U-Net MXNet.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Nerve - Segmentation w/ U-Net"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "kenstler@\n",
    "\n",
    "Ultrasound nerve-segmentation using U-net architecture for pixel-wise classification. Adapted from Keras implementation [here](https://github.com/jocicmarko/ultrasound-nerve-segmentation). Follow instructions on that repo to prepare the data using the `data.py` script, and keep it in the same directory as this nb. We will use methods from said script to load data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import mxnet as mx\n",
    "from mxnet import ndarray as F\n",
    "\n",
    "from __future__ import print_function\n",
    "\n",
    "import os\n",
    "from skimage.transform import resize\n",
    "from skimage.io import imsave\n",
    "from sklearn.cross_validation import train_test_split\n",
    "import numpy as np\n",
    "from data import load_train_data, load_test_data\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ultrasound images and corresponding masks are loaded here."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_size = 32\n",
    "img_rows = 96\n",
    "img_cols = 96"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_shape = (batch_size, 1, img_rows, img_cols)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_X, train_Y = load_train_data()\n",
    "train_X, valid_X, train_Y, valid_Y = train_test_split(train_X, train_Y, test_size=0.2, random_state=1984)\n",
    "test_X, test_Y = load_test_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "_, class_counts = np.unique(train_Y, return_counts=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4508, 420, 580)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_X.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4508, 420, 580)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_Y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f10dc87d390>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAVgAAAD8CAYAAAAylrwMAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f112c8d0910>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(train_X[6], cmap=plt.cm.gray)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f10bcf7fbd0>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAVgAAAD8CAYAAAAylrwMAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nS/pcnehWbncAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f112c8ac750>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(train_X[6]*train_Y[6], cmap=plt.cm.gray)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Preprocessing includes mean-centering, normalization, and resizing original images to 96x96."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_mean = np.mean(train_X) \n",
    "train_std = np.std(train_X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "def preprocess(imgs, mean = train_mean, std = train_std, test=False):\n",
    "    imgs_p = np.ndarray((imgs.shape[0], img_rows, img_cols), dtype=np.uint8)\n",
    "    for i in range(imgs.shape[0]):\n",
    "        imgs_p[i] = resize(imgs[i], (img_cols, img_rows), preserve_range=True)\n",
    "\n",
    "    imgs_p = np.expand_dims(imgs_p, axis=1)\n",
    "    if test:\n",
    "        imgs_p = imgs_p / 255.\n",
    "        imgs_p = np.round(imgs_p).astype('uint8')\n",
    "    else:\n",
    "        imgs_p = imgs_p.astype('float32')\n",
    "        imgs_p -= mean\n",
    "        imgs_p /= std\n",
    "    return imgs_p"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We're also going to flatten our masks."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/ubuntu/src/anaconda2/envs/mxnet1/lib/python2.7/site-packages/skimage/transform/_warps.py:84: UserWarning: The default mode, 'constant', will be changed to 'reflect' in skimage 0.15.\n",
      "  warn(\"The default mode, 'constant', will be changed to 'reflect' in \"\n"
     ]
    }
   ],
   "source": [
    "train_X = preprocess(train_X)\n",
    "train_Y = preprocess(train_Y, test=True)\n",
    "train_Y = train_Y.reshape((len(train_Y), img_rows*img_cols))\n",
    "valid_X = preprocess(valid_X)\n",
    "valid_Y = preprocess(valid_Y, test=True)\n",
    "valid_Y = valid_Y.reshape((len(valid_Y), img_rows*img_cols))\n",
    "test_X = preprocess(test_X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "def post_process(imgs, mean = train_mean, std = train_std):\n",
    "    imgs_p = imgs.copy()\n",
    "    imgs_p *= std\n",
    "    imgs_p += mean\n",
    "    imgs_p = imgs_p.astype('uint8')\n",
    "    if len(imgs_p.shape)==4:\n",
    "        return imgs_p[:,0,:,:]\n",
    "    else:\n",
    "        return imgs_p[0, :, :]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f112c878190>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP4AAAD8CAYAAABXXhlaAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n4Vep0odSf/hVqvSh1B9+lSp9KP8fS4fWfCywwH8AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f10bcfe4e10>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(post_process(train_X[1]), cmap=plt.cm.gray)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f112c7e77d0>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP4AAAD8CAYAAABXXhlaAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f10bcf3f4d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(post_process(train_X[1])*(train_Y[1].reshape(img_cols, img_rows)).astype('uint8'), cmap=plt.cm.gray)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will be using the dice coefficient as our loss by optimizing it's negative."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def dice_coef(y_true, y_pred):\n",
    "    intersection = mx.sym.sum(mx.sym.broadcast_mul(y_true, y_pred), axis=(1, 2, 3))\n",
    "    return mx.sym.broadcast_div((2. * intersection + 1.),(mx.sym.sum(y_true, axis=(1, 2, 3)) + mx.sym.sum(y_pred, axis=(1, 2, 3)) + 1.))\n",
    "\n",
    "\n",
    "def dice_coef_loss(y_true, y_pred):\n",
    "    intersection = mx.sym.sum(mx.sym.broadcast_mul(y_true, y_pred), axis=1, )\n",
    "    return -mx.sym.broadcast_div((2. * intersection + 1.),(mx.sym.broadcast_add(mx.sym.sum(y_true, axis=1), mx.sym.sum(y_pred, axis=1)) + 1.))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Unet architecture w/ batch normalization."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_unet():\n",
    "    data = mx.sym.Variable(name='data')\n",
    "    label = mx.sym.Variable(name='label')\n",
    "    conv1 = mx.sym.Convolution(data, num_filter=32, kernel=(3,3), pad=(1,1), name='conv1_1')\n",
    "    conv1 = mx.sym.BatchNorm(conv1, name='bn1_1')\n",
    "    conv1 = mx.sym.Activation(conv1, act_type='relu', name='relu1_1')\n",
    "    conv1 = mx.sym.Convolution(conv1, num_filter=32, kernel=(3,3), pad=(1,1), name='conv1_2')\n",
    "    conv1 = mx.sym.BatchNorm(conv1, name='bn1_2')\n",
    "    conv1 = mx.sym.Activation(conv1, act_type='relu', name='relu1_2')\n",
    "    pool1 = mx.sym.Pooling(conv1, kernel=(2,2), pool_type='max', name='pool1')\n",
    "    \n",
    "    conv2 = mx.sym.Convolution(pool1, num_filter=64, kernel=(3,3), pad=(1,1), name='conv2_1')\n",
    "    conv2 = mx.sym.BatchNorm(conv2, name='bn2_1')\n",
    "    conv2 = mx.sym.Activation(conv2, act_type='relu', name='relu2_1')\n",
    "    conv2 = mx.sym.Convolution(conv2, num_filter=64, kernel=(3,3), pad=(1,1), name='conv2_2')\n",
    "    conv2 = mx.sym.BatchNorm(conv2, name='bn2_2')\n",
    "    conv2 = mx.sym.Activation(conv2, act_type='relu', name='relu2_2')\n",
    "    pool2 = mx.sym.Pooling(conv2, kernel=(2,2), pool_type='max', name='pool2')\n",
    "    \n",
    "    conv3 = mx.sym.Convolution(pool2, num_filter=128, kernel=(3,3), pad=(1,1), name='conv3_1')\n",
    "    conv3 = mx.sym.BatchNorm(conv3, name='bn3_1')\n",
    "    conv3 = mx.sym.Activation(conv3, act_type='relu', name='relu3_1')\n",
    "    conv3 = mx.sym.Convolution(conv3, num_filter=128, kernel=(3,3), pad=(1,1), name='conv3_2')\n",
    "    conv3 = mx.sym.BatchNorm(conv3, name='bn3_2')\n",
    "    conv3 = mx.sym.Activation(conv3, act_type='relu', name='relu3_2')\n",
    "    pool3 = mx.sym.Pooling(conv3, kernel=(2,2), pool_type='max', name='pool3')\n",
    "    \n",
    "    conv4 = mx.sym.Convolution(pool3, num_filter=256, kernel=(3,3), pad=(1,1), name='conv4_1')\n",
    "    conv4 = mx.sym.BatchNorm(conv4, name='bn4_1')\n",
    "    conv4 = mx.sym.Activation(conv4, act_type='relu', name='relu4_1')\n",
    "    conv4 = mx.sym.Convolution(conv4, num_filter=256, kernel=(3,3), pad=(1,1), name='conv4_2')\n",
    "    conv4 = mx.sym.BatchNorm(conv4, name='bn4_2')\n",
    "    conv4 = mx.sym.Activation(conv4, act_type='relu', name='relu4_2')\n",
    "    pool4 = mx.sym.Pooling(conv4, kernel=(2,2), pool_type='max', name='pool4')\n",
    "    \n",
    "    conv5 = mx.sym.Convolution(pool4, num_filter=512, kernel=(3,3), pad=(1,1), name='conv5_1')\n",
    "    conv5 = mx.sym.BatchNorm(conv5, name='bn5_1')\n",
    "    conv5 = mx.sym.Activation(conv5, act_type='relu', name='relu5_1')\n",
    "    conv5 = mx.sym.Convolution(conv5, num_filter=512, kernel=(3,3), pad=(1,1), name='conv5_2')\n",
    "    conv5 = mx.sym.BatchNorm(conv5, name='bn5_2')\n",
    "    conv5 = mx.sym.Activation(conv5, act_type='relu', name='relu5_2')\n",
    "    \n",
    "    trans_conv6 = mx.sym.Deconvolution(conv5, num_filter=256, kernel=(2,2), stride=(1,1), no_bias=True, name='trans_conv6')\n",
    "    up6 = mx.sym.concat(*[trans_conv6, conv4], dim=1, name='concat6')\n",
    "    conv6 = mx.sym.Convolution(up6, num_filter=256, kernel=(3,3), pad=(1,1), name='conv6_1')\n",
    "    conv6 = mx.sym.BatchNorm(conv6, name='bn6_1')\n",
    "    conv6 = mx.sym.Activation(conv6, act_type='relu', name='relu6_1')\n",
    "    conv6 = mx.sym.Convolution(conv6, num_filter=256, kernel=(3,3), pad=(1,1), name='conv6_2')\n",
    "    conv6 = mx.sym.BatchNorm(conv6, name='bn6_2')\n",
    "    conv6 = mx.sym.Activation(conv6, act_type='relu', name='relu6_2')\n",
    "    \n",
    "    trans_conv7 = mx.sym.Deconvolution(conv6, num_filter=128, kernel=(2,2), stride=(1,1), no_bias=True, name='trans_conv7')\n",
    "    up7 = mx.sym.concat(*[trans_conv7, conv3], dim=1, name='concat7')\n",
    "    conv7 = mx.sym.Convolution(up7, num_filter=128, kernel=(3,3), pad=(1,1), name='conv7_1')\n",
    "    conv7 = mx.sym.BatchNorm(conv7, name='bn7_1')\n",
    "    conv7 = mx.sym.Activation(conv7, act_type='relu', name='relu7_1')\n",
    "    conv7 = mx.sym.Convolution(conv7, num_filter=128, kernel=(3,3), pad=(1,1), name='conv7_2')\n",
    "    conv7 = mx.sym.BatchNorm(conv7, name='bn7_2')\n",
    "    conv7 = mx.sym.Activation(conv7, act_type='relu', name='relu7_2')\n",
    "    \n",
    "    trans_conv8 = mx.sym.Deconvolution(conv7, num_filter=64, kernel=(2,2), stride=(1,1), no_bias=True, name='trans_conv8')\n",
    "    up8 = mx.sym.concat(*[trans_conv8, conv2], dim=1, name='concat8')\n",
    "    conv8 = mx.sym.Convolution(up8, num_filter=64, kernel=(3,3), pad=(1,1), name='conv8_1')\n",
    "    conv8 = mx.sym.BatchNorm(conv8, name='bn8_1')\n",
    "    conv8 = mx.sym.Activation(conv8, act_type='relu', name='relu8_1')\n",
    "    conv8 = mx.sym.Convolution(conv8, num_filter=64, kernel=(3,3), pad=(1,1), name='conv8_2')\n",
    "    conv8 = mx.sym.BatchNorm(conv8, name='bn8_2')\n",
    "    conv8 = mx.sym.Activation(conv8, act_type='relu', name='relu8_2')\n",
    "    \n",
    "    trans_conv9 = mx.sym.Deconvolution(conv8, num_filter=32, kernel=(2,2), stride=(1,1), no_bias=True, name='trans_conv9')\n",
    "    up9 = mx.sym.concat(*[trans_conv9, conv1], dim=1, name='concat9')\n",
    "    conv9 = mx.sym.Convolution(up9, num_filter=32, kernel=(3,3), pad=(1,1), name='conv9_1')\n",
    "    conv9 = mx.sym.BatchNorm(conv9, name='bn9_1')\n",
    "    conv9 = mx.sym.Activation(conv9, act_type='relu', name='relu9_1')\n",
    "    conv9 = mx.sym.Convolution(conv9, num_filter=32, kernel=(3,3), pad=(1,1), name='conv9_2')\n",
    "    conv9 = mx.sym.BatchNorm(conv9, name='bn9_2')\n",
    "    conv9 = mx.sym.Activation(conv9, act_type='relu', name='relu9_2')\n",
    "    \n",
    "    conv10 = mx.sym.Convolution(conv9, num_filter=1, kernel=(1,1), name='conv10_1')\n",
    "    conv10 = mx.sym.sigmoid(conv10, name='softmax')\n",
    "    \n",
    "    net = mx.sym.Flatten(conv10)\n",
    "    loss = mx.sym.MakeLoss(dice_coef_loss(label, net), normalization='batch')\n",
    "    mask_output = mx.sym.BlockGrad(conv10, 'mask')\n",
    "    out = mx.sym.Group([loss, mask_output])\n",
    "#     return mx.sym.Custom(net, pos_grad_scale = pos, neg_grad_scale = neg, name = 'softmax', op_type = 'weighted_logistic_regression')\n",
    "#     return mx.sym.LogisticRegressionOutput(net, name='softmax')\n",
    "    return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "ctx = [mx.gpu(i) for i in range(8)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "unet_sym = build_unet()\n",
    "unet = mx.mod.Module(unet_sym, context=ctx,data_names=('data',), label_names=('label',))\n",
    "unet.bind(data_shapes=[['data', data_shape]], label_shapes=[['label', (batch_size, img_cols*img_rows)]])\n",
    "unet.init_params(mx.initializer.Xavier(magnitude=6))\n",
    "\n",
    "unet.init_optimizer(optimizer = 'adam', \n",
    "                               optimizer_params=(\n",
    "                                   ('learning_rate', 1E-4),\n",
    "                                   ('beta1', 0.9),\n",
    "                                   ('beta2', 0.99)\n",
    "                              ))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 332,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_iter = mx.io.NDArrayIter(data = train_X, label=train_Y, batch_size=batch_size, label_name='label', shuffle=True)\n",
    "\n",
    "val_iter = mx.io.NDArrayIter(data = valid_X, label=valid_Y, batch_size=batch_size, label_name='label', shuffle=False)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "import logging\n",
    "logging.getLogger().setLevel(logging.DEBUG)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Training loop. Saves best model on validation loss."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 335,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1: Moving Training Loss -0.08519, Validation Loss -0.13640\n",
      "Epoch 2: Moving Training Loss -0.14094, Validation Loss -0.16745\n",
      "Epoch 3: Moving Training Loss -0.17312, Validation Loss -0.18187\n",
      "Epoch 4: Moving Training Loss -0.19683, Validation Loss -0.19442\n",
      "Epoch 5: Moving Training Loss -0.21657, Validation Loss -0.21293\n",
      "Epoch 6: Moving Training Loss -0.23555, Validation Loss -0.22890\n",
      "Epoch 7: Moving Training Loss -0.25108, Validation Loss -0.23755\n",
      "Epoch 8: Moving Training Loss -0.26500, Validation Loss -0.24491\n",
      "Epoch 9: Moving Training Loss -0.27997, Validation Loss -0.25844\n",
      "Epoch 10: Moving Training Loss -0.29100, Validation Loss -0.26039\n",
      "Epoch 11: Moving Training Loss -0.30139, Validation Loss -0.27129\n",
      "Epoch 12: Moving Training Loss -0.31167, Validation Loss -0.27418\n",
      "Epoch 13: Moving Training Loss -0.32149, Validation Loss -0.27037\n",
      "Epoch 14: Moving Training Loss -0.32888, Validation Loss -0.28313\n",
      "Epoch 15: Moving Training Loss -0.33569, Validation Loss -0.29314\n",
      "Epoch 16: Moving Training Loss -0.34326, Validation Loss -0.29004\n",
      "Epoch 17: Moving Training Loss -0.34832, Validation Loss -0.29716\n",
      "Epoch 18: Moving Training Loss -0.35574, Validation Loss -0.30385\n",
      "Epoch 19: Moving Training Loss -0.36387, Validation Loss -0.30531\n",
      "Epoch 20: Moving Training Loss -0.37251, Validation Loss -0.31384\n",
      "Epoch 21: Moving Training Loss -0.38630, Validation Loss -0.32578\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:Saved checkpoint to \"best_unet-0000.params\"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best model at Epoch 22\n",
      "Epoch 22: Moving Training Loss -0.39848, Validation Loss -0.33716\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:Saved checkpoint to \"best_unet-0000.params\"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best model at Epoch 23\n",
      "Epoch 23: Moving Training Loss -0.41848, Validation Loss -0.33903\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:Saved checkpoint to \"best_unet-0000.params\"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best model at Epoch 24\n",
      "Epoch 24: Moving Training Loss -0.43883, Validation Loss -0.37535\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:Saved checkpoint to \"best_unet-0000.params\"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best model at Epoch 25\n",
      "Epoch 25: Moving Training Loss -0.45622, Validation Loss -0.39797\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:Saved checkpoint to \"best_unet-0000.params\"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best model at Epoch 26\n",
      "Epoch 26: Moving Training Loss -0.47870, Validation Loss -0.44039\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:Saved checkpoint to \"best_unet-0000.params\"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best model at Epoch 27\n",
      "Epoch 27: Moving Training Loss -0.50846, Validation Loss -0.46196\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:Saved checkpoint to \"best_unet-0000.params\"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best model at Epoch 28\n",
      "Epoch 28: Moving Training Loss -0.53855, Validation Loss -0.48380\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:Saved checkpoint to \"best_unet-0000.params\"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best model at Epoch 29\n",
      "Epoch 29: Moving Training Loss -0.56428, Validation Loss -0.51343\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:Saved checkpoint to \"best_unet-0000.params\"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best model at Epoch 30\n",
      "Epoch 30: Moving Training Loss -0.55425, Validation Loss -0.54715\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:Saved checkpoint to \"best_unet-0000.params\"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best model at Epoch 31\n",
      "Epoch 31: Moving Training Loss -0.56531, Validation Loss -0.55644\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:Saved checkpoint to \"best_unet-0000.params\"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best model at Epoch 32\n",
      "Epoch 32: Moving Training Loss -0.59385, Validation Loss -0.56605\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:Saved checkpoint to \"best_unet-0000.params\"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best model at Epoch 33\n",
      "Epoch 33: Moving Training Loss -0.60645, Validation Loss -0.61197\n",
      "Epoch 34: Moving Training Loss -0.62961, Validation Loss -0.57995\n",
      "Epoch 35: Moving Training Loss -0.63495, Validation Loss -0.59262\n",
      "Epoch 36: Moving Training Loss -0.64299, Validation Loss -0.60536\n",
      "Epoch 37: Moving Training Loss -0.65929, Validation Loss -0.60055\n",
      "Epoch 38: Moving Training Loss -0.65807, Validation Loss -0.60428\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:Saved checkpoint to \"best_unet-0000.params\"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best model at Epoch 39\n",
      "Epoch 39: Moving Training Loss -0.65419, Validation Loss -0.65319\n",
      "Epoch 40: Moving Training Loss -0.66291, Validation Loss -0.63549\n",
      "Epoch 41: Moving Training Loss -0.65988, Validation Loss -0.58070\n",
      "Epoch 42: Moving Training Loss -0.60071, Validation Loss -0.58907\n",
      "Epoch 43: Moving Training Loss -0.58932, Validation Loss -0.59185\n",
      "Epoch 44: Moving Training Loss -0.58802, Validation Loss -0.59340\n",
      "Epoch 45: Moving Training Loss -0.58856, Validation Loss -0.59432\n",
      "Epoch 46: Moving Training Loss -0.58921, Validation Loss -0.59490\n",
      "Epoch 47: Moving Training Loss -0.58970, Validation Loss -0.59529\n",
      "Epoch 48: Moving Training Loss -0.59005, Validation Loss -0.59555\n",
      "Epoch 49: Moving Training Loss -0.59030, Validation Loss -0.59574\n",
      "Epoch 50: Moving Training Loss -0.59047, Validation Loss -0.59587\n",
      "Epoch 51: Moving Training Loss -0.59059, Validation Loss -0.59596\n",
      "Epoch 52: Moving Training Loss -0.59068, Validation Loss -0.59603\n",
      "Epoch 53: Moving Training Loss -0.59074, Validation Loss -0.59607\n",
      "Epoch 54: Moving Training Loss -0.59078, Validation Loss -0.59611\n",
      "Epoch 55: Moving Training Loss -0.59081, Validation Loss -0.59613\n",
      "Epoch 56: Moving Training Loss -0.59083, Validation Loss -0.59615\n",
      "Epoch 57: Moving Training Loss -0.59085, Validation Loss -0.59615\n",
      "Epoch 58: Moving Training Loss -0.59086, Validation Loss -0.59617\n",
      "Epoch 59: Moving Training Loss -0.59087, Validation Loss -0.59617\n",
      "Epoch 60: Moving Training Loss -0.59087, Validation Loss -0.59618\n",
      "Epoch 61: Moving Training Loss -0.59088, Validation Loss -0.59618\n",
      "Epoch 62: Moving Training Loss -0.59088, Validation Loss -0.59618\n",
      "Epoch 63: Moving Training Loss -0.59088, Validation Loss -0.59618\n",
      "Epoch 64: Moving Training Loss -0.59088, Validation Loss -0.59618\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-335-0247ed205134>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     15\u001b[0m         \u001b[0mloss\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0munet\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget_outputs\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     16\u001b[0m         \u001b[0munet\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mupdate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 17\u001b[0;31m         \u001b[0mcurr_loss\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mF\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmean\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mloss\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0masscalar\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     18\u001b[0m         \u001b[0mcurr_losses\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mcurr_loss\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     19\u001b[0m         moving_loss = (curr_loss if ((i == 0) and (e == 0))\n",
      "\u001b[0;32m/home/ubuntu/src/anaconda2/envs/mxnet1/lib/python2.7/site-packages/mxnet/ndarray/ndarray.pyc\u001b[0m in \u001b[0;36masscalar\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1809\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m \u001b[0;34m!=\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1810\u001b[0m             \u001b[0;32mraise\u001b[0m \u001b[0mValueError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"The current array is not a scalar\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1811\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0masnumpy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1812\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1813\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mastype\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/ubuntu/src/anaconda2/envs/mxnet1/lib/python2.7/site-packages/mxnet/ndarray/ndarray.pyc\u001b[0m in \u001b[0;36masnumpy\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1791\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1792\u001b[0m             \u001b[0mdata\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mctypes\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdata_as\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mctypes\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mc_void_p\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1793\u001b[0;31m             ctypes.c_size_t(data.size)))\n\u001b[0m\u001b[1;32m   1794\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mdata\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1795\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "epochs = 100\n",
    "smoothing_constant = .01\n",
    "curr_losses = []\n",
    "moving_losses = []\n",
    "i = 0\n",
    "best_val_loss = np.inf\n",
    "for e in range(epochs):\n",
    "    while True:\n",
    "        try:\n",
    "            batch = next(train_iter)\n",
    "        except StopIteration:\n",
    "            train_iter.reset()\n",
    "            break\n",
    "        unet.forward_backward(batch)\n",
    "        loss = unet.get_outputs()[0]\n",
    "        unet.update()\n",
    "        curr_loss = F.mean(loss).asscalar()\n",
    "        curr_losses.append(curr_loss)\n",
    "        moving_loss = (curr_loss if ((i == 0) and (e == 0))\n",
    "                               else (1 - smoothing_constant) * moving_loss + (smoothing_constant) * curr_loss)\n",
    "        moving_losses.append(moving_loss)\n",
    "        i += 1\n",
    "    val_losses = []\n",
    "    for batch in val_iter:\n",
    "        unet.forward(batch)\n",
    "        loss = unet.get_outputs()[0]\n",
    "        val_losses.append(F.mean(loss).asscalar())\n",
    "    val_iter.reset()\n",
    "    val_loss = np.mean(val_losses)\n",
    "    if e > 20 and val_loss < best_val_loss:\n",
    "        best_val_loss = val_loss\n",
    "        unet.save_checkpoint('best_unet', 0)\n",
    "        print(\"Best model at Epoch %i\" %(e+1))\n",
    "    print(\"Epoch %i: Moving Training Loss %0.5f, Validation Loss %0.5f\" % (e+1, moving_loss, val_loss))\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 218,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 21: Moving Training Loss -0.28446, Validation Loss -0.22898\n",
      "Epoch 22: Moving Training Loss -0.28671, Validation Loss -0.22918\n",
      "Epoch 23: Moving Training Loss -0.28803, Validation Loss -0.22628\n",
      "Epoch 24: Moving Training Loss -0.29093, Validation Loss -0.22953\n",
      "Epoch 25: Moving Training Loss -0.29446, Validation Loss -0.22560\n",
      "Epoch 26: Moving Training Loss -0.29743, Validation Loss -0.22602\n",
      "Epoch 27: Moving Training Loss -0.29959, Validation Loss -0.22668\n",
      "Epoch 28: Moving Training Loss -0.30181, Validation Loss -0.22123\n",
      "Epoch 29: Moving Training Loss -0.30410, Validation Loss -0.23067\n",
      "Epoch 30: Moving Training Loss -0.30666, Validation Loss -0.23304\n",
      "Epoch 31: Moving Training Loss -0.30911, Validation Loss -0.23213\n",
      "Epoch 32: Moving Training Loss -0.31153, Validation Loss -0.23487\n",
      "Epoch 33: Moving Training Loss -0.31382, Validation Loss -0.23522\n",
      "Epoch 34: Moving Training Loss -0.31535, Validation Loss -0.23875\n",
      "Epoch 35: Moving Training Loss -0.31678, Validation Loss -0.24137\n",
      "Epoch 36: Moving Training Loss -0.31896, Validation Loss -0.24539\n",
      "Epoch 37: Moving Training Loss -0.32070, Validation Loss -0.24293\n",
      "Epoch 38: Moving Training Loss -0.32226, Validation Loss -0.24461\n",
      "Epoch 39: Moving Training Loss -0.32375, Validation Loss -0.24726\n",
      "Epoch 40: Moving Training Loss -0.32572, Validation Loss -0.24115\n"
     ]
    }
   ],
   "source": [
    "epochs = 20\n",
    "smoothing_constant = .01\n",
    "for e in range(epochs):\n",
    "    while True:\n",
    "        try:\n",
    "            batch = next(train_iter)\n",
    "        except StopIteration:\n",
    "            train_iter.reset()\n",
    "            break\n",
    "        unet.forward_backward(batch)\n",
    "        loss = unet.get_outputs()[0]\n",
    "        unet.update()\n",
    "        curr_loss = F.mean(loss).asscalar()\n",
    "        curr_losses.append(curr_loss)\n",
    "        moving_loss = (curr_loss if ((i == 0) and (e == 0))\n",
    "                               else (1 - smoothing_constant) * moving_loss + (smoothing_constant) * curr_loss)\n",
    "        moving_losses.append(moving_loss)\n",
    "        i += 1\n",
    "    val_losses = []\n",
    "    for batch in val_iter:\n",
    "        unet.forward(batch)\n",
    "        loss = unet.get_outputs()[0]\n",
    "        val_losses.append(F.mean(loss).asscalar())\n",
    "    val_iter.reset()\n",
    "    val_loss = np.mean(val_losses)\n",
    "    print(\"Epoch %i: Moving Training Loss %0.5f, Validation Loss %0.5f\" % (e+21, moving_loss, val_loss))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You'll notice I stopped early. The model will often times end up converging to a bad minima, that is it predicts everything as zero."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 336,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f0d5ab6e4d0>]"
      ]
     },
     "execution_count": 336,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX8AAAD8CAYAAACfF6SlAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n+YuIxCElfxGROPT/AQsjug7b4J9FAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f0d5d7bb850>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(moving_losses)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We've saved the model at the lowest point here, before it plateaus into converges to this bad minima.\n",
    "\n",
    "Let's check this model out. At ~65% dice coefficient, this gets pretty close to the results achieved in the implementation this was adapted from."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "sym, arg_params, aux_params = mx.model.load_checkpoint('best_unet', 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "unet.set_params(arg_params, aux_params)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "unet.save_checkpoint('unet_nerve_segmentation', 39)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 473,
   "metadata": {},
   "outputs": [],
   "source": [
    "def post_process_mask(label, p=0.5):\n",
    "    return (np.where(label.asnumpy().reshape(img_cols, img_rows) > p, 1, 0)).astype('uint8')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 504,
   "metadata": {},
   "outputs": [],
   "source": [
    "batch = next(val_iter)\n",
    "\n",
    "unet.forward(batch)\n",
    "\n",
    "outputs = unet.get_outputs()[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 509,
   "metadata": {},
   "outputs": [],
   "source": [
    "idx = 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "True Label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 510,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f0d2b241d10>"
      ]
     },
     "execution_count": 510,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAsMAAAFTCAYAAADcAgGcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4xLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvAOZPmwAAIABJREFUeJzsvXvUXmV57ntNCCBC1UAOJF9CzueE\nJAiBSECMoHTrQKsoupYOd2vrP3bvtbs6xm5XOzr0D6u7HbtdbUfHoAN3NwM7dOuSpaCIHFqESgzh\nlBMJOX75cj4QCohK1Mjcf5D58Htu3ufNF5K85Mu8fmN09M77zfnM5zRfpu91zfuu6rqWMcYYY4wx\nbeSMN7sDxhhjjDHGvFn4YdgYY4wxxrQWPwwbY4wxxpjW4odhY4wxxhjTWvwwbIwxxhhjWosfho0x\nxhhjTGvxw7AxxhhjjGktx/UwXFXVDVVVbayqaktVVX96ojpljDHmxOPvbGOMeT3VGy26UVXVmZI2\nSbpe0i5Jj0v6ZF3X609c94wxxpwI/J1tjDGdGXYc5y6StKWu635Jqqrqm5I+JKn4xXruuefWb3vb\n2yRJZ5555mudGJZ346yzzup4Ps/5zW9+k+JXXnklxVVVZefwb6XjzjjjjKN+Lkm//vWvOx7H/0HB\nc+L/0OD1f/WrX3Vsi8ecc845KY5z9Ja3vCXFhw4d6nj9w4cPd7wGry3l88q/8XPG7KOUr0VpLFxT\nHt9tjuKYGzgu9iuuPf/N67/wwgspHj58+KD6UrpmaeyxL6X9zjXm9dmulK93aV15/ssvv9xxHPHf\nnJfSfJPYr9L1S/8jO/aFY2HMfvEapXWIxzEu3UdsK/a3098OHjyol156KV/YocUxf2dXVeUSpcaY\nIUtd14P6zj6eh+E+STvx712Sruh2wtve9jbdfPPNkqS3v/3t6fMRI0Zkx40dO7bj+b/1W7+V4p//\n/Ocp/sUvfpHi+BDCB9iXXnopxXy44H8s+R9YXk+S9u7dm2L+R5XXOPfcc1Mc/8PPfu7c+drUsc98\niJk2bVqK3/GOd2RtzZ07N8Xr17/237K3vvWtKX722WdTzIeLXbt2ZW1xnPxb8z9c4vXZR0n66U9/\nmmLOHx/SR40aleIXX3wxxXzoiOdccMEFKebDyvPPP59i7qP4P6LYlzFjxqT4u9/9boo/+tGPpjg+\n6HGcBw4c6HhN7kPuifigxvm78MILUzx58uQUc78899xz2fmzZs1KMdeV+4395Z6I68U+jx49umO/\nSg+5PFfKx8y14zkcF4+R8nnlmPkdcPbZZ6d4//79Keb/kJHyOec9PWnSpBRzjP/xH/+RYt7D8TrN\nvvjiF7+oIc4xf2cbY0wbOJ6H4U5P26/7FaGqqs9J+pz0+odLY4wxPeOYv7ONMaYNHI9neLGkL9Z1\n/f4j//5vklTX9VdK54waNapufhnmr3Djx4/PjvvlL3+Z4viLaENfX1/Hz/lrT2ybvwyfd955KeYv\nSvw8/srMfvGXJ/5SyV9T+euzlP8yzF89eX3+6vjMM890PEaSNm/enGLOJdvl5/wFdsKECVlb/OWN\nv/Rt27at41g4R1L+yzB/BeR4+avfxRdfnOItW7ZkbXG99u3b1/Ga/KWT681f0iVpYGAgxXEtO/X3\nZz/7Wfa30q+23JPsY8kKIuVzwV/J+Usr55i/mEr5L/78H5WlX+L5iy/XV8rvA+5djoW/zNJ6E/ch\n14L95zV4frTocCycP/76TbgP4j7m+Zxv7ukpU6akmAoFrTORZsx/9Ed/pM2bNw9Zm8Qb+c62TcIY\nM5QZrE3ieLJJPC5pWlVVk6qqOlvSJyR97zjaM8YYc/Lwd7YxxnTgDdsk6ro+XFXVH0q6T9KZkv7f\nuq7XnbCeGWOMOWH4O9sYYzrzhm0Sb4SpU6fWf/u3fyup+5vclF4pN1NSpiTKl2J2796dtcUXp3g+\nZe/SC3C0RUi5RMzrsy1KxZS2Yz9L52/dujXFlJQp6caxUB6nTYPzSDk6WkkoPVPe5ouNlNppxYj9\npyTNttjfUnYAKbcqlCwn559/foq5XnEfPfTQQylesmRJx/5yjLRVxL7wmuwX9wjXiGst5VYSvvhF\nawL3apwXriXnlWvBNabdJmZF4TxxXvmCKG1ItLhE+wL3Me+P0gt0MWMF/8a15J7kXHAfxZfeeE2e\nU7I3cR6jHYs2i+Y+/vM//3P19/cPWZvEG8E2CWPMUKYXNgljjDHGGGOGNH4YNsYYY4wxrcUPw8YY\nY4wxprUcT57hY+bQoUMpXRg9ldHTSH8qY/oL165dm2L6G2NKKx5HHyL9uBdddFGKS/5EKfeK0q9I\nHyR9lzENFb2XTOpPbyrTjtFby8+lcvo5ejp5/W6Vuwg9qDyHqafotZTy8bNfpbRZXO/o+6RvdM+e\nPSnmXJTyVUeP9uLFi1P8/e9/P8VLly5NMfdLTNfHtG+8Pn2+hJ7f6F8uVcAbOXJkiun9jvPC63Mf\ncx/19/enmOsQ/cf0CZfSqTHl2sGDB1Mc07Sxz7wmPda81+L9SS817wN+P3BPc+9E7zo92/Rlc7+w\nX9zH9FjHtpr5i303xhhzeuBfho0xxhhjTGvxw7AxxhhjjGktPbVJDBs27HXSpvT6imaUlCllUhJe\nsGDBUY+X8mpflJ5pjaDUy883btxY7BdlcMreHMuGDRuy8ykJs0IY5VdKyuPGjVOJiRMndmyrVJmP\n9gOm6ZLydG6lNGu0aXC+pFweZ8y2KM1Tto5V3yjJs89ce9onaF2J6edYwY9rt2rVqhRz7DG9Fs/h\nPmLaLaapo10kWmxKVf44L7RCxBRktEDwb+zzpEmTUsy9Hu0jXCOew75wf3NPxqqKCxcuTDHnP46/\ngVYnKbctcL9zj11++eUppk0ipghk2jTex5w77mP2MaaMY/XBpi/d7EXGGGOGLv5l2BhjjDHGtBY/\nDBtjjDHGmNbSU5vEz372My1fvlxSLrfGbBKlCmWUhzdt2pTiUjW3eA7bpTxLSZhyaaxAR3vB5s2b\nO/aRcjhlWymXWdkXfl6q1sUxSvnb9hwL5Xz2hVkL4hzNmzcvxVwL2hQoYR84cCA7n3Izr88+U0Kn\nNSLOEftMy8S0adM6xjwm9otSOccyY8aMFNNmQfuDlFsNaIWhrYVZCHhMt/Uqrf3OnTs7tivl68Ks\nFVxLzgWtM9FOQKsSx8xx8Xpco5hRYfv27SnmfmXMPsZ54d7lPc19xP5zHefMmZO1xfuV42dMuw33\ndNyHnJemj7ZJGGPM6Yl/GTbGGGOMMa3FD8PGGGOMMaa1VLE4wMnk3HPPrZu3tCmbxzfnKVmW5FbK\nq3xbPmZToAxOGZltrVu3LsWUZKNsTgmfb/FT0qYMzIwP8W+0LfA6MdNDQ8yUUHpbn1DmZ7ucOymX\nvmn5iNJxA/su5dI154IZAXh9zjGLO0TYLvvCbBSc4zjfU6dOTTH3CPu1e/fujp9L+fwxCwHXkXNH\n6w/3nVReY1pGBmsf6SThS2UZP+4dtsUMJ8y+MWHChI7HjB8/PmuLmSZoMeJa8DtmYGCg2BdaSTh+\nfk6bA9dOym0qnAtmReF+K2VxkfL90tzrd9xxhw4cOFCpRVRV1bv/QBhjzAmmrutBfWf7l2FjjDHG\nGNNa/DBsjDHGGGNaS0+zSbzjHe/QjTfeKCkvyDB27NjsOP6bUjWLYFBqp+UhFnGgrFp6851v19P+\ncO2112ZtXXfddSnmm+uU8yk1r169Ojt/x44dKd62bVvH/nNeKIHHt/Bnz57dsf+UnVn0g2OPsjll\nbI6FlgnaMjin8W+UntluKe4G7RAxu0ID5ztmJeHemTlzZsfj2JdoUeEeobzOOeZccu2arCkNtJYw\n6wXtFFzvaIPhvJYKcDADAy0fzF4h5Xtp3759KaY1gucMHz684/Wk/H7jGK+++uoUr1+/PsXcX1I+\nfxw/x0v7A9eI+0PKbSrc+7R50DJBG82ll16atcU90uy9+++/X8YYY04//MuwMcYYY4xpLX4YNsYY\nY4wxraWn2STOOeecurFAUPamvCzlb3IzYT7f6Kdcu2XLlhTzjXZJ2rVrV4ppgSgVFaBsHSVhvr3O\nmJYJ9j3aP0q2h+eeey7FMaNBAy0LsW/8G8dImAUgrjn7wj7yGpTDKVvH9jgXI0aMSDGldsZxjijP\nl+wbpeIplMOl3GpAOwDXmJkKosWGWRtoRWG/aJ9gRoK4DpwztsvCD8x2QhuNlM8ZbTUcIy0EvKdi\nZgvaCbhe3Hucl7lz56aY96OUrwttEitWrEgxLTq0rsTzCftFOwTHEvch15g2Ku7jRYsWpZi2EM5J\nvH5zzuLFi/Xkk086m4QxxgwRnE3CGGOMMcaYo+CHYWOMMcYY01r8MGyMMcYYY1pLTz3DVVXVTWUo\npk7qBv2CjOk1ZYqk6D9mGipWpSpV/iKxUhu9mozpN6RnOPoh6emkT5f+ZXpTS+mtpNyfyXlh2jP6\nHln5i5/Htul1pZf6oosuSnFTRbCB/lJWHqMflnTbc/TAcv7ps92zZ0+KOV+xIhnnm+OnR5u+01hZ\nj17TK664IsX0CdPnyzhW1uMac+8zTRv7Hyu90X9dqiDHdXjiiSdSzHRisS9jxozp+DnvqQ9+8IMd\n+yhJTz75ZIrp3WfKO/Y3rj3HNWvWrBRz/Dt37kwx09LF+4t7gfcn54jj5V6dM2dO1tbatWtT/Nhj\nj0mSvvSlL2lgYMCeYWOMGSLYM2yMMcYYY8xR8MOwMcYYY4xpLT2tQNfX16fPf/7zkvK0RrG6Giu1\nUYalPE47wNNPP53imNKKsijTQrHyFeVwpm+LFckIU0RRXqYVI0qvTCfHCnK0KVA2fiOwz6VqcjFl\nHKVrjottce6jZYPQakBrQslyEauj0RrBtHhMlUU5nv2NafU4TlpZ5s2b17G/rHAo5fvq3nvvTTHH\nz3Hx+rFKH9Ou8ThaCGixidX0+vr6Ujxt2rSO19+6dWuK3//+96d41apVWVucM9pPeB/SrkL7R5xj\n2kG4jxYsWJBiplaLqfRojeBcsF/cu7RBRXsTU6vxONohaIvhvPA7R8r3S2MzGay1yxhjzNDCvwwb\nY4wxxpjW4odhY4wxxhjTWnpqk5Bek8X5VvqUKVOyY/gm+TXXXJNiSuiUaynpRrmTEi2retECQPmT\nn0c7wPTp01NMC0TMztAQLQ+XXXZZiuMb/g20FvD6sboaZXhK6oz5Rj0l5Zglg3+jbYHx5ZdfnuIo\nF9OOQJsI14jjveuuu1JM+4WU2wnI3r17U0yZn5YLZtWQclsMz6H9gvMd7SO03HCOWAWNbVGCj5XW\n2Bbnj3ahj33sYynmeONx3BesGsc9uWnTphTPnDkza2vDhg0dz+fc8/orV65M8fve976sLWZ34Bhp\neWCWh1gNj/uV1gjub96TnewLDbRpsC+0n5T6Fe0fPK6xokTrijHGmNMDf7sbY4wxxpjW4odhY4wx\nxhjTWnpadOPtb397vXjx4iZOn1Nal3IplW+SM/k+MyVQvowWAJ5P6fTSSy9NMaXXa6+9NsWxaAQt\nFyxqwDmk1B5lXBZF4DUpydLaQJsCxyHlMjLP57wy0wCJGTfYf2bv4Fg4rzEDBAssMOsD7S6U07mO\nMSMAZXvGzA7AtWcGhGhX4VxyjkpzH88fNWpUx5gZFHh9thvh+evXr08x15U2j2jZKGWQ4HEslEG7\nULTrcO/ffffdKf7xj3+cYtpyrr766hRH+8fcuXNTTCsNrQ20J61YsSI7n1ldOP+0fNAKs2bNmhQz\ny4WUzyVj2koYc9/TShFprvPlL39Z27dvd9ENY4wZIrjohjHGGGOMMUfBD8PGGGOMMaa19NQmQcmN\nsjslcCm3E/Ateton+OY75fQI/zZ58uQUUyqmvDt//vwU33///VlbzHxAWZVj4XxGy8aMGTNSzLf1\nmfWAUjvtB5GSvEzZn2/Icx6iTYK2B8rxLJxAO0O0NjzyyCMpZuEDytvMtMC1Y2ELKS98QfsL7QiX\nXHJJimkBiBk32E9aG0oFSGLxF9pcaOWJlpVO16NFQ8oznnD8XO9SLOX7inAfsoAI7xVaZ2JbtA3E\n/drAOeI8Srm145Of/GSKaeWgZWL58uXZ+bTMcI7ZZ94HHFe0V3EfM2MHP+cYeUxcL1pZmj39gx/8\nQAcPHrRNwhhjhgi2SRhjjDHGGHMU/DBsjDHGGGNaS0+LbgwfPlzvf//7JeUyKiVkSRo7dmyKKfVT\numTWAkq9kyZNytq68cYbU7xkyZIUs4AGJVIm/m/62rBq1aoUUy7mObQGRDsCZX/aPyhbL1u2LMUc\ne8wMQema/adlgkUFmBmDcyfl1gRmqeB883qx+ADHybYpY1NOp92kVGQjQnmcsjnXntkrpFwSf+65\n5zr2hdkRov2DMjrnlRYG2k/6+vpSHPdhqTBJybJAS0/sJ60hXDvuQ7ZLG46UzwttNZw/3pPr1q1T\nCfY5ZsBooE0iZljhHiW0NtD+wYwu8Xq06HD+eE/yXqHdJVqtaGNqrCgxi4oxxpjTA/8ybIwxxhhj\nWosfho0xxhhjTGvxw7AxxhhjjGktPfUMn3nmmSmNGj2Y9PpJ0qZNm1JMfyp9vqwUxypYixYtytrq\nVhWsEzx+165d2d/oz6SfdMqUKSmmHzVWtWJ6MVYIYyUtHkMPaEwjRe8jU4LxmhwLK8Cx71Lux2Wl\ntJLXkz7T2H+uJfvI9HlsK/o+6WeNY27geNkX9lfKK7LRZ8t0XvTWvuc978nOp0+Xc8lx0fvNccU0\nbRwX+8Lj2P9Y6Y3XpJ+Va8eqbVwTVuyT8jEzlSCvwRRo9HXTIyzlvvZ77rknxZzLlStXppieXyn3\nctN7zvnmMZwjXlvK9xjnlR5zfofwetwTUj6XzXdV3PfGGGNOD/zLsDHGGGOMaS1+GDbGGGOMMa2l\npzaJV155JUmuTKlEm4EkXXbZZSmmrMnKdJSdmRIr2iJiKqcG2jTWrFmT9bEhyqLXXHNNilkBjima\naDOI1+bfaFug3Ltt27YUMyXWxIkTs7bGjBmT4sFUMeNYKK1L5Qp6hGOMVch4zsKFC1O8YsWKFHON\nWF2M6eqkPMUVU33RWrB79+6O/YqWC84xoYTOubjhhhuy42IKuYZSFTNCG4yUrx/X4nvf+17HfkVr\nA6GFglaazZs3dzwmpvij7YDrMmvWrBTTPsH7g1YhKb+PORff/OY3U8z9Ha0NXDPuI9p9aD3i3EV7\nFS0YtBsxzRotJrw/477n35rvisGmATTGGDO08C/DxhhjjDGmtfhh2BhjjDHGtJaqJIufDBYsWFA/\n8MADkvKqUqxIJkkDAwMp/ulPf5piyuuUZLtJ5cwoQChJU1KlPBtl0aeeeirFlOp5HCXp+OY8ZXdW\nlGM8Y8aMFC9durTYF0rHXMPBrGc8pnQ+JXHK9ry2lNs/Hn744RSzYhetMPfee2+Kt27dmrVF+wmh\nfYVzQTk79ouVx7jGpaprHIeUV+1jdgRmGWHFPa5drBjIeeW88HNaRGImjQ0bNnS8Pm0SJVsMLSbx\n+qw0R3sS7yNmY4i2FkL7Ba/JPsaMHyUbEyshMoMF90G8v7gvOH+XXnppijkv/G5hu1JuUWosJ3/9\n13+tHTt2tKoMXVVVvfsPhDHGnGDquh7Ud7Z/GTbGGGOMMa3FD8PGGGOMMaa19DybRPNmNt/qjsUt\nKIvSzkBJuq+vr+M1+Ha8lMvLtFksWLAgxZS0v/a1r6WY8q6US7yMKc/zLX4WBpFyGZw2EUq6lOaZ\nkYBv9EvS6tWrU8yCAbR5cC5o0aB9QMptJqVsGjw/2ix4HDMSEBaBoJwds39QnuY1uSdKloeY/YMZ\nQ9guM3HQDhALXXD+mBmDGU64jsxeETN2cPwcC/cY15tjjH1m26UCJJy72BfaJGg3YnYGxtwfs2fP\nztri9ZctW9ax/zyf96OUZ6Ngn5k5huezvzGTB+9j2nr6+/vVCd7D0V7FdWmu42wSxhhzeuJfho0x\nxhhjTGvxw7AxxhhjjGktPc0mMXHixPoLX/iCpFzOp4Qt5XYCyrC0VvAYSu38XMplWb5VXkrqTysG\nry3lmSIoqbP/JfuDlEvCP/nJT1LMYgmUbvmmf3zrnrI7x8KYsjGtFJSdpdw+wrlkW+xLtKIwiwAl\n+ZLNgUUQNm3alLXFPl9xxRUppi3mkksuSTHX4Yknnsja4loycwD7dd5556U42iyYOYHWAu4LSvVj\nx45NcbR/XH755SmmFYZ9YZYNWjykXMbnnE2aNCnFtBZQ5o82pFIBFM4X+0hi0QzuV+7jUiGauHf4\nb9oQWKSEc8yCL7wH4jVpk+B6cb+U7CJSbgVq9u6dd96pZ5991tkkjDlFuOeee7J/8zv7z/7szzqe\n81d/9Vcp5nf2pz/96RPcO3Mq4GwSxhhjjDHGHIWjPgxXVTW+qqofVVX1TFVV66qq+i9HPr+gqqoH\nqqrafOT/Dz9aW8YYY04u/s42xphjYzC/DB+W9Md1Xc+SdKWkz1dVNVvSn0r6t7qup0n6tyP/NsYY\n8+bi72xjjDkGjtkzXFXVXZL+8cj/XVvX9d6qqsZIeqiu6xndzp0yZUr95S9/WVLugY0+Qnpt6SPk\ncfSW0isZq4jRC0j/8MyZM1O8cOHCFNOD2Q3OG/2027ZtS/Hy5cuzc5iuiufQt0l/J/2Y9DdKuZ+X\n6dDogaKXl9ejHzNeh+fQa8n5jtW6Sj5MVpejB5Y+3eirZsVA+lYvvvjijtdjOjNWk5PycdHLzEpx\nTz75ZIrXr1+fnU+POv3eXC96WOkXj75sjplp0riOMbUb4XV4H9ALTp8y/bTse+wbz+fanX/++Snm\nvMb14n7h/cU9UtqrUn7vco14HL36/N6I9xfnhW1NmzYtxRwj90esfEgvdzP+b37zm9q/f/+Q8wwf\nz3e2PcPmVOOOO+5I8Zvxnf2Rj3zkjXTbvEmcFM9wVVUTJS2UtELS6Lqu9x652F5JowrnfK6qqieq\nqnqCLzEZY4w5uRzvd3av+mmMMW8mg34YrqrqfEn/U9L/Udf1oJ9q67q+ta7ry+q6viwm/zfGGHNy\nOBHf2Sevd8YYc+owqAp0VVWdpVe/VL9e1/V3jny8v6qqMZDcDpRbeJVf/OIXKfXJ/v370+cxrREr\nt1HqmDBhQoqZnouS5sDAQNYW7QGsvEbZn5Lw97///RQzVZSUV32LcnEDf/2mBCzllg/KtUxXRamG\n6bQoNUv5nHGMrLbFtpg2jNaCCOeCMeXp2Bf2n6ntmHJuy5YtKeYcMQWWJC1evDjFlLM4F5S3eQ0e\nL+VrxPR57AvXeM+ePdn5rKLGOaO0VpL2Y1o+tkUrDe06PIfV6KR8zmgH4LzQssC91y19Heef1+BY\nKD3yXpPytadNg7YYptWL8J5gOjjuV9okOPaYWo3X5PcG145j5z6K9zOrDDZzHMd+KnOivrONORV4\n4IEHUswUmm/Gd7Y5PRlMNolK0j9Leqau67/Fn74n6TNH4s9IuuvEd88YY8yx4O9sY4w5Ngbzy/BV\nkj4taW1VVU1G6z+T9H9J+h9VVX1W0g5JHzs5XTTGGHMM+DvbGGOOgaM+DNd1/Yik0tt47z2Wi1VV\nlWRKZgqIcueKFStSzCwCtFZQEqdUHt/IpzxOHn300RR/9atfTTHfKo8SPu0BlFCYJYLZEGLVuJJE\ny3iw2QU4Z7wOrRGUyikbxyp9lIcoF8+aNSvFK1euTDHHK+VrRNmKmTGY5YJ2E86jJN13330d2+Xa\n01pA6fqqq67K2qKthtk/7rrrtR/ESvYDKbefsDpbid27d6eYNh4pn9f58+d3PJ9rSsuAlK8r33Km\nNYLWAn4e4X5jZhHuXcI3trtZBXhP8jiuMa8t5fcEKwvyPqSVJVbTI7xfOF+0JHGOeI2YVYdSbGPT\niPfNqcqJ/M425s2CFoZT6Tv7X/7lX1LsqnWnD0PHBGeMMcYYY8wJxg/DxhhjjDGmtQwqm8SJ4uc/\n/3myQDCJNd8Wj/+mJMy30vlGKLMDbN++PWuLb5RSbqUsynaZQYFvuku5xErplXYKvl3PMUq5XMw3\n70tZG5gpIMpBlJBK/aIEzbfjN27cmLXFcTJLxbJly1LMMb7nPe/JzudcPv300ymO2TgaON6YAYJv\n7VJSZ9J0ztFnP/vZFNMWEaFszviZZ55JMbMRSLn9ZsSIESnmHLH/HFfMoMBiLlyLxx57LMVcF66p\nlO8r2gF4r1Ai5N6JFgD+jdLgDTfckOJHHnkkxbQmxLeqeX3OH/cr7QWcR0maO3duirmnKXHSesN5\niFloOOccI6/PmOdHSZVtxesYY04+p+p39pw5cwY3ADOk8C/DxhhjjDGmtfhh2BhjjDHGtJae2iTe\n+ta3auHChZLKxRmk/M1NyrjMYnDPPfekeN26dSmOibMpKTOzBJNw005QyqwQ26I8TCsGswDE8ynJ\n8G17ysOlbBK0VcTr0OZAawaldkroUQ4aP358iik7UZ6mLSXaLGiN4BpRema/KKFz7mOfuUeYGYNS\nebfsAqRUtIPXi5IbLSe06/BtZq4ji7pcf/31WVvc43feeWeKOZfMmMG2pHyeRo16rYpuqZALM1PE\nOeZ6M2MH36zeuXNnirn3ohWFa8m5nDZtWop5H8Q5ZpEY7n3uHcqa8XzC65QKanC/8D5gtpV4fjPH\nlGqNMSceWroYn0rf2eb0xL8MG2OMMcaY1uKHYWOMMcYY01p6apM455xz0huaLKwR+da3vpVi2hko\n/e7duzfFlDtjoQzKyJShKbVTWqE0E9/op9TOjAa0TJSkXimXaijvlN58pzUh2iQo9fT19aWY9gna\nN1iEIRYPYDYOSkucy6eeeirFsRgIpXKOi4VVOEauVyy6wbnkurD/lLQHBgZSHMfFOWNbN998c4q5\np77//e9n53P+aC1gn7k/b7rpphRHm8O9996bYhaxYOEHWnTievPfXGPuKd4THO873/nOrC32jbaW\nH/zgBx2vR3tPtBPw35wvwgwvscAO9w73HteSeyJmeCGUUmmr4T3NrCRsK2ah4ZibN9GdVcKYE0ss\nwkNO1e9sc3riX4aNMcYYY0xr8cOwMcYYY4xpLX4YNsYYY4wxraXnnuEpU6ZIKvsWJWn16tUpZoqn\nUuUYevnoBZZyHya9h/QklrwqEyqaAAAgAElEQVSO9OzGa9IDSy8y/bT0PEm5d5Hpnngc+0VPI1Oj\nxbboIY3V/BroTeU4pHyOCX3J9PnG9HX0FnPOOBZenz7RWB2tVFmPa1Tyk0afLeFcst1Pf/rTHfsl\n5WnP6BmjL5z+M1ZGin3hOUxnxj1d2gdSvkZsi/PF1IH0xtI/K+X+a3q8Z86cmWKu6Y4dOzr2Q8r9\ne6WKifTXR785x8xUfhwj72/2N6aMoy+Q1+dYShXo4rsG/Ftzr8b72RhzfAyV7+y/+7u/K/bTnB74\nl2FjjDHGGNNa/DBsjDHGGGNaS09tEi+//HKSLijpxpQmtDpQlqUMylRII0aMSPHSpUuztiil8pyS\nbYBSb4TXp82BdgLKsFHGpfRKybUk4VBSjtYEplOjDE65l/I0+0vZW8pTwLFfrLjF87l2sZ+8JsfF\nc2h5iKl1OMe0f7AvnCN+zjWV8lRdlNlozXjyySdTTJuAJF122WUpZpqdTZs2dWyLc/Tggw9mbdFO\nwApytExwHu+///7sfF6H+5X7gmt/ww03pDhaZygtbtu2LcVcuzVr1qSYVo64D7nf2Rb3EeVH3qvx\nb1xL2j9oEeL1onWIsirnm9Yd7gOON7bFvdP0JVp6jDHHzt///d+n+FT+zv785z/fZRTmdMO/DBtj\njDHGmNbih2FjjDHGGNNaemqTqOs6SZm0QqxcuTI7jpW0aC2gPE6pmTIu5V0pf6uecumECRNSzLfY\neUyURUtV3Cjpso/doNQd3/Zv4NgpIUt5Bg5WYWO7HDutHNGawEpxfFufElZJgo59Y0ybBKGETqla\nyueY46cc9uEPfzjF8+bNSzH3jZS/TXz99denmOvKY2K2gCuuuCLFn/jEJ1LM7CFcB1psYvVCzjGt\nBlw7VpOL601rA88fO3Zsijmun/zkJykePXp01hbXj/uYeySuS8Pu3buzf/N8zh8tH6VYyq0dXOOm\nUqWUz+Xtt9/ecRxSLrHSYsP9XsqyEavL8d/NvDibhDHHz1D5zjbtwr8MG2OMMcaY1uKHYWOMMcYY\n01qqXr4hPXbs2PoP/uAPJOWyRyww0N/fn2LKwzE7QwOlEkq98ZxoD+h0Dt9u55umUi4RU4blNWiZ\niIn8KcmwbUrH7Atl3yjjUkantYFST6ngSJxHSlWlzBhsi5JT/BvHwswB/JzZJGhRkfI5Gz9+fIqv\nuuqqFFNOp0WGNgMpz5LBrA3cB9xrUSbjv6+99toUv/vd704xs51Qto+FZDZs2JBiZtZgX3gO5Xwp\n3y9ce16T1ga2G/cxr891KRV8KVkhpHy/cx/yfK5p7AuzOJQK6dDOwPuT2WHi+cxKUkrez/mKWWT4\n/dSMeevWrXr55Zdb5ZWoqsopNMxxc9ttt6X4VP3Opi3DnD7UdT2o72z/MmyMMcYYY1qLH4aNMcYY\nY0xr6Wk2iV//+tdJkqeMGWuCU5alXEp5n3JrqThDt/MpnVK2KRULiDDrRKk4RsymEPvWQGsCx0XZ\nPL6Fz77xOhwXx05rQpSqS4VCaHPgHEepnBIWJXFmo+BY2MdusvsFF1zQ8bj77rsvxZSz43zTYsOs\nCzNmzEgxs5owlsrj53FMAM8MCHG9aHkh7POSJUs69l3K53Xz5s0ppv2Fc8d9GG1IpfkvxaXCGPGa\n3FcluxGPl/L1434dPnx4imnL4bzEOeWc01rBfcT+c9/GohudMmDE+8YYMzhO1e/s3//93x/cAMxp\nj38ZNsYYY4wxrcUPw8YYY4wxprX03CbRyPV8i5RvoUu5VM833yn101rQLXE2pRLGpbfoKTt36n8D\npR5Kv4xZQCKezzFzvJSnu2UEoGxesimQUpaKTm03cF5Z5IRv7Uv5XLCIRKkYByUr2k2kfF44ri1b\ntqS4ZBGJ1gTOJdf+scce63iNaE2YNm1airdt25ZijpFvPHNccU5jNpBO13/qqaeK57NvGzduTHGn\nrAdSPhcxMwXnj/PCtngf8L6LRTe4R7gW7EvM1EBKRW44L4xL+0jK55jX5zV4f7KtaIlin5tzepl5\nx5ihzk033ZTi0vfEm/2dbUyDfxk2xhhjjDGtxQ/DxhhjjDGmtfhh2BhjjDHGtJaeeobruk7+XPr7\nmOpJyn2/rEhDXx9TL9HTGFM/0V9EfxK9yKWUaxG2TT8TPZn0vDK9Vfw3Y7ZLbxX71a1CFj28HGPJ\nLx191Tyf88I1mj17dopjJb/7778/xUynRj8XY85R9H2yLxdddFGKb7zxxhQ/9NBDKV69enXH46V8\nzpiOjH5c9jfOMf1nu3btSvGmTZtSvGjRohRz7lasWJG1xWtyj7GKWskHLpVThbGCUkyh1hB92fQD\nM6UY/bRMfcjUg3GOuJfYFr1/XO+YWo3Hcb/xOpzXbvcX+0mPPPc+v0M4D3G+O6WWs2fYmMFz8803\np/hU+s7+i7/4i0H137QL/zJsjDHGGGNaix+GjTHGGGNMa+mpTeKVV15JMmdJQpdyewBTelESYdoU\n2gx4fITnU1KlXFpKzyTlkg4laUq/bCvK1hwXZVnKPqSUmkzKJWleh32OknRDTCPFcyjh79ixI8Ub\nNmxIcbRZ0JrBtjleznepX1JuAeB8c/y0E0yZMiXFTNcn5Wv50ksvpZhzTwvBNddck53P/vO4pUuX\ndoy//OUvpzjaP7he3IeU3intx31MyZDrzb3DtjhHHHvsG9Mc8fqcI+6PaBUoVbOj3YXr3S31IeeF\ndghapZjaLaZl4vmMmUqQ12ffYyq7bungjDFH51T6zr7tttuOqe+mffiXYWOMMcYY01r8MGyMMcYY\nY1pLz7XARjqhpEp5VsqlX8omtCBQTqekSjk6/o1vpTMbBaE8GmVYvp3Kv1GaoaQbocTMfpWkW1oO\nYkYAykulqlrjx49PMeXwMWPGZG1xzkrV5Ch7xwp2tEDQ8kKbBcdbqpIX+/n444+nmDYN2icmTpyY\n4q1bt2Zt7d27N8UcI+eI51955ZXZ+Rzzhz/84RRzj3zlK19J8Zo1a1LM7A9SvvdKmUS4jpxTKZ8/\njov7jfcU55j7IF6fe4xrz8+7vb3N+5P2DR4XMzUQzjHPZx9L1ohoY+L3A+0XJFa7bOhWga7JrFGq\nImiMeZVrr702xd/4xjdS/GZ8Z3/84x8ffMdN6/Evw8YYY4wxprX4YdgYY4wxxrSWntokDh8+nIoE\n0BoRZfeYOaHT55TTaS2I1gaeU5J7aVPgG6xREmbbjEtScewLpVjaHErZBdhuLDBAmwihTYESOsdI\nOVnKbR5MWl6SsNetW9fx2hG2O2LEiBSXiivEfnLMXBdaCJgZIWY6GDduXIq5Dy6//PIUv+td70px\nlPO/+93vduznjBkzUsz5Lu2j2DeOi9fkesWMGyyuUbLY0BrBNea5Um7hYDEV3pNbtmxJMYvKrF+/\nPmuLe5ztcr7ifiPPPfdcijmXzO7AueDcxcwUvL8osfJeo92I44rfQfyuaPq/b9++4jiMMW/+d/bC\nhQvfSLeN8S/DxhhjjDGmvfhh2BhjjDHGtJae2iSqqkpSZsm+IOXyOuV5yqiUVyjD8q39eJ1SZonS\n59HmwOtQhqVUQ6mWVg4ptx0wowPHxf6y3fgmO+VlXod2BPaf8lW0JnC++XYv16VUGETKx3XxxRd3\nPIYyWfN2fry2lEv9pYwEHAvXrltbkyZNSjHX6Gtf+1qKOXeS9Md//Mcpprz+ne98p+NY2G7ch9zH\nLKDBdeSejmvENebfaG2gXYhzFLNJxGwiDf39/SlmURvKmjGDA+dswoQJKeYb37QzPP3009n5e/bs\nSTH3K+eClgm2FftCOwjXguewmAmPiVlg+LfmnoySrjEmp9ff2f/4j/94nD025lX8y7AxxhhjjGkt\nfhg2xhhjjDGtxQ/DxhhjjDGmtfTUM1zXdfJOlvyNUu4PoleWflp6JelZnTp1atYWj6Mnkh5YegFL\nXmIp9ySy3ehdbIi+T1bFKlWqK3klS9eI5/Aa7CP9zvSpSrmflXPPdWHFvpj6jv9mBSFen+mxOC+x\n8hdTXJWqoJW8uTE9FqEXjXNE7/cHPvCB7Bz6hB955JGOMfcxvamxAtr+/ftTzPWih5Ueu27+VI6F\nnrvS+LmPpDxFWGlfldLq9fX1ZccxnRqrTNGnzHRs9B/HttlP7inGHG+co9I9SY80Py+lcovXbP7W\n7R40xvTmO9s+YXMy8C/DxhhjjDGmtfhh2BhjjDHGtJaql+mCzj///Hru3LmSpJ07d6bPmXbryHEp\npoxCGZoxq7HFqlRsi1IL5VlK+EzvxGOkXCYtVaOjBSBWEaOkTngO5Vm2G2Xc0lxwvthfWjyifSNa\nFTpdn5YJystSbifhGNnn0nzFSnqf+tSnUvyxj30sxbQmMD3XwMBAiqM1geNiTKmdUtxHPvKR7PzJ\nkyen+Nvf/naKaXEppQOLfWEVtJJU3816w/6X7gnuHaYsi3uH16H9hVYaVo9iu5wvqXyv8vobN25M\ncayGx/FTPmV1OFpvuL/idxfXgucz5jxyHmLqQu7ppt2BgQEdOnSoVV6JqqqcT868IW655ZYUH+93\nNquBGnMs1HU9qO9s/zJsjDHGGGNaix+GjTHGGGNMa+lpNokzzjgjSaGUgfkWvZRLlpT0KQlTnqWE\nz8+lznJn05dOcencCPvM6mC8RpReeR1aMDguytAcS2yLMjaroJXmLlo2COVpVhFjW7QGRJsFx8Wx\nMGtFKVPC6NGjs7Z43IwZM1JMCZ99XLt2bYqj5MY9RmsEr8kMCMyyIEnf+ta3Usw9Vlov2gTini7t\nC37O7BW0L0i5nYRrwX3Atrq9pc15oe2A68X9wvsg2iRoc6AUyj1Cm0asOMW5oJWEMdviOsSsJoRj\nKWXc4LXj/cW5bOTa0veEMeb1fPCDH0zxG/nO/uIXv3jyOmdMYNDf7lVVnVlV1cqqqu4+8u9JVVWt\nqKpqc1VV36qq6uyjtWGMMaY3+DvbGGMGx7H81PFfJD2Df/+VpP9e1/U0Sc9L+uyJ7Jgxxpjjwt/Z\nxhgzCAZlk6iqapykD0j6S0n/tXpVy1gq6T8dOeR2SV+UdEvHBo5w+PDh9DY5pVfaDKTyG9+lJNyl\n4g7xuFLib0qvLMIQZVhK0pTKo22goZtUTumWhQvYbqkAhpTPxcSJE1NMeZlv4TNjBwtAxH9zjFde\neWWKKeFHO0GpiEXJosLxMpakyy67LMUcMyV4zivtA5wHKZ8jtkWbBLNUrFixonj+/PnzU8zCE5T1\nKLXH4hKcI8a0END+QOuLJG3bti3FlPo5Ls437Qtx7/AeYUxrBueV88V1lKTdu3enmPutZGHoZjVg\nX0qZW0jMJsFzOK9xLjudH7N38Dupub+HUtGNE/WdbcwbhYV3jDnVGewvw38n6f+U1PzX6kJJL9R1\n3fwXb5ekvk4nGmOM6Tn+zjbGmEFy1Ifhqqo+KOlAXddP8uMOh3bMR1lV1eeqqnqiqqonur3wYowx\n5vg5kd/ZJ6WDxhhzijEYm8RVkm6squp/kfQWSW/Tq786vKOqqmFHfmkYJ2lPp5Prur5V0q2SNGzY\nsHr79u2SchmTSfGlXK6lXFkqSME3xyOUlCkjU86nZYIybrQ/UDan9Mo38ksSdGyPmSLYL84La7vH\n7AA8jvOyePHiFFPab4qdSFJ/f3/W1j/8wz+kmMVQKHPRJhGLkdBmUbJJcI1HjRqV4phNgtYQ2glo\ngaBNYfny5Sk+cOBA1hYLPNBa8a//+q8p3rx5c4qj/YP7kAU4Ro4cmWLaEWhziPYP9i0WwWigbB//\nhyP/TTsC4T6kpB+LqpQyjnC/8f6iFSK+/c37oJQxg2sci6zwHuHe4TmrV69OcalYThwLj+MacR7Z\nF66vlGcZac7funWrhggn7DvbRTeMMW3gqL8M13X93+q6HlfX9URJn5D0YF3X/1nSjyTddOSwz0i6\n66T10hhjzKDwd7Yxxhwbx5M480/06osZW/SqH+2fT0yXjDHGnAT8nW2MMR04pqIbdV0/JOmhI3G/\npEXHcn5VVUkipuUg2iRojaDUTqmbEum8efNSTMuDlMu6lJcpCQ/2zXdek9IzM0NQAmehCCkfM20H\nzABROj72hXPBJOaUmjnGH/7whymmzUDKbRLMJnH77bd3jGktkMoFPTh+yva0GTCWcnma67Vy5coU\nM+vDhg0bUkyLh5TPGeeCdoJuhSpok6CFguN95zvfmeJ///d/TzHXQcrHTzsEMz10K9hCSkU/eE1e\nI2Zd4N7l3ue4aH/gPRTvL16f92rJjrBgwYLsfN47tJns2LEjxSy+QstGpGSpKu3PkiVJyvdOs0Zx\nHocCx/udbYwxbcAllYwxxhhjTGvxw7AxxhhjjGktfhg2xhhjjDGt5Zg8w8dLVVXJf8fUR9GrSb/e\nmDFjUlxKXUV/ZfSN0o/L4+ibpAeT/sLo06UnkanRWG2M42J/4zXZT/pkGdNzG9O8MT0Zfb5sd9my\nZSnmfMV0ZoRrQS8y+zJp0qTsHKYw45zRK0q/JT3LU6dOzdoaGBhI8YMPPphipm/jfNP/G1OW0Y9L\nrys9sNxfcb055/Tjcl75+a5duzp+HhkxYkTHa3LvRH9qySfMsbDKINsqVXCTytUTCfd69DKzn2yL\n80rPbzePOOeC46IXmTH3nZTfu9wXvO8Yc0/HcdGT34xxKHqGjTHGHB3/MmyMMcYYY1qLH4aNMcYY\nY0xr6alNQnpNSqWMGeVOSqRM8cQUbLQssAIdq01JufzJdEuU1NkXystRNqf0TQsEr0mpNsrT/DfT\ndo0bNy7FlPMpy0bZnVXkmLaLUjUlZVbPYtoqSfre976XYs4L565bqi/KzZSXOS6yZ89rha8ok8f+\nM90V15hWDlohYoo8joVrxLR2pUp+Up5OrZSmbc2aNR37RVtD7Cf3N48rrZ2Uzyuvz7Xn/cH+8h6K\nbfOa0a7U0C3lG8dCCwL3MW0W0coS77EGWmw4Ru6p2BfuK6Zg431E21ScF3LxxRenuJmXwVhKjDHG\nDD38y7AxxhhjjGktfhg2xhhjjDGtpee6XyNFs/JUlKcpf1JW5dvylIopg3aTyikPUzanVE3ZNkq6\nPJ/9L1Wm6yar8m/MlEA5vFTRS8rHzD5TUua4KPv+zd/8TdYWz//Sl76U4qeffjrFlJfjvLBtXn/m\nzJkp5rwwEwZjKa86x3mlbE5bC+eRVgop3yOcC16DcxwzHXDOeR3aLGjzoJWlm/WH88eYFh1mCIn/\nLlVM5Bh5vbgPS1kbeE/wHNoc2McI7wnuz3Xr1qV44sSJ2TncL8yEwiqHW7Zs6dhfZtWQ8u8NjpHr\nxUweXK9t27ZlbXGPNZamaF0xxhhzeuBfho0xxhhjTGvxw7AxxhhjjGktPbVJ/OY3v0lSLGVYSuhS\nLh3TNkD4hjvl4fhGPGXYkgWB0i/7Qtk3Qhm4VDQjwmIRvD4tCMxSwWvQJiDlci8LajDLxpVXXpni\nm266KcXxDf4777wzxcuXL08xpXlaDubOnZudP23atI79ZzaG6dOnp5iZMGjFkHKpm3PEjBWl42Mm\nkVJ2A8rrHGO0JowdO7bjdSih08rBY2jpkcrFKdgW7Q8xEwn/xn1csjYwe0U3uF84L7wHeUy8V0vX\nZH+5P2lZkPJx8n5jX8aPH5/iAwcOpDhmg6CFgvc9j+P1+F0R7y8e1+zjkj3FGGPM0Ma/DBtjjDHG\nmNbih2FjjDHGGNNaKsq3J5uzzjqrbt7M5nWjpEu5NMqyDbQ2TJkyJcVR7uQb6iw2QRmVUjWl+fi2\n+qRJk1JMqwHlU8rAkydPzs6fNWtWx+tQnmfhABYYWLRoUbEtZmSgHYBQwr/llluyv331q19NMa0N\nLAwyYcKEFI8ePTo7n5J6qWBKyf5ACV6S+vr6Ovafx9FmQaLNgf1nv0pZAaIthtYM9p/WCFoWuA9i\nQQjuMcrz3N+0VsQiK6SUSYT9KhWCidcsZTwpZWNgYQ8ptyhx/nl/0tKzePHi7HxaKEqFNmgvYjEN\nWiakfI7Z54GBgY7nM/tI3BOdMszs3LlThw4dqtQiqqrq3X8gjDHmBFPX9aC+s/3LsDHGGGOMaS1+\nGDbGGGOMMa2lpzaJc845p26yLcS35QltE7QqUN4d7JvdlD8pFzOmhN4NyriU8ynJsoBGzNrA8ykP\nMxsDLQ9XXXVVsS3KvWxrzZo1KV67dm2KaReJBQYoN9OOwKwRXIeYEYBjpqTNDBRcu1IGhHgcLR8c\nP6/BdYx7omSxKe2deDxtErxPKMeXimbEa3C/0wLAa7LdaLMoZXTg3uW6ct9HGxItAJx/zjEtD92y\nVHBPs13OC8fOvSrlGTvYFq/Je4392r59e9YW55z7lQVQOF9xjgnvl6bdTZs26Re/+IVtEsYYM0Sw\nTcIYY4wxxpij4IdhY4wxxhjTWvwwbIwxxhhjWktPK9CdccYZOv/88yXl/sjo2aX3j/5Kpr6iJ5HE\nNFL0cfKa9CTSA0vfZPQUsl+lKnn0RPJ60uvTiHX6nJ7hp556KsVxjrZu3Zri9evXp5heYvo76XmN\nPvH58+enmBX09u7dm+Jdu3alOFb5Y/+b9ZWkiy++OMX0k7Jd+o2lfI5LHlZ6eelHjdUK2RbjUjqx\nmEqPa0wPLsc/ceLEFHOOYtqvUgU6+qe5X6JHnHuU/Vq4cGHHvmzYsCHFnG+pXOWQ99dgPM5Svi/Z\nx5LHmfejlO/Xkk+YfWHKuW6+bK4X9wXTMHKO4/3FvjRrWfrOMcYYM7TxL8PGGGOMMaa1+GHYGGOM\nMca0lp7aJM4+++xUVY1SbbQTULJklahShSlKmpR0pXJqNsqo/JztdrNcUK5lpbNufaElgBYGjpHz\ncvPNN6c4WjZYqY4WglK6KPYrSt20M3C+KKFTao52Dx7H82kB+OhHP5pipkbj2KV8zjjHHBcrCXIe\nuqXrI5TXaeuIMjjHyfR33IfcE5yjaAfgXuLa03LB8TItnZRbHVjRjftl3rx5KeZ6sxKglK8Rj+P1\nSzaDaCeg1aBkQ6J1KFaw43WYSm/kyJEppg2IFRKjLYZtcb9wjpkyjXMXqxd2srLE7wNjjDGnB/5l\n2BhjjDHGtBY/DBtjjDHGmNbS0wp0F1xwQX399ddLyuXSlStXZsdt2bIlxZQ7Ke+yWhXfgo/yNKVf\nyuA8J1bV6nR8hH1hzHH99Kc/zc4pSfq0LVC2p/2C/ZWk6667LsUzZ85M8W233Zbixx9/PMWl6nnx\nbwsWLEhxf39/irtZMZjd4ZJLLknx8uXLU0x5mmOZM2dO1hbHQgvFsmXLUkzLAKVrrrWUWy64z0vr\nzbmXctme1gTOH/tC2T1Ca0jJItTNylKqzMe5HD9+fMe2Yr+4D9kv3ju0P3Be4xyX5pJWFB7DbClS\nbqHgPmTWh1I2CWbMiP0vVRnkPcWxR6sW56/5286dO3Xo0CFXoDPGmCGCK9AZY4wxxhhzFPwwbIwx\nxhhjWktPbRLnnntuPXny5CZOn0fZnYn4KX2ycAPlTkqc8Q3zUnYHSq88pvTWf+wnMw2wXfaxVGQj\nwqwL7BfHFd/CpyS+ePHiFD///PMpfuaZZ1JM2XjMmDFZW/wbZWj2n/M9Y8aM7Pzt27enuJQZg9YC\nWg4o2Uu5HaCUVYTZEbheURovZTrgevGceC9wjkv7jfuzZH2JcI6516OthpSyebBICu0qpTFKuR2E\nc0QrC/dkqR+xba5rKTNFtPvwbywgwrlnu+z7pk2bsrZY9IT94rowGwXHGO0fnOPGLrVv3z798pe/\ntE3CGGOGCLZJGGOMMcYYcxT8MGyMMcYYY1pLT20SZ511Vt1IuXyrnG+RS/kb+nzjm5I4bRaUNGPh\nBcrFpQIctElQmo+SMNtmuzyfsjOvJ+XZKWi5oG2BmQL4pn3sC4sycF5o0+D1eW32V8rngnYMWhMO\nHDigEly/d73rXSn+7d/+7RQzG8TGjRtTTCuHlI+Fc0n7ByXw0hil3M7Atko2C85dbJvnU+rnOnK+\n4n1FmwXXkjYBWi6iZYJ7j21z7Up7Lxbd4L5izGuysAnHyKI0UnkuuEY8J2ZtINyXtOI01iopHyOz\nnUj5WvL+4Fxs3ry54zExCw1p5vuFF17Qr3/9a9skjDFmiGCbhDHGGGOMMUfBD8PGGGOMMaa1+GHY\nGGOMMca0lmFHP+QEXmzYMI0ePVqSNHXq1PT5pEmTsuPoo3zsscdSTO/jYP2w9GfSb0ivZSlVVUxJ\nxdRN9JfSK0kv8XnnnZedz3HxOKYgYzo3jjd6NZluquSNZaoq+kzpw459ufzyy1M8ceLEFD/44IMp\njuOiz/iKK65IMdOm0avJlFgHDx7M2uL88/r047L/XIfYFtePnlCOt5QiL8I5pn+Z6cy4RvSeS/k+\nLKVjY0z/a+wn9xv3O8/n9WKKQHqDS3uXXuJSxTwp34elCnpcr+iF5pxxXrj3S3s9pq+jf5ke7VKa\nNc7R1q1bs7ZYcbG5b+OaGGOMOT3wL8PGGGOMMaa1+GHYGGOMMca0lp6mVjv77LPrRgafOXNm+pzS\nupRL4uvWrUsxZVTaJCjTRymzlMaKcitlZM4Hj5dyOwYlZcbdrA2UpEsptdh/pixj5S1JmjVrVopZ\nSYupyngOrQ2f+tSnsraWLFmSYtoh2N8PfehDHfsV+dGPfpTib3/72x37RTk7VtZrbDRSbvkopfqi\nfYCpsqS8qhj3C+eYaxfvBVowuF6U2tlH7rVuKcRKthZ+Hvcxj6Pszz5yLJyjuHdo4eBxHBdtAtw7\ntIVI+ZxxjrknaZ+gfUHK14gx7TolK0qsXMlKiPFvDdxvnJdYcY/WjtWrV0t6tULfoUOHnFrNGGOG\nCE6tZowxxhhjzFHww7AxxhhjjGktPc0mUdd1kpIpw8bqT8wiwIpslFjnz5+f4vHjx6eYmQokadWq\nVSlm5TNmOii9bR4rmpUqfFHSpgUgVv4iHEupahxlY2ayiG3TKkBJefbs2SlmFa8JEyZkbVEG//jH\nP57i+LZ+Q7QAfOlLX5ydCBwAACAASURBVEox53v69Okp/sM//MMU7969u2O7Um5HoARP2Zt2hFJm\nBSmXxPv6+lLM9d63b1+KYzYK2nU4R1xj9pf7JVaz45yxn5Tnu1XDK9kkOEe0SbBfcR15HV6fdiHa\nEThe9kPK9+vYsWM7trtz586O/Yp95lquWbMmxZwLXj/aP0rZJNgu+1XKNBOv2WS72b9/v4wxxpx+\n+JdhY4wxxhjTWvwwbIwxxhhjWktPbRLSa1IyrRHMGCHlb5LffPPNKb700ks7tknZl8UCpNxqQEmc\nn1P+pKQa33xnv6ZNm5ZiSr333Xdfiin1SrlczMINlKRLMjDflJdymwf7ctVVV6V40aJFKWaxA0rb\n0uvH2QlmmbjjjjuyvzHTA6V2Zi2ghYBydsyaQDsH15JFO37yk5+kmLaQbgVX5syZk2IW9li5cmWK\naaOR8mIT3C8cIzNrcE9zfaVywRdaHmJBCsI9Fm1Fna7B46O8z/3GsdAawD3BtqJ1iGPhfJUK4cS+\ncC8QXrOULSbO1969e1PMfcQxjhw5smNfYoEdXj/+zRhjzOmFfxk2xhhjjDGtxQ/DxhhjjDGmtfTU\nJnHmmWcmyZMSJ2VrKX8rvWSNePTRR1P88MMPpzhmXWBGAGawYFJ9Wg7GjBmT4ph1YcGCBSlm1gVm\nUOjv708xM1NIubzNt9opwRPaGeKb85S6S8UDaNO48cYbO46jG5zLH/7whymmHC3l4xw1alSKKYGv\nX78+xVz7ESNGZG1Rque8sJgHLSO8NtdOyuX1FStWpJjZN9jHWLSDsj/7xTku2T/ienGP8ThK8My0\nELMuMKsKZX/uKdoRStknpHycbKtUWITzGAuT8Bzex7wmrRS8H2P/ud9pVyplcYhZTXgdWpI437w/\nS9lKpHz+mrhbIRVjjDFDF/8ybIwxxhhjWosfho0xxhhjTGvpeTaJ5i1tStUsmiFJS5cu7XguJc57\n7rknxZTQY0YAvhVO+8WVV16Z4g984AMpppUiSrpPPPFEipnFgLI5pVe+0S7l2RUo41J+pexMqTla\nAChJ881/SsWUt2MWgMHAuViyZEnHPkr5unDOKLW/733vSzEl/2gl4Xo99dRTKabNgNegnB0zOJQk\neGYV4ZrEQhclewDb5byWMjBIuW2C57AYCG0O0drAsVHS53HMtNBJ5m9gARHuPbZFi8jFF1+c4rj2\npQIonEteI+5DzhMtNpwL2moYR2sD17i030uWom6ZSBorSbSIGGOMOT3wL8PGGGOMMaa1DOphuKqq\nd1RVdUdVVRuqqnqmqqrFVVVdUFXVA1VVbT7y/4+erNYYY8xJx9/ZxhgzeAb7y/DfS7q3ruuZkuZL\nekbSn0r6t7qup0n6tyP/NsYY8+bj72xjjBkkR/UMV1X1NknXSPpfJamu619J+lVVVR+SdO2Rw26X\n9JCkP+nW1hlnnJF8ffRKsoKalPte77777hQ/9NBDKV6+fHmKt2zZkmL6AKW84hS9ye9973tTPGnS\npBTTkxhTiDENGGP6f/k5fZeSNHv27BTPnDkzxfTAMuUb/az0WEu5Z5k+yMWLF6eYabN2796d4jiu\nG264IcUlXyTTnHVLh8Z0bJxvnl9KTRavz7bYf3qZ6Z2mx1fKvaqbNm1SJ5h2i9XkpNwbTA8px8v+\nciyxLf6b/XrhhRdSzL3KMUq575fncy75eSfPawN9utxXPI73IPsY26LXdtu2bSkePXp0iuml5p6O\nfab3nHNcShMXqxdyvbgX6EuPPuMGpiqM12zmcqh4hk/kd7YxxrSBwfwyPFnSs5Juq6pqZVVV/09V\nVedJGl3X9V5JOvL/R3U6uaqqz1VV9URVVU+4rKkxxpx0Tth3du+6bIwxbx6DeRgeJulSSbfUdb1Q\n0s91DPJaXde31nV9WV3Xl8U3to0xxpxwTth39snqoDHGnEoM5ul0l6RddV03Jbzu0KtfrPurqhpT\n1/XeqqrGSDpwtIbOOuuslD6J6bWeeeaZ7LivfOUrKb7wwgtTfPDgwRRTXuVDNtMrSXm6pquuuirF\nv/M7v5PiUhqmKIvGdFkNpSp5J5KY0oryOCVlzith1TGmopOkf/qnf0oxq31F+0pD/B81tIOwah/t\nG7SfMFUXP5ekxx9/PMVMrUZVgdI+90Fsq5SmjhYA2hdiajXuC1bto02Esn9pTaQ8NRpT/F1zzTUd\n+//AAw9k59OqUErtxmtwvNHawNRqtIlwLjkuXpvHS/kac8w8h5aNeK9xL02fPr3j9dnu1KlTUxwt\nD7Qb8XxadDgXtNjEKpC8ZjNfHNMpzgn7zjbGmDZw1F+G67reJ2lnVVUzjnz0XknrJX1P0meOfPYZ\nSXedlB4aY4wZNP7ONsaYY2OwvoX/TdLXq6o6W1K/pN/Vqw/S/6Oqqs9K2iHpYyeni8YYY44Rf2cb\nY8wgGdTDcF3XqyR18o+9t8NnRaqqSvJjqVKZlGdKoBRKy0PMiNAQq23NmTMnxe9+97tTzOwClNNp\nE4gyLN84pzw8ZcqUFJfeyD9eotRNOwQl3c2bN6eYb+f/+Mc/TvHAwEDWFuds3LhxKaaczrfzo1zM\nzBi0CnCNKO1zvWPWhHXr1qWY68L5plTPOGbc4PncUxwL7QRx77DqH6V+2nUox3PvRDsB7QBsi+Pn\nHDEriJRXzeMaM+MG9wHj+OIqrQqclwMHXlPNS9abaCfgv0uZKXh+tEnQmsFr8l6jfYRWkJjVhNfh\n+YsWLUoxs5JwHWNmCo6l6ddQeufhRH1nG2NMG3AFOmOMMcYY01r8MGyMMcYYY1pLT3W/w4cPJ2my\n9Ha/lEuhzOhAiZNS++TJk1McZXdmOrjzzjtTTHmYki77FbNHsCAGs1ZQdmdbUXZn25RxORbGlOmj\nZaNUHIQSPGV3zkMsMMACCbQW0DIxY8aMFD/99NPZ+ZS+OUZK4JTgWaQkZuygvYBzzD5T0qZ9IFpn\nuK/YLm0eJWuBlK8315XWCK4DLQNx71B251xwjJxvZtyQ8sIyHBf7ResR54IWk9gX7jfOK9ui/SBa\nBbjejJkFhn2k1UnKLUZcF9plmPGDGSB4f0j5XuD3AMdL6w+JlgveR809Ee9nY4wxpwf+djfGGGOM\nMa3FD8PGGGOMMaa19NQmUdd1ks5HjhyZPo+FMq6++uoUUyLds2dPiinJMoMBYymXdSnJU96mjEo7\nwfDhw7O2WISCkjr7SDtDtDZQRmehDtpCKMVu3bo1xTE7AduifYSw/5SjKdNLuR1gyZIlKaakTJme\nhSKkfM6++93vpphzTNmca0crh5TbNErFUJgFgG1FmwPbLmUtiNcng8keUFo7Xk/K9wKtCU888VrF\nWxYsmThxYnY+C9NwXko2j2hHINzvbIv2izdiCeBcck+x3ZiJhPPH47jGtIzQYsNMGlK+3+fPn5/i\n9773tQQKq1ev7thHxlLnghxxfxljjDk98C/DxhhjjDGmtfhh2BhjjDHGtBY/DBtjjDHGmNbSU8/w\nK6+8ktJS0ecaU6vRR0noO6Xn9jOf+UyKY3W1e+65J8X0ITL1Ff2cTG9Fz2vsJ/9GLzLTUEXfZcmD\ny8pl9EWz3ei15PXZf/oa2RbTZsV0ZvRaltJjMRVd9E7efvvtKWZKLvo7eX16ZjlfUj5nHDM9uBwv\nfb1xvtlnpspiW9wH3eaFvnb6mjlfTLkW54jjZ//pN1+/fn2KV65cmZ3Pfcn+l7y9nOPx48dnf5s+\nfXqKeb9wLpgukHMcfdTsS2mNeU6813kc54Xt8v7guwbRl821oN/+lltuSTF9xvQJc3/EvjSp4eL3\ngTHGmNMD/zJsjDHGGGNaix+GjTHGGGNMa+mpTWLYsGGp2hml5m5pwyirUlJlerCHH3644+dSLj0z\njRNl0EmTJqWYKcgos0u5RMuqWmyLsjVtCvHflN23bduWYlopWEWLErSUS/o8h3IvJWHOcbQDzJs3\nL8VTp05NMau+8fzNmzdn57PyF+eYsvWLL76YYs53tH8wTRvh+EtV/qKEz/3Cczj+bueXqvnRPsG4\nW8o2SvK8DtOcbd++XSV27NiR4lidroFp0tgu7zUpXyOew/WmTYLnx9SFHDPnslTVMe5j7ovSHLNd\nttXf35+1xfuolJaPVhZadKJ9g/dU069oyzDGGHN64F+GjTHGGGNMa/HDsDHGGGOMaS09r0DXyJyU\nYflGvSTNnDkzxXyDmxIt5dVly5alOFazo7TJa1ISptTLbAjMpiDlb6gzUwIlYcqwtDlI+dv67Cdl\ne0rFlKdjdgL+jVYDSvuUyin7RgmZlc+YcWPVqlUdz48V72gnYdYDrhfbZd/HjBmTtTVjxowU035R\nqjLI8ca1p7WD1pvSnoh2HcrzbJvHleT4bpYLQosN1552ACmf/71796aYY2HVNH4esyBwLbj3uMdo\nZzj//PNTHPc054L7jXud1pdosyhVkuT8sf/8nHtFkr7whS+kmOvKdeT9ye+QeH/x+6FZ47/8y7+U\nMcaY0w//MmyMMcYYY1qLH4aNMcYYY0xr6alN4je/+U2SJikPx8ILlCwpo9OmQKmYMj8zCEh5BghK\nt5RLmdSf0ny0b5CSHYJyPK8ndX+rvoFvtZcKGki5nWPUqFEp7uvrSzGzNlASZh8lacWKFSmmHYBz\n2WQBkXIrhJQXceA5HP9jjz2WYhZ6iLI514vWAsrjlOZ5jZjBgfYVnlOyJsS9Q3sAbTFcI1oD2G60\nRXD9uL9pGelmi+FeoIRPC8DGjRvVCfYx9p/nc/y0ZRBmeYhtc404F1zTmD2E8zR27NgUf+QjH0nx\n0qVLUzx79uyO/eoFt95665t2bWOMMScP/zJsjDHGGGNaix+GjTHGGGNMa+l50Y0mQwMl1fiG+vLl\ny1NMiXX8+PEpptRLC0CEhQx4Dm0OlKRLb7RLuSTPLAKUmmnrYHEDKZfBeQ4ldH7OLA2U7KVcxqaE\nzqIjtGWwmMbcuXOztubMmZNiFod49NFHO17/xhtvzM6nVH7ttdem+O67705xqeBIhLYFzjFtIZwj\nWgZi4QRmSmB2Aa43bTHRTsAsCJTzaSfg3uX1o0WG5wwmg0S0I9DyQZsLz+E68vNYiIZ7n3u8lFmi\ndH/E63D++Hkp44Yk/e7v/m6Kf+/3fq9jv0pEKwr3FcfI63Pued/Ftef6xbU0xhhzeuFfho0xxhhj\nTGvxw7AxxhhjjGktPbVJnHHGGUmOpLRP2VvKpUxaIChrzps3L8XMZhAtF4Syd+ktfsrulKZjn3kc\n+8UsFyNGjMjOp1xLqZvj5zW7ZSfgv9l/Zk1gsQLaDPjWviStXLkyxevXr08xx8X4G9/4Rnb+DTfc\nkGJmBqG8Xsq6EO0fPL+UmYLWBmamoK1EKltpaE3gfHF94/mlfvEYrgMtB1I+ZloI2Bb3VIQ2D1px\nuJaXXHJJimmzoPVFym1I27Zt69iXkg1nwYIFWVv82+bNmzv2ff78+SmeOHFi9jdmP3n44YdTzDni\n/cGCL7EwCeeP/eI+5H3DPR0zrDBjSrP23b5bjDHGDF38y7AxxhhjjGktfhg2xhhjjDGtxQ/Dxhhj\njDGmtfTUMyy95qWkN5beVin3dNIrSn8nPbCE3kwp9wXGCmUN9CTymJiqi3+jJ5H+UF6PHmUp92Sy\nLV6fHlh+Hv2RTD1Ffyc9nawad/DgwRQ/9NBDWVvRK9tAD2yTEk/KfaZS7vtlv5jSi+vYLVUXvZv0\naHLtuS6cl7iPmF6L55TSdsX1Zj/Z/9KeYruDTcfFa7KtuFfZF84Fqw/29/enuORrlqTrrrsuxfRZ\nr127NsXch1yTmIKMe5R+YN6HXOMdO3Zk53OPcL1Y/ZF9LL1DIElbtmxJMfc0Pdacl9K+lfLvl6bP\n0bdvjDHm9MC/DBtjjDHGmNbih2FjjDHGGNNaemqTOHz4cJLrKfVGawMtFJQmKWVSUqUkGqtdsW3K\nvUzDVKrCFWV39ov95+eUt2O6JtoOSlXMKIdTgmZqLSlPqbVkyZIUU8Z+4YUXUkwrRYSSMM9hHyln\nX3TRRdn5lMGZ0otS99VXX53i1atXp5ipsqR8LWmBYGozxrRvRLsH907JCsK161ZxkDErnXFNOXcx\nzRvP554sWWyiDYj/3r9/f4o595wXrgPThEnSqlWrUnzllVemmHuE9xf3QUwZN3ny5I7957yuW7cu\nxXEfl/Y7LUalSnHRYsNUhjyf/aethH3keOPfdu3aJam8h4wxxgxt/MuwMcYYY4xpLX4YNsYYY4wx\nraWnNolhw4Zp5MiRr/s8Vsh6/vnnU0wZnBIrpWa2GSuaUV4u2RRoraD0Gq0FJXmbb+uXKsNJuQzM\nmNJv7H9DnLcbb7wxxZT3t27dmmJKv43UK+Vv5Mf+c8y0fPCcuF5PPvlkijlmStIzZ85MMeX8mJ0g\nyvCdrsk9wXnhvonQ1kJoxYh2HcK5iFacBlpv4ri4d2mLYbscOzOBSPm+5HqXPud6ca9HaFPgXHDt\nSvtAyq0DrL7IueRcROtRab+XqsPRrhOhNYK2FN5rzDjBa8Q5osWn2W+lLCTGGGOGNv5l2BhjjDHG\ntBY/DBtjjDHGmNbSU93vrLPOSvIv37zv6+vLjqNsT/mS51B6pbzL4hJSLoUyCwChpNutIAQlXdoB\nShJ8LLxAiZdj5FvwlGLZ3+nTpxf7TNmdcjxld9ofos2BbdFawfMpNXcrUMDzmSli/fr1Hc9n36Xc\nZsKxULamzYKyO9dOKttaKLtzf8UiDoR7jP0v2W1iwZVS22yXx3C+pXyP8RxaGFgoY8KECSmOVgTu\n60cffTTFvKc4FvYrFn/h/cY9zfM5lnhP8d8luxHvI7YbbS3sGzNolIqk0OIR9+G+fftS3Ky3i24Y\nY8zpiX8ZNsYYY4wxrcUPw8YYY4wxprVUpbf3TwbnnXdePXv2bEm5nM4E91Iul5YKL/DNfcqoPCZC\nmbNUSICfR0mY1+HfKM+XiihIuUTLMVJ2pwzMMUYom3NcvD7bKhUMif1iX3gcbSHsu5RLypwXns9r\n8BhaX2KfKe+X1pVrwmwG8Zqldrutd6nIAtsqZSiJEj6vSfsHLRvsb7RJ8N/cb1xXHsO5j5kpWBCE\n92Gp+Az3cbQKcCzM8kHrDON4r5fsRrwmrTDce9GGxDmnrYZ2I2bJYFsbNmzI2tq+ffvrrrN8+XK9\n+OKL+UVPc6qq6t1/IIwx5gRT1/WgvrP9y7AxxhhjjGktfhg2xhhjjDGtpafZJM4++2yNHz9eUi6X\nxiIGJXm9ZDPg+VHaptxKeblUoIBtRTsBZXTKxaViDVESHjt2bIopD/P67CPl4ZjBgdYESvDsMwtS\ndGur9LY/x8u+Rwmfa7Ft27YUl2wpJWleKmcxoJ2CsjetETFrQskOwTFyvij5d+tzKbsCZftYsIVw\nvbmnmc0hWlFK9hOuBeeOfY9zzOuwLWbA4H1Em8XUqVOztmin4ByVCufE4hal4hiEc8k9Fe9Pts39\nzvuT4+Kc7tmzZ1DXNMYYc/rhX4aNMcYYY0xr8cOwMcYYY4xpLX4YNsYYY4wxraWnnuEzzjgjeSGn\nTJmSPqd/NkKvKqtEMfUR/YH0LUq5d5H+RLbLCnD0k9LbKZW9g/T5MlXW8OHDs+PoaeTf6H2kv5Ee\nzuhLpm+WPkj6ZEupsmIVsFKqM57P6nn0nEbogeVa0M/K82NqNY6Lc1m6BuPYFsfMeSmtcaxgV/Kz\ncr14DXrEY1ulscTjGqKPvlSdjjHnm/PI+0Mq+7/p7S3N8c6dO7O2OK+lFHtcl+jz5Rzzb6ySWJr7\n6Kvmv3lPsF9si/uQY5fy+625D+KaGGOMOT3wL8PGGGOMMaa1+GHYGGOMMca0lp7bJBppctOmTenz\naEdgiqt3vvOdKf7kJz+Z4gcffDDF3/72t1McLQAleZr2iZL8GdNA8TjK0JSRKePG1GqsokYLAqVb\nyu6UfaNNgnNE+wblYcbsF+0m8d8cI6/JuegmdbNflOfZF1Yki7YWyu4cP+eIa9dJzu70b+4xfj5u\n3LhiXziuUpU+QmvAwYMHs7/RDsFxlWwScY24rynvsy/ck6wyF+8JWh1oV+L4OV7OXewXU5CVbCJs\nN+5j9r9U5ZBzyc/j9wb3KPvCSo78PqD1J+7pTpU5SykUjTHGDG38y7AxxhhjjGktfhg2xhhjjDGt\npec2iUZWpeQYs0mwGtQPfvCDFFMGfs973pNiSvDLli3L2lqzZk2KeU3Ku/yc8nCsIkbptCTp8nzK\nsFIu9TM7AKVnXoOSbrRsUCLm9WfNmpViSsVN5T9J2r17d9YW2y5VFOMcse/xbzyfcj6levarW6U1\nSu28Jq0YnNMo4dM+wnZ5DsceZXD2rZSpgP2ixSRmEuFa8jjOV6mP8fq0ktB6w/miNSDeXxzzunXr\nOn7Oa3C9mHlFyvc7s59w7XnfRmvDqFGjUsysD7TScFzsC+cr9o39GkwFOfZdkvr6+lLcrJezSRhj\nzOnJoH4Zrqrqj6qqWldV1dNVVf1/VVW9paqqSVVVraiqanNVVd+qqqqz+dEYY0xP8Xe2McYMnqM+\nDFdV1Sfpf5d0WV3XcyWdKekTkv5K0n+v63qapOclffZkdtQYY8zR8Xe2McYcG4O1SQyTdG5VVb+W\n9FZJeyUtlfSfjvz9dklflHRLt0bquk5SLOVdSppSLjEzYf8dd9yRYkqqbGvGjBlZW5SxS4U6KHtT\n0o3yKmV0ZhoovWVOqVrKpd9SH0sWgvi2O6Vvyre7du3q2EfOMedLKsvLHFcpG0I8jueXZH+ez/mW\ncgtEqWgH16WU2UHKpXr2hdcvZe+Q8vnj+TFLSKfj457gdUqyPefoggsuyP5WKtJSslywaEW02HDO\neB9wT9E2wHkZPXp01hbnopSJpFvGj1gQpIH7gudwf3G+pXJhGO4jzj3vT+47Kb8n4703RDgh39nG\nGNMGjvrLcF3XuyX935J26NUv1BclPSnphbqumyeEXZL6OrdgjDGmV/g72xhjjo3B2CSGS/qQpEmS\nxko6T9Jvdzj09Yk5Xz3/c1VVPVFV1RODeZHFGGPMG+dEfmefvF4aY8ypw2BsEtdJ2lbX9bOSVFXV\ndyS9S9I7qqoaduSXhnGS9nQ6ua7rWyXdKkljxoypG2mS8ijfXJdeL+s2MCPBtm3bUsw3vymPSnkW\nBVoueI1SUv4odVOipbWCbVFqjvYPHhdl8AZKzbxGfJOdEnEpOwKP2b9/f8djpHxeKCnz+pTNo2zM\nOaedglL3xIkTU8w57jZHnMuSRSUWlChBa0jJ/hDhXLAvpcIgbCuuF+eCEjznmHMX93GpmAvXPs5l\nAwtwSPl685qlDAwcY8y6wLZLhWhKVggpnzPehzFjSQPXMf6Pa/6b+4V9pOWCcxotMjyuGfMQKrpx\nwr6zq6rq+MBsjDGnE4PJJrFD0pVVVb21evUp4r2S1kv6kaSbjhzzGUl3nZwuGmOMOQb8nW2MMcfA\nYDzDKyTdIekpSWuPnHOrpD+R9F+rqtoi6UJJ/3wS+2mMMWYQ+DvbGGOOjUFlk6jr+guSvhA+7pe0\n6IT3yBhjzHHh72xjjBk8Pa1A97Of/SxViKOPj55fSZo2bVqKn3/++RQfPHgwa6uhv78/xTFFEv2p\n8+bNS3Gpgly3NFDsCxkYGEgx/aTxfPpDWd2O/lJ6oem7jOm86PUsVbajJ5J+TKbgitdnHzl3nK+Y\nXovXYV+4xpMmTUox/dKPPfZY1tbatWs79oVrFNe4Uz+kfM44ZlZBK3l2pdwDW/JiM+Y8xvXi36I/\ntYHjin0pVfljv0oV0jjGeH3OGavBsV3u++hLphed/adHnPPN6pLxHO4djp++YvqCY2o1tlXyHLOt\nnTt3pjjeE53mqNSmMcaYoc2gKtAZY4wxxhhzOuKHYWOMMcYY01p6apN45ZVXUrqqKN2SUj5iSpys\n5kbLQUmClvJUWZQ8+Tml4lhBjv2Kcm8D01a99NJL2d8o/dJ2UJLamV4rVmrj/LH/hDaNUjotKU99\nRXmc12cFtG4WAB7HynZsd9OmTSnesWNH1hZleM4/15hWlIsuuijF0ZbCimR79+5NMeerZH+Q8jXi\ncdwH/Jx9j+nMmBKMdgCuK6X6mMaLfeM1uZa8Ji0DseIg9zivyfuLFhmuIy0PUj6XbKtUyTCmjOPf\naIPiuHiNkSNHpjimeeM+5vcA9yvH1c1606myYcmGYowxZmjjX4aNMcYYY0xr8cOwMcYYY4xpLW+a\nTYIS5a5du7LjKE+X3jAvvVHPzA5SLq8zpjzLmNJrlEqfffbZjn0klLMpL0u5zMpxMTMG7RvdMgWU\nJG1COwGPL1Uqk/I5oiTN/u7bty87p3R9Xoc2DUrS0U7A6zBmv2if4DrQFiGVq+5xLrrN42DsCDyf\nmRmizaFUVbHUr2ixof2Ee6xkN6K1INpadu/enWLee7Qece2YPSRm7OC/OZaSjSdmZOA50QrUcOGF\nF6a4ZMWQ8v3C6zOmTYN9ifPNPjd7r1u1QmOMMUMX/zJsjDHGGGNaix+GjTHGGGNMa6lisvmTybnn\nnltPnjxZUi5XUgKXcumUVoUZM2akmBYESshR6qYMTOmzVKiiVMwiEuX9BloIuiXyL9ksKK/TWsBY\nyqVvyrfMHMC5i1I5KRUAoWWEcniUp3lNWiCmT5+eYo6Lb/3HQiZcP2YXKGVtYNytKALnqHRctDJw\nv1Cq57zy+pzHOEfRNtHAPcG5j1kbOJc8jnPJe4pZNmJfaJ9hZgnuSVpMumXJoJWGMYuncB/FbC9c\nl1L2DsbcHzFLBteL7dJ+QUsSj4n3Oq/T2FK2bNmil19+OU85cppTVVXv/gNhjDEnmLquB/Wd7V+G\njTHGGGNMa/HDsDHGGGOMaS09zSYhvSZRU96NxDfWG/gWfWO3kKRp06aleMuWLdk5LLZA20JJAqed\nIPaRUiol4TFjxnT8fNWqVdn5lIt5Tcq17CPlfMreUi5dUwamzYPzFQttEFob2EeOhefPmjUrO5/Z\nHcaNG5fia665VqIxugAADY1JREFUJsWU8Fmw5Otf/3rW1sqVK1PMN/xpE6FNgAUwYvaP0lxw7nhO\ntLVwH3ItStkoeI1ogyllMuH1eb1oa2HbJSsN14GZNWLGCbbFzBK0BvCcUmEQKZ9/9oW2js2bN3e8\nhpSvJS0Y3COldewG57WUoaSUqUbK16/pYzfblDHGmKGLv92NMcYYY0xr8cOwMcYYY4xpLT21SRw+\nfDhlD+Cb4FF+pHRKeXrjxo0p5lvwY8eOTTGlYimX+ilDlzIw8M31+LY6JVq+kb9jx44UU96lFUPK\n5XlehxYC9p9WkJiNgBYOtsu2KGEza0OUmvmG/ZIlS1LMjAI8JhZH+MQnPtHxmmvWrOnY1ooVK1Ic\nC66wn1yjaIFo4N6J813KGsG2GEc7ATMfcL65FtyrzH4RsybQHsBrso+cozheZinh37gPOXe0WXBP\nSHlxDV6f59DWwTnm+sbzORfck9xvvHY8jnYd9pl2Ga5RvNc5F+wX14vX4xrFjBuc48Yy0S1biTHG\nmKGLfxk2xhhjjDGtxQ/DxhhjjDGmtfhh2BhjjDHGtJaeV6CbOHGipNwfGP2V7BNTHNHTSH8gUzI1\n7TfQx0jPMM9hWiV6YxlLedox+gfZR54TvdD0Ppa8mn19fSmm1zH2hXC+eA49tDwm9ovV/ObOnZvi\nkj9z4cKFxevTE0r/NH3V9ALv3Lkza4upvth/+k55fqkanZTvK55D3yvnK+5D+ro5xlK7XNPoLy3t\nXbbFa3CvxeO4FvS1lzzp0ePN9WY/S75y+vPjuOjtpc+2lHItVvlje6wkyT5zLp599tmO15PyteA9\nzX08mMqPUn6PNH7vgYEBHTp0yBXojDFmiOAKdMYYY4wxxhwFPwwbY4wxxpjW0tPUalVVJbm3JFtL\n5UpPlIF5PqXPmDaMbb344ospZtU4tsUKaoylcuWvUr+ijMuqczyHVg6eT2mdto4IpWZaE2gLYYq5\nKJvTgkErBy0bnDumRpPyeerv708x15XXp1Q+MDCQtRVl9E5t8RjaB+J8Ux4vpdGi5SBaAAZTGbC0\nV7tZf7iPmHKNY+SekMqpzrjfuUc49miFYpo4/q2Uvo6fx3uCKd9oaylVd4vpCrnHGbNfF1xwQYq5\ndrGaHcdFGxbXi/cE9z3XWsrXr7FvxDSAxhhjTg/8y7AxxhhjjGktfhg2xhhjjDGtpac2ibquk8zZ\nLYsFZVnKwDFbQAMl5AjlZkqfe/fuTTHfrt++fXuKmQFByjMlUNItWR54jJRL1xwj5V1ek2/qR9n9\noosuSjFlYLZLeZoy8IEDB7K22GdK3c/8/+2dS8hdVxXHfysvaiuhVmOoSbUtFGsRtFKkPhCpDnwU\n60BRsVDEzgSrKFKdOXAgiI+BCNIqHYgPYsHiQJDagaOgNQMfn+JHfTQxppFYFQc2xeXgnPv1fzd3\n3wThOzc56/+b5Nz7nXP2Xnvts9i563/W3tpaeS+tLgDL0gjts9qrKW1tv32LX8dMU/CKygF0XLTS\nACzLDtR+TeGr71p/adUHpVftRP3Qyj30s16j/dd53Kbte3IItaX3TLWymJ796iPtl/pRZTiwLKvp\nSUF07rT+1ut1jFTOoBIdrXih94Vlf2lfes9dL57A8hgtrmklWMYYY+aBfxk2xhhjjDFl8WLYGGOM\nMcaUZVKZxJ49e3bS6Jq6bNOVmkbWVGgv3appVD2G5TfR9XpNaWt6WFOtbZpc5QSa0u2lYds359VO\nlWmo/KK3uYPaActyDrVL07vaf227Tafr31TCoClslT+cOXNm6XqtMKApfH2jX8dY5QBtX9R/WrFD\nbVEJgabNdRxhOdXe873Oo3XShra6w4Je1YR2c4deZQvti9quG1DAsuRDjzV1r/1Vu1r5iM4lraig\n/tKx621S0rapc18lJ3p9W7FDr9E2Va7T25ij7Yteo/ftya7aZ0rRcVnMsV6lE2OMMZc3/mXYGGOM\nMcaUxYthY4wxxhhTlkllEvBcWllTn+1b4ZqS1xS+psHbaxa06WxNPbdv1a9C0/btZiDaL0VTwpqS\nbfvSbhKw6jxNpyttirYnM9Fx7W0A0kv5w3J6WM/T69txUDnF9vb2zvHhw4d3jjUFrxIRlULAsl3a\nTi8FrpKLNu2t92rT8xf6HvrVCdQXmnbXtH27gYWe10vha1WTdRVSdLxViqPSDL2vbvYCy3KEXrWV\nnnyklX9oJRRtU6/X527dxiY6R3p26TxoZRJtBY4FKmXRa9ZVk1j3N2OMMfPCvwwbY4wxxpiyeDFs\njDHGGGPKMqlM4vz58zuVCFTm0KaUNa2p6WV9k1zT+Zr6XPe2uqa6NaWq99L2NJ286P8ClVDoJhSa\nwm5TrefOnVvZZk8aoX1v09NKrwqA2qsSjYMHDy5drzIP9Yuep/dqK3b0+qxjrGl/3UyjTW1rql7/\npn3Rc9QPOm9gWQLQG5dear09ryff0LmnPmqrZGjf9Fjn66lTp3aO20okaktvIxo9R+/bjkvv2enJ\nN3SM1slK9PnoVZNox0X7rM+HojZqX1rpkP5Nx0/tV5lFT/bUtrkYo3UbBRljjLl88S/DxhhjjDGm\nLF4MG2OMMcaYsngxbIwxxhhjyjKpZnj//v075bZUu9dqZns6Xz1WDamWXGvLLfVKVOl5qsFUPaxq\nU9f1S/W4PS0yLNus56n9qtntlTaD/u5qvfJYujPcOo22Xq+6yd450B8/1dD27qvaUujb3NtBrr1e\n0XZUX6pj3yuN1v5N/a3lxPQaHaNWX6r+Un/3+rVOy6z3WqcFX9DqfLXN3rhq+zqOqomH5fFXm7Wc\nms77to+9Xfu0nd4Oi63vdY71NM89HXjrr1U7X7baa2OMMfPA0d0YY4wxxpTFi2FjjDHGGFOWje1A\np2XS2hSlpku1LJOmrS+23NOqEkkXS5uq1nb0WNPmvd3oYDkt20vR6q5nWh5KZQ7rztP0rvZFU8ht\nmTaVmai0oyd5aFPdmgZXdEc4vZem6dtrezu1aYq6JyFo/dsrNabHWl6rleu0pbtW9aX3fTsn1a6e\n/EK/byU26n+VRqhkQuehzo/WDpUO6dzp7b6obbe72WmfVdqgvjh06NDOcTv3tG+950DntPponWxB\n76XzSOUben3bL312FjaqrcYYY+aDfxk2xhhjjDFl8WLYGGOMMcaUJabcVSkizgL/Bv42WaOXHi+i\nrv2VbYfa9s/B9pdl5qELnzYfxpj9J+bhv/8X216XyvbPwfaLjtmTLoYBIuLnmXnbpI1eQlS2v7Lt\nUNv+yrbPgcr+s+01bYfa9lez3TIJY4wxxhhTFi+GjTHGGGNMWTaxGP76Btq8lKhsf2Xbobb9lW2f\nA5X9Z9vrUtn+UrZPrhk2xhhjjDHmUsEyCWOMMcYYU5ZJF8MR8baI+F1EbEfE/VO2PTURcV1EPBYR\nWxHx64i4b/z+moj4cUT8fvz3BRe61+VKROyNiBMR8cPx8w0RcXy0/bsRceBC97hciYirI+JYRPx2\nnAOvq+L7iPj4OOd/FRHfjogrKvl+Tjhm14rZUDduV47Z4Lg92WI4IvYCXwXeDtwCfCAibpmq/Q3w\nLPCJzHwFcDvwkdHe+4FHM/Mm4NHx81y5D9iSz58HvjTa/nfgwxvp1TR8BfhRZt4MvIphHGbv+4g4\nAnwUuC0zXwnsBd5PLd/PAsfskjEb6sbtkjEbHLdh2l+GXwtsZ+YTmfkM8B3grgnbn5TMPJ2ZvxiP\n/8XwYB1hsPmh8bSHgHdvpoe7S0QcBd4JPDB+DuAO4Nh4ypxtPwi8CXgQIDOfycynKeJ7YB/wvIjY\nB1wJnKaI72eGY3ahmA1147ZjNlA8bk+5GD4CPCmfT47fzZ6IuB64FTgOHM7M0zAEX+DFm+vZrvJl\n4FPAf8fPLwSezsxnx89z9v+NwFngm2O68YGIuIoCvs/MU8AXgD8zBNN/AI9Tx/dzwjG7VsyGunG7\nbMwGx22YdjEcK76bfSmLiHg+8H3gY5n5z033Zwoi4k7gqcx8XL9ecepc/b8PeA3wtcy8lWEL8lmm\n11pGTd1dwA3AS4CrGNLsLXP1/Zyo9MzuUDFmQ/m4XTZmg+M2TLsYPglcJ5+PAn+ZsP3JiYj9DEH1\nW5n58Pj1mYi4dvz7tcBTm+rfLvIG4F0R8UeG1OodDL84XD2mYGDe/j8JnMzM4+PnYwyBtoLv3wr8\nITPPZuZ54GHg9dTx/ZxwzB6o8NxC7bhdOWaD4/aki+GfATeNbyceYBBnPzJh+5Myaq0eBLYy84vy\np0eAe8bje4AfTN233SYzP52ZRzPzegY//yQzPwg8BrxnPG2WtgNk5l+BJyPi5eNXbwF+QwHfM6TZ\nbo+IK8dnYGF7Cd/PDMfsgQrPbem4XTxmg+P2tJtuRMQ7GP6nuRf4RmZ+brLGJyYi3gj8FPglz+mv\nPsOgQfse8FKGCfjezDy3kU5OQES8GfhkZt4ZETcy/OJwDXACuDsz/7PJ/u0WEfFqhpdQDgBPAB9i\n+M/n7H0fEZ8F3sfwdv4J4F4GrVkJ388Jx+x6MRtqxu3KMRsct70DnTHGGGOMKYt3oDPGGGOMMWXx\nYtgYY4wxxpTFi2FjjDHGGFMWL4aNMcYYY0xZvBg2xhhjjDFl8WLYGGOMMcaUxYthY4wxxhhTFi+G\njTHGGGNMWf4H5m+6agO+AGcAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f0d2b8f5a10>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "width = 12\n",
    "height = 12\n",
    "plt.figure(figsize=(width, height))\n",
    "plt.subplot(121)\n",
    "plt.imshow(post_process(batch.data[0][idx]).asnumpy(), cmap=plt.cm.gray)\n",
    "plt.subplot(122)\n",
    "plt.imshow(post_process(batch.data[0][idx]).asnumpy()*post_process_mask(batch.label[0][idx]), cmap=plt.cm.gray)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Predicted"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 511,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f0d2ae17a50>"
      ]
     },
     "execution_count": 511,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAsMAAAFTCAYAAADcAgGcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f0d2b646a90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "width = 12\n",
    "height = 12\n",
    "plt.figure(figsize=(width, height))\n",
    "plt.subplot(121)\n",
    "plt.imshow(post_process(batch.data[0][idx]).asnumpy(), cmap=plt.cm.gray)\n",
    "plt.subplot(122)\n",
    "plt.imshow(post_process(batch.data[0][idx]).asnumpy()*post_process_mask(outputs[idx]), cmap=plt.cm.gray)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "These results are qualitatively promising. However, while highly specific this model isn't very sensitive. Further exploration can be done to select a more qualitatively pleasing model at an earlier iteration that balances positive / negative rates out."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "MXNet 1.0",
   "language": "python",
   "name": "mxnet1"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}