0FmFnzOBjMLOFgMLOEg8HMEg4GM0s4GMwsUVowSLpF0iuSFiTtLOt1zKx4KuM4BknnAD8BbgYWgWeBOyLi5cJfzMwKV1aP4XpgISJei4jfAI8AW0t6LTMr2LklPe864PWe3xeBDw1b+DytjPO5oKRSzAzgLd48GRFr8ixbVjBoQNtZYxZJc8AcwPm8hw/pppJKMTOAH8Rj/5N32bKGEovA+p7frwCO9i4QEfMRsSkiNq1gZUllmNk0ygqGZ4ENkq6SdB5wO7CnpNcys4KVMpSIiLcl3QP8J3AO8GBEvFTGa5lZ8cqaYyAi9gJ7y3p+MyuPj3w0s4SDwcwSDgYzSzgYzCzhYDCzhIPBzBIOBjNLOBjMLOFgMLOEg8HMEg4GM0s4GMws4WAws4SDwcwSDgYzSzgYzCzhYDCzhIPBzBIOBjNLOBjMLOFgMLOEg8HMEg4GM0s4GMws4WAws4SDwcwSDgYzSzgYzCzhYDCzhIPBzBIOBjNLOBjMLOFgMLOEg8HMEufO8mBJR4C3gHeAtyNik6RVwLeBK4EjwKci4s3ZyjSzKhXRY/iLiNgYEZuy33cC+yJiA7Av+93MWqSMocRWYHd2fzewrYTXMLMSzRoMAXxf0nOS5rK2yyPiGEB2e9mgB0qak3RA0oEznJ6xDDMr0kxzDMCNEXFU0mXAk5L+O+8DI2IemAe4SKtixjrMrEAz9Rgi4mh2ewL4HnA9cFzSWoDs9sSsRZpZtaYOBkkXSLqwex/4KPAisAfYni22HXhi1iLNrFqzDCUuB74nqfs834qI/5D0LPCopLuAnwOfnL1MM6vS1MEQEa8Bfzag/X+Bm2Ypyqp3au/VSdvFWxZqqMSaYNbJR7OJ9AeQw6eZfEi0mSUcDEvUqb1XDxwemOXhoURL5d3oT+29ujHd9UE1N6k++x0HQwvM+s1f9cbneYT2czAsE2UOK8Y9t4c07eNgsESV3/DuTTSTJx8tt5NzN1T+mu5t1MM9BpvIybkbWD3/9Fm3K3ijkOd2CDSHeww2sTp6DlYtB0MLNHUc3g2IM4+vmfo5xh1vMei9u2dRPg8lWuLiLQuVbRDd12lCIPW+59514F2i5XKPoQXacBTjLL2GIjR9/bSNg6Hh2vSBn3VIYc3hoUTD9XeR69qA8k44nnl8DSu2FbOXop/DozoOhhaoa4M4tfdqzjy+htXzT0/82P0bH3v3/uaDt737++aDtxVWXz+fd1EcDyVsrEl3T/aGwqDfy+aexezcY1jGBm2wZX6jV6lJe1bayD2GhvO332y8/qbjHkOL7d/42NTf8Hm7992JxDx7HJ7bdd9UtRTJQVAM9xgabJIPeRHj+FEhU9aehio4LCbnYGipQUFQ1CRf1ZOFoxRVi8NhMh5KNFiew6An3XBGLV9FIEwz/JllyNSr//BqG87BUJEyPpS9xwd0jTpeIO+GX/aeiXEb+qjeUFG1+e9ojOZgqEERJwAtld2K1kyeY2iAace/VQ8LqpiAHFf3/o2PNWoOZKlyj6EhJj2cN8/GMekGVHcvxBt8czgYKjBNj2C5zKI3JQw8v3A2B0MFuh+6pl9mvYoTnfpfq04Og+EcDBUq4ipMTdig6lRUaDkURnMwVGzWcBi0i7Iprtt1d9LWhMOkB/Ep2qN5r0QLVXGcQVEGhYU1n3sMNcg75zBKNxya0nvo7Rlct+vu0noKVc6DLGdjg0HSg8DHgRMR8adZ2yrg28CVwBHgUxHxpiQB/wxsAX4N/GVEPF9O6QbpBtKEoGjq8MHyyzOU+AZwS1/bTmBfRGwA9mW/A9wKbMh+5gB/QlqqjoDphtzmg7cN7BF028s8LNo6xgZDRDwF/LKveSuwO7u/G9jW0/5QdOwHLpG0tqhibenqDYX+tjI5HAabdo7h8og4BhARxyRdlrWvA17vWW4xazs2fYm23Hk+oXpFTz5qQFsMXFCaozPc4HzeU3AZy1MT5heK1H0//e+r6KDwrsvUtMFwXNLarLewFjiRtS8C63uWuwI4OugJImIemAe4SKsGhoeNtlSCoK7rM9hw0x7HsAfYnt3fDjzR036nOjYDp7pDDivWcg2FrqLPsvRcw9nGBoOkh4GngT+WtCjpLuArwM2SXgVuzn4H2Au8BiwA9wN/U0rVLVfE36Is8xuzqm/jca+Tp46iwsFDibONHUpExB1D/ummAcsGsGPWomy4qnoKk4RD2QczWfV8SLTNZJJQuG7X3WcdIu0Nv7l8SHTL1H0o9KRXcRp3roQPcW4m9xgqVsRp1235pq3qBKq2rI82cTC0zKBDgqc9VHjYYcdtNGs4eK/E2TyUaLlBIZFnIxl0CHL3cb1/jq7oocMkcxJVniDmvRJnczC0WNHf7oOerxsS4wKiyEComkMh5WBoqVGhUMYEZf8fte0GxazzCJOEW90Tr8uJ5xhsKnn++nXXsN5CW+czlgP3GCpWxAVhm2JcODR5+NDLJ1GlHAxL1Czd7e4wYZJeQV5l9hJ8clVxHAwVWio9hUmUsaF297z074FxKBTHcwwVKOKkqUk1YSOp4kSvJrzPpcg9hpItx15CFX/8tkieX0i5x1CiJodCmw6tLosDYTgHgy1rDofBHAwlunjLQmM+eMPOsRilyCHBsN7JrL2W5d7rKYuDoQJ1hUNdl2TvN27jrXrjblJgN5WDoSJN+zDmCYiij2PoD4A6vu2b9H/QZA6GijXhg+ldfDaOd1cucbOEwIptb5Tea7Bmco/BRppkAnLFtjcafQxDE3prbeEeQw3adiJVnp7DoECooneQ51RsB8LkHAw16X5Y2xIQRQwr6rqeQv86dlCM52Cwynnys/k8x2BLwiRh05ZeWp3cY7BKNLGXcHLuhqRt9fzTNVTSPA6GGvmbqx5nHl/DybniL0KzlDgYatDGQCjjak7WXJ5jsEK0JTjaUmfd3GOwUjVxbsHGc4+hBm3cjz7qiMamH/HY5d5Cfu4x1KRtRz9COedO1Ml7IIZzj8GWjd5ejUNhNPcYbGK9G9WgYwGs/cb2GCQ9KOmEpBd72nZJ+oWkg9nPlp5/+4KkBUmvSPpYWYUvBW2ca+j/ph01t+CJx/bKM5T4BnDLgPavRcTG7GcvgKRrgNuBD2SP+VdJ5xRV7FLUpnAYVGvTrkzl6z0UY+xQIiKeknRlzufbCjwSEaeBn0laAK4HPKAboa6JyCI36CZMpo4LhaU0cVq2WSYf75F0KBtqXJq1rQNe71lmMWtLSJqTdEDSgTOcnqEMa4Mm9Sw88TjetJOP9wF/D0R2+1XgrwENWDYGPUFEzAPzABdp1cBlrF3ybPhF9Cw8XCjfVD2GiDgeEe9ExG+B++kMF6DTQ1jfs+gVwNHZSjSbjINjdlP1GCStjYhj2a+fALp7LPYA35J0L/A+YAPwo5mrNMvBe0GKMzYYJD0MfARYLWkR+DvgI5I20hkmHAE+AxARL0l6FHgZeBvYERHvlFO6WcckgeD5hXzy7JW4Y0DzAyOW/zLw5VmKMsvLvYRy+JBoa51Bf3rPiuVDopexU3uvbswuxElNGgqd99nO91oH9xiscm0No+XEwdAQ3ljK43U7OQfDMuYNxoZxMDRId0P1Blscr8vpOBgaxh/k4nhdTs/B0FBlf6jr3GiqOAvToTAbB0ODlXVGYt0bTVmv76FYcXwcQwvk+aDn+RZu0gZT1vUbmvQe28zBsER0N4hhG1sTN5hhNdV9wRdzMCw5/d/ETQyEcfprHhcUbXyPTedgWIKW2oay1N5PG3jy0cwSDgYzSzgYzCzhYDCzhIPBzBIOBjNLOBjMLOFgMLOEg8HMEg4GM0s4GMws4WAws4SDwcwSDgYzSzgYzCzhYDCzhIPBzBIOBjNLOBjMLOFgMLPE2GCQtF7SDyUdlvSSpM9m7askPSnp1ez20qxdkr4uaUHSIUnXlv0mzKxYeXoMbwOfj4g/ATYDOyRdA+wE9kXEBmBf9jvArcCG7GcOuK/wqs2sVGODISKORcTz2f23gMPAOmArsDtbbDewLbu/FXgoOvYDl0haW3jlZlaaieYYJF0JfBB4Brg8Io5BJzyAy7LF1gGv9zxsMWszs5bIHQyS3gt8B/hcRPxq1KID2mLA881JOiDpwBlO5y3DzCqQKxgkraATCt+MiO9mzce7Q4Ts9kTWvgis73n4FcDR/ueMiPmI2BQRm1awctr6zawEefZKCHgAOBwR9/b80x5ge3Z/O/BET/ud2d6JzcCp7pDDzNohz9+uvBH4NPCCpINZ2xeBrwCPSroL+Dnwyezf9gJbgAXg18BfFVqxmZVubDBExH8xeN4A4KYBywewY8a6zKxGPvLRzBIOBjNLOBjMLOFgMLOEg8HMEg4GM0s4GMws4WAws4SDwcwSDgYzSzgYzCzhYDCzhIPBzBIOBjNLOBjMLOFgMLOEg8HMEg4GM0s4GMws4WAws4SDwcwSDgYzSzgYzCzhYDCzhIPBzBIOBjNLOBjMLOFgMLOEg8HMEg4GM0s4GMws4WAws4SDwcwSDgYzS4wNBknrJf1Q0mFJL0n6bNa+S9IvJB3Mfrb0POYLkhYkvSLpY2W+ATMr3rk5lnkb+HxEPC/pQuA5SU9m//a1iPin3oUlXQPcDnwAeB/wA0l/FBHvFFm4mZVnbI8hIo5FxPPZ/beAw8C6EQ/ZCjwSEacj4mfAAnB9EcWaWTUmmmOQdCXwQeCZrOkeSYckPSjp0qxtHfB6z8MWGRAkkuYkHZB04AynJy7czMqTOxgkvRf4DvC5iPgVcB/wh8BG4Bjw1e6iAx4eSUPEfERsiohNK1g5ceFmVp5cwSBpBZ1Q+GZEfBcgIo5HxDsR8Vvgfn43XFgE1vc8/ArgaHElm1nZ8uyVEPAAcDgi7u1pX9uz2CeAF7P7e4DbJa2UdBWwAfhRcSWbWdny7JW4Efg08IKkg1nbF4E7JG2kM0w4AnwGICJekvQo8DKdPRo7vEfCrF0UkQz/qy9CegP4P+Bk3bXksJp21AntqdV1Fm9QrX8QEWvyPLgRwQAg6UBEbKq7jnHaUie0p1bXWbxZa/Uh0WaWcDCYWaJJwTBfdwE5taVOaE+trrN4M9XamDkGM2uOJvUYzKwhag8GSbdkp2cvSNpZdz39JB2R9EJ2avmBrG2VpCclvZrdXjrueUqo60FJJyS92NM2sC51fD1bx4ckXduAWht32v6ISww0ar1WcimEiKjtBzgH+CnwfuA84MfANXXWNKDGI8DqvrZ/BHZm93cC/1BDXR8GrgVeHFcXsAX4dzrnsWwGnmlArbuAvx2w7DXZ52AlcFX2+TinojrXAtdm9y8EfpLV06j1OqLOwtZp3T2G64GFiHgtIn4DPELntO2m2wrszu7vBrZVXUBEPAX8sq95WF1bgYeiYz9wSd8h7aUaUuswtZ22H8MvMdCo9TqizmEmXqd1B0OuU7RrFsD3JT0naS5ruzwijkHnPwm4rLbqzjasrqau56lP2y9b3yUGGrtei7wUQq+6gyHXKdo1uzEirgVuBXZI+nDdBU2hiet5ptP2yzTgEgNDFx3QVlmtRV8KoVfdwdD4U7Qj4mh2ewL4Hp0u2PFulzG7PVFfhWcZVlfj1nM09LT9QZcYoIHrtexLIdQdDM8CGyRdJek8OteK3FNzTe+SdEF2nUskXQB8lM7p5XuA7dli24En6qkwMayuPcCd2Sz6ZuBUt2tclyaetj/sEgM0bL0Oq7PQdVrFLOqYGdYtdGZVfwp8qe56+mp7P53Z3B8DL3XrA34f2Ae8mt2uqqG2h+l0F8/Q+Ua4a1hddLqS/5Kt4xeATQ2o9d+yWg5lH9y1Pct/Kav1FeDWCuv8czpd7EPAwexnS9PW64g6C1unPvLRzBJ1DyXMrIEcDGaWcDCYWcLBYGYJB4OZJRwMZpZwMJhZwsFgZon/B5JwFDdo6TyQAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "n_vertex = 2000\n",
    "n_class = 4\n",
    "f_label = np.zeros(n_vertex)\n",
    "\n",
    "for i in range(segments.max()+1):\n",
    "    h = int(np.where(segments==i)[0].mean())\n",
    "    w = int(np.where(segments==i)[1].mean())\n",
    "    f_label[i] = label[h, w]\n",
    "\n",
    "label_new = np.zeros((label.shape[0], label.shape[1]))\n",
    "for i in range(segments.max()+1):\n",
    "    label_new[np.where(segments==i)] = f_label[i]\n",
    "plt.imshow(label_new)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 246,
   "metadata": {},
   "outputs": [],
   "source": [
    "def feature_extract(img, segments, f_range=5, n_vertex=256, tao=0.1):\n",
    "    step = math.floor(f_range/2)\n",
    "    n_segments = np.max(segments) + 1\n",
    "    centroid_list = []\n",
    "    for i in range(n_vertex):\n",
    "        if i < n_segments:\n",
    "            h_mean = int(np.where(segments==i)[0].mean())\n",
    "            w_mean = int(np.where(segments==i)[1].mean())\n",
    "            centroid_list.append((h_mean, w_mean))\n",
    "        else:\n",
    "            h_mean = int(np.where(segments==0)[0].mean())\n",
    "            w_mean = int(np.where(segments==0)[1].mean())\n",
    "            centroid_list.append((h_mean, w_mean))\n",
    "    \n",
    "    features = np.zeros((n_vertex, f_range*f_range))\n",
    "    for i in range(n_vertex):\n",
    "        h, w = centroid_list[i]\n",
    "        features[i] = img[h-step:h+step+1, w-step:w+step+1, 0].flatten() \n",
    "        \n",
    "    aj_arr = np.zeros((n_vertex, n_vertex))\n",
    "    for i in range(n_vertex):\n",
    "        for j in range(n_vertex):\n",
    "            diff = features[i] - features[j]\n",
    "            diff = np.power(diff, 2)\n",
    "#             for k in range(diff.shape[0]):\n",
    "#                 diff[k] = math.pow(diff[k], 2)\n",
    "            l2_norm_square = diff.sum()   \n",
    "            aj_arr[i, j] = math.exp(-l2_norm_square / (2*tao*tao))\n",
    "    \n",
    "    return features, aj_arr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 247,
   "metadata": {},
   "outputs": [],
   "source": [
    "def back_to_label(segments, class_output):\n",
    "    output = np.zeros((segments.shape[0], segments.shape[1]))\n",
    "    for i in range(class_output.shape[0]):\n",
    "        output[np.where(segments==i)] = class_output[i]\n",
    "    return output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 260,
   "metadata": {},
   "outputs": [],
   "source": [
    "f, a = feature_extract(img, segments, 7, 300, 7)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 259,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f911d5d6898>"
      ]
     },
     "execution_count": 259,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "class_output = np.asarray([3]*50 + [2]*30 + [1]*70)\n",
    "output = back_to_label(segments, class_output)\n",
    "plt.imshow(output)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 203,
   "metadata": {},
   "outputs": [],
   "source": [
    "def vertex_set(img, segments, filter_size, num):\n",
    "    step = math.floor(filter_size/2)\n",
    "    num_of_segments = np.max(segments) + 1\n",
    "    centroid_list = []\n",
    "    for i in range(num_of_segments):\n",
    "        h_mean = int(np.where(segments==i)[0].mean())\n",
    "        w_mean = int(np.where(segments==i)[1].mean())\n",
    "        val_mean = img[h_mean-step:h_mean+step+1, w_mean-step:w_mean+step+1].mean()\n",
    "        centroid_list.append((h_mean, w_mean, val_mean))\n",
    "        centroid_list.sort(key = itemgetter(2), reverse=True)\n",
    "    \n",
    "    features = np.zeros((num, filter_size*filter_size))\n",
    "    for i in range(num):\n",
    "        h, w, v = centroid_list[i]\n",
    "        features[i] = img[h-step:h+step+1, w-step:w+step+1, 0].flatten() \n",
    "    \n",
    "    return features, centroid_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 204,
   "metadata": {},
   "outputs": [],
   "source": [
    "def adjacency_array(features, tao):\n",
    "    n = features.shape[0]\n",
    "    aj_arr = np.zeros((n, n))\n",
    "    for i in range(n):\n",
    "        for j in range(n):\n",
    "            diff = features[i] - features[j]\n",
    "            for k in range(diff.shape[0]):\n",
    "                diff[k] = math.pow(diff[k], 2)\n",
    "            l2_norm_square = diff.sum()   \n",
    "            aj_arr[i, j] = math.exp(-l2_norm_square / (2*tao*tao))\n",
    "    return aj_arr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 205,
   "metadata": {},
   "outputs": [],
   "source": [
    "features, centroid_list = vertex_set(img, segments, 7, 150)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 206,
   "metadata": {},
   "outputs": [],
   "source": [
    "aj_arr = adjacency_array(features, 0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 207,
   "metadata": {},
   "outputs": [],
   "source": [
    "def back_to_label(segments, class_output, centroid_list):\n",
    "    output = np.zeros((segments.shape[0], segments.shape[1]))\n",
    "    for i in range(class_output.shape[0]):\n",
    "        h = centroid_list[i][0]\n",
    "        w = centroid_list[i][1]\n",
    "        lab_num = segments[h, w]\n",
    "        output[np.where(segments==lab_num)] = class_output[i]\n",
    "    return output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 208,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f911d932eb8>"
      ]
     },
     "execution_count": 208,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "class_output = np.asarray([3]*50 + [2]*30 + [1]*70)\n",
    "output = back_to_label(segments, class_output, centroid_list)\n",
    "plt.imshow(output)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [],
   "source": [
    "info = np.zeros((img.shape[0], img.shape[1], 3))\n",
    "n = 7\n",
    "step = math.floor(n/2)\n",
    "# for c in centroid_list:\n",
    "#     for i in range(c[0]-3, c[0]+3+1):\n",
    "#         for j in range(c[1]-3, c[1]+3+1):\n",
    "#             info[i, j, 0] = img[i, j, 0] \n",
    "#             info[i, j, 1] = img[i, j, 1] \n",
    "#             info[i, j, 2] = img[i, j, 2] \n",
    "\n",
    "for c in centroid_list:\n",
    "    info[c[0]-step:c[0]+step+1, c[1]-step:c[1]+step+1] = img[c[0]-step:c[0]+step+1, c[1]-step:c[1]+step+1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f911dc862b0>"
      ]
     },
     "execution_count": 134,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(info)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 538,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "image = nib.load('/home/tony/Documents/IBSR_preprocessed_2d/IBSR_01/image_80.nii.gz').get_data().astype(np.float32)\n",
    "label = nib.load('/home/tony/Documents/IBSR_preprocessed_2d/IBSR_01/label_80.nii.gz').get_data().astype(np.int64)\n",
    "plt.subplot(121)\n",
    "plt.imshow(image)\n",
    "plt.subplot(122)\n",
    "plt.imshow(label)\n",
    "image = max_min_normalize(image)\n",
    "image = np.expand_dims(image, axis=2)\n",
    "\n",
    "image = np.concatenate((image, image, image), 2)\n",
    "segments = slic(image.astype('double'), n_segments=2000, compactness=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 539,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f9118b3b518>"
      ]
     },
     "execution_count": 539,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(mark_boundaries(image, segments))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 540,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f9118e1e780>"
      ]
     },
     "execution_count": 540,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# transform label to feature\n",
    "n_vertex = 2048\n",
    "n_class = 4\n",
    "f_label = np.zeros(n_vertex)\n",
    "\n",
    "for i in range(segments.max()+1):\n",
    "    h = int(np.where(segments==i)[0].mean())\n",
    "    w = int(np.where(segments==i)[1].mean())\n",
    "    f_label[i] = label[h, w]\n",
    "    \n",
    "# for i in range(segments.max()+1):\n",
    "#     f_label[i] = round(label[np.where(segments==i)].mean())\n",
    "    \n",
    "plt.imshow(mark_boundaries(max_min_normalize(label), segments))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 541,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f9118ef7128>"
      ]
     },
     "execution_count": 541,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# turn back feature to label\n",
    "label_new = np.zeros((label.shape[0], label.shape[1]))\n",
    "for i in range(segments.max()+1):\n",
    "    label_new[np.where(segments==i)] = f_label[i]\n",
    "plt.imshow(label_new)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 503,
   "metadata": {},
   "outputs": [],
   "source": [
    "def label_transform(label, segments):\n",
    "    #f_label = np.zeros(n_vertex)\n",
    "    label_new = np.zeros((label.shape[0], label.shape[1]))\n",
    "    for i in range(segments.max() + 1):\n",
    "        where = np.where(segments==i)\n",
    "        h = int(where[0].mean())\n",
    "        w = int(where[1].mean())\n",
    "        label_new[np.where(segments==i)]  = label[h, w] \n",
    "    return label_new"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 504,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f9119257b38>"
      ]
     },
     "execution_count": 504,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "label_new = label_transform(label, segments)\n",
    "plt.imshow(label_new)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 405,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[0, 0],\n",
       "        [0, 0]])"
      ]
     },
     "execution_count": 405,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import torch\n",
    "a = torch.zeros(2, 2).long()\n",
    "a"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 451,
   "metadata": {},
   "outputs": [],
   "source": [
    "output = torch.LongTensor(f_label).cuda()\n",
    "target= torch.LongTensor(label).cuda()\n",
    "seg = torch.LongTensor(segments).cuda()\n",
    "\n",
    "img = torch.zeros_like(target).long()\n",
    "\n",
    "mask0 = torch.zeros_like(target)\n",
    "mask1 = torch.ones_like(target)\n",
    "\n",
    "# for i in range(seg_tensor.max() + 1):\n",
    "#     seg_tensor = np.asarray(seg_tensor.cpu())\n",
    "#     where = np.where(seg_tensor==i)\n",
    "#     h = int(where[0].mean())\n",
    "#     w = int(where[1].mean())\n",
    "#     seg_tensor = torch.LongTensor(seg_tensor).cuda()\n",
    "#     img += torch.where(seg_tensor==i, label_tensor[h, w]*mask1, mask0)\n",
    "\n",
    "for i in range(seg_tensor.max() + 1):\n",
    "    img += torch.where(seg_tensor==i, output[i]*mask1, mask0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 433,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f91194f9748>"
      ]
     },
     "execution_count": 433,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "img = np.asarray(img.cpu())\n",
    "plt.imshow(img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 464,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[0.7041, 0.7984, 0.1155, 0.7155, 0.4653, 0.4735, 0.1769],\n",
       "        [0.9559, 0.2194, 0.7668, 0.0571, 0.4287, 0.5186, 0.4793],\n",
       "        [0.5714, 0.4119, 0.4528, 0.4247, 0.0440, 0.8828, 0.8793]])"
      ]
     },
     "execution_count": 464,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x = torch.rand(3, 7)\n",
    "x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 493,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[1., 0., 0., 1.],\n",
       "        [1., 1., 1., 0.],\n",
       "        [0., 1., 1., 1.],\n",
       "        [0., 1., 0., 1.]])"
      ]
     },
     "execution_count": 493,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "a = torch.FloatTensor([[0., 0., 0., 1.],\n",
    "           [1., 0., 1., 0.],\n",
    "           [0., 1., 0., 1.],\n",
    "           [0., 1., 0., 0.]])\n",
    "i = torch.eye(a.size(0))\n",
    "a_hat = a+i\n",
    "a_hat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 494,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[2., 0., 0., 0.],\n",
       "        [0., 3., 0., 0.],\n",
       "        [0., 0., 3., 0.],\n",
       "        [0., 0., 0., 2.]])"
      ]
     },
     "execution_count": 494,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "d = torch.diag(a_hat.sum(1))\n",
    "d"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 495,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[0.5000, 0.0000, 0.0000, -0.0000],\n",
       "        [0.0000, 0.3333, 0.0000, -0.0000],\n",
       "        [0.0000, 0.0000, 0.3333, -0.0000],\n",
       "        [0.0000, 0.0000, 0.0000, 0.5000]])"
      ]
     },
     "execution_count": 495,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "torch.inverse(d)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 500,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[0.5000, 0.0000, 0.0000, 0.5000],\n",
       "        [0.3333, 0.3333, 0.3333, 0.0000],\n",
       "        [0.0000, 0.3333, 0.3333, 0.3333],\n",
       "        [0.0000, 0.5000, 0.0000, 0.5000]])"
      ]
     },
     "execution_count": 500,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "torch.mm(torch.inverse(d), a_hat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 477,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[1.5103],\n",
       "        [3.0652],\n",
       "        [2.9216]])"
      ]
     },
     "execution_count": 477,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(a.sum(1, keepdim=True) + 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 476,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[1.5103, 0.0000, 0.0000],\n",
       "        [0.0000, 3.0652, 0.0000],\n",
       "        [0.0000, 0.0000, 2.9216]])"
      ]
     },
     "execution_count": 476,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "e = torch.eye(3) * (a.sum(1, keepdim=True) + 1)\n",
    "e"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 502,
   "metadata": {},
   "outputs": [],
   "source": [
    "i = torch.eye(a.size(0))                                                                                                                                                             \n",
    "a_hat = a + i\n",
    "d = torch.diag(a_hat.sum(1))\n",
    "adj_arr = torch.mm(torch.inverse(d), a_hat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5\n"
     ]
    }
   ],
   "source": [
    "a = 5\n",
    "def test_func(b):\n",
    "    b += 1\n",
    "    return 1\n",
    "c = test_func(a)\n",
    "print(a)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([5, 6, 7, 8, 9])"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.arange(5, 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "torchenv",
   "language": "python",
   "name": "torchenv"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}