test_yolo_confidence.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Dataset functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import pandas as pd\n",
    "import SimpleITK as sitk\n",
    "\n",
    "# this data loader is in a file now\n",
    "class Abus23DataLoader():\n",
    "    def __init__(self, dataset_path, labels_csv=None):\n",
    "        self.data = self.load_abus23(dataset_path, labels_csv)\n",
    "        self.dataset_path = dataset_path\n",
    "        self.used_data = self.data\n",
    "        self.cidx = 0\n",
    "        \n",
    "    def load_abus23(self, dataset_path, label_file=None):\n",
    "        if label_file is not None:\n",
    "            dataset = pd.read_csv(os.path.join(dataset_path, label_file))\n",
    "        else:\n",
    "            files = [f for f in os.listdir(dataset_path) if f.endswith(\".nrrd\")]\n",
    "            files_id = [int(f.split(\".\")[0].split(\"_\")[-1]) for f in files]\n",
    "            files_path = [f for f in files] #os.path.join(dataset_path, f) \n",
    "            dataset_dict = [{\"case_id\": f_id, \"data_path\": f_path} for f_id, f_path in zip(files_id, files_path)]\n",
    "            dataset = pd.DataFrame.from_dict(dataset_dict)\n",
    "        print(\"Dataset columns:\", dataset.columns)\n",
    "        return dataset\n",
    "    \n",
    "    def set_subset_ids(self, list_id = [], id_label = 'case_id'):\n",
    "        if list_id:\n",
    "            self.used_data = self.data[self.data[id_label].isin(list_id)]\n",
    "        \n",
    "    def get_data_entry(self, idx):    \n",
    "        return self.used_data.iloc[idx]\n",
    "    \n",
    "    def get_item(self, idx):\n",
    "        entry = self.get_data_entry(idx).to_dict()\n",
    "        output = {}\n",
    "        output[\"id\"] = entry['case_id']\n",
    "        if 'label' in entry:\n",
    "            output[\"class\"] =entry['label']\n",
    "        image_full_path = os.path.join(self.dataset_path, entry['data_path'].replace('\\\\','/'))\n",
    "        output[\"image\"] = sitk.ReadImage(image_full_path)\n",
    "        output[\"image_path\"] = image_full_path\n",
    "        if 'mask_path' in entry:\n",
    "            mask_full_path = os.path.join(self.dataset_path, entry['mask_path'].replace('\\\\','/'))\n",
    "            output[\"mask\"] = sitk.ReadImage(mask_full_path)\n",
    "            output[\"mask_path\"] = mask_full_path\n",
    "            \n",
    "        return output\n",
    "\n",
    "    def get_keys(self):\n",
    "        return self.used_data.columns.tolist()\n",
    "        \n",
    "    def __getitem__(self, idx):\n",
    "       return self.get_item(idx)\n",
    "   \n",
    "    def __len__(self):\n",
    "        return len(self.used_data)\n",
    "   \n",
    "def get_validation_ids(val_file):\n",
    "    with open(val_file) as fp:\n",
    "        lines = fp.readlines()\n",
    "        #print([os.path.basename(i)for i in lines])\n",
    "        patients = set([int(os.path.basename(case).split('_')[0]) for case in lines])\n",
    "    return list(patients)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Data processing functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "import SimpleITK as sitk\n",
    "import numpy as np\n",
    "import cv2\n",
    "\n",
    "def normalize_8bits(image: np.ndarray):\n",
    "    return (255.0 *(image - image.min()) / (image.max() - image.min())).astype(np.uint8)\n",
    "\n",
    "def get_slices(data, norm_fn = normalize_8bits):\n",
    "    data_array = sitk.GetArrayFromImage(data)\n",
    "    return [norm_fn(data_array[i, ...]) for i in range(len(data_array))]\n",
    "\n",
    "\n",
    "# Create the volume from slices\n",
    "\n",
    "def volume_from_slice(slices, format=None):\n",
    "    mask_3d = np.stack(slices)\n",
    "    output_mask = sitk.GetImageFromArray(mask_3d)\n",
    "    \n",
    "    if format is not None:\n",
    "        #sitk.GetPixelIDValueAsString(format)\n",
    "        castImageFilter = sitk.CastImageFilter()\n",
    "        castImageFilter.SetOutputPixelType(format)\n",
    "        output_mask = castImageFilter.Execute(output_mask)\n",
    "        \n",
    "    return output_mask"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Prediction functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from ultralytics import YOLO\n",
    "import numpy as np\n",
    "import cv2\n",
    "\n",
    "# Load a model\n",
    "class YOLOPredictor:\n",
    "    def __init__(self, model_file, conf_th = 0.5):\n",
    "        self.model = YOLO(model_file)  # pretrained YOLOv8n model\n",
    "        self.conf_th = conf_th\n",
    "        \n",
    "    def set_conf_th(self, conf_th = 0.5):\n",
    "        self.conf_th = conf_th\n",
    "        \n",
    "    def __call__(self, slice, conf_th=None):\n",
    "        return self.predict(slice, conf_th)      \n",
    "        \n",
    "    def predict(self, slice, conf_th=None):\n",
    "        assert len(slice.shape) == 2\n",
    "        \n",
    "        if conf_th is None:\n",
    "            conf_th = self.conf_th\n",
    "        \n",
    "        cv2.imwrite(\"temp.png\", slice)\n",
    "        results = self.model(\"temp.png\", verbose=False)[0].cpu().numpy()\n",
    "        \n",
    "        slice_mask = np.zeros(slice.shape)\n",
    "        if results.masks is not None:\n",
    "            \n",
    "            pred_mask_data = results.masks.data\n",
    "            for i in range(len(pred_mask_data)):\n",
    "                \n",
    "                pred_box_conf = results.boxes[i].conf  # confidence score, (N, )\n",
    "                if pred_box_conf < conf_th:\n",
    "                    continue\n",
    "\n",
    "                m = cv2.resize(pred_mask_data[i, ...], dsize=(slice.shape[1], slice.shape[0])) # interpolation=cv2.INTER_CUBIC)\n",
    "                slice_mask = np.logical_or(slice_mask, m).astype(\"float32\")\n",
    "                \n",
    "        return slice_mask       "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Run inference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset columns: Index(['case_id', 'label', 'data_path', 'mask_path'], dtype='object')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 20/20 [43:37<00:00, 130.86s/it]\n"
     ]
    }
   ],
   "source": [
    "from ultralytics import YOLO\n",
    "import json\n",
    "from tqdm import tqdm\n",
    "\n",
    "# ABUS 23\n",
    "dataset_path = \"datasets/Train\" #\"datasets/DATA\"\n",
    "label_file = \"labels.csv\"\n",
    "validation_file = \"datasets/abus23_25_png/val_seg.txt\"\n",
    "\n",
    "# Yolo model\n",
    "yolo_weights = \"/home/joel/abus23/runs/segment/train10/weights/best.pt\"\n",
    "\n",
    "# Volume\n",
    "confidance_th = np.arange(0.1,1,0.1)\n",
    "# Output folder\n",
    "output_folder = os.path.join(\"results_masks\", \"abus23_25\", \"raw_stack_train10\")\n",
    "output_file_template = \"MASK_{:0>3}.nii.gz\"\n",
    "output_format = sitk.sitkUInt8\n",
    "\n",
    "# Create output folder\n",
    "os.makedirs(output_folder, exist_ok=True)\n",
    "\n",
    "for conf in confidance_th:\n",
    "        os.makedirs(os.path.join(output_folder, f\"{int(conf*100)}\"), exist_ok=False)\n",
    "\n",
    "# Load dataset\n",
    "dataset = Abus23DataLoader(dataset_path, label_file)\n",
    "\n",
    "# Get validation cases\n",
    "if validation_file is not None:\n",
    "        val_ids = get_validation_ids(validation_file)\n",
    "        dataset.set_subset_ids(val_ids)\n",
    "\n",
    "# Load yolo predictor\n",
    "yolo_predictor = YOLOPredictor(yolo_weights)\n",
    "\n",
    "# For each item in the dataset\n",
    "gt_files = []\n",
    "pred_files = {conf: [] for conf in confidance_th}\n",
    "for item in tqdm(dataset):\n",
    "        \n",
    "        # Get image slice\n",
    "        image_slices = get_slices(item['image'])\n",
    "        \n",
    "        for conf in confidance_th:\n",
    "                # Get predicted slices\n",
    "                yolo_masks_slices = [yolo_predictor(slice, conf_th=conf) for slice in image_slices]\n",
    "\n",
    "                # Create 3D volum\n",
    "                mask_volum = volume_from_slice(yolo_masks_slices, output_format)               \n",
    "                \n",
    "                # Copy metadata from predited image\n",
    "                mask_volum.CopyInformation(item['image'])\n",
    "                \n",
    "                # Save NRRD mask prediction\n",
    "                mask_file = os.path.join(output_folder, f\"{int(conf*100)}\", output_file_template.format(item['id']))\n",
    "                sitk.WriteImage(mask_volum, mask_file, useCompression=True )\n",
    "                \n",
    "                # Save file names for evaluation\n",
    "                pred_files[conf].append(mask_file)\n",
    "                \n",
    "        if 'mask_path' in item:\n",
    "                gt_files.append(item['mask_path'])\n",
    "                \n",
    "with open(os.path.join(output_folder,\"cases.json\"), \"w\") as fp:\n",
    "        json.dump({\"gt_files\": gt_files,\n",
    "                   \"pred_files\": pred_files,\n",
    "                   \"confidance_th\": confidance_th},fp)\n",
    "        "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Validation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from TDSCABUS2023.Metrics import segmentation\n",
    "\n",
    "\n",
    "def Validate(pred_list, gt_list, cvs_pred_file = None, csv_gt_file = None):\n",
    "    \n",
    "    print(\"Segmentation:\")\n",
    "    print(\"------------------------------------------\")\n",
    "    \n",
    "    scores = {'DiceCoefficient': [], 'HDCoefficient': [], 'score': []}\n",
    "    for pred, gt in zip(pred_list, gt_list):\n",
    "        try:\n",
    "            result = segmentation.score_case(gt, pred)\n",
    "        except Exception as e:\n",
    "           result = {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0} #HD coefficient if fail?\n",
    "        print(\"Case:\", os.path.basename(pred), \"  Results:\",  result)\n",
    "        \n",
    "        for k, v in result.items():\n",
    "            scores[k].append(v)\n",
    "        \n",
    "    for k, values in scores.items():\n",
    "        values = np.array(values)\n",
    "        print(f\"\\n{k}:\")\n",
    "        print(f\"   - Min: {values.min():0.4f}\")\n",
    "        print(f\"   - Max: {values.max():0.4f}\")\n",
    "        print(f\"   - Mean: {values.mean():0.4f}\")\n",
    "        \n",
    "    return {k: np.mean(v) for k, v in result.items()}\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{0.1: ['results_masks/abus23_25/raw_stack_train10/10/MASK_007.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_009.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_010.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_014.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_022.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_025.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_031.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_039.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_053.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_061.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_062.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_066.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_067.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_074.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_075.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_080.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_087.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_089.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_090.nii.gz', 'results_masks/abus23_25/raw_stack_train10/10/MASK_099.nii.gz'], 0.2: ['results_masks/abus23_25/raw_stack_train10/20/MASK_007.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_009.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_010.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_014.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_022.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_025.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_031.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_039.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_053.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_061.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_062.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_066.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_067.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_074.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_075.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_080.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_087.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_089.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_090.nii.gz', 'results_masks/abus23_25/raw_stack_train10/20/MASK_099.nii.gz'], 0.30000000000000004: ['results_masks/abus23_25/raw_stack_train10/30/MASK_007.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_009.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_010.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_014.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_022.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_025.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_031.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_039.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_053.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_061.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_062.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_066.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_067.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_074.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_075.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_080.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_087.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_089.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_090.nii.gz', 'results_masks/abus23_25/raw_stack_train10/30/MASK_099.nii.gz'], 0.4: ['results_masks/abus23_25/raw_stack_train10/40/MASK_007.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_009.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_010.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_014.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_022.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_025.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_031.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_039.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_053.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_061.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_062.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_066.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_067.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_074.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_075.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_080.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_087.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_089.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_090.nii.gz', 'results_masks/abus23_25/raw_stack_train10/40/MASK_099.nii.gz'], 0.5: ['results_masks/abus23_25/raw_stack_train10/50/MASK_007.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_009.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_010.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_014.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_022.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_025.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_031.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_039.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_053.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_061.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_062.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_066.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_067.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_074.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_075.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_080.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_087.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_089.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_090.nii.gz', 'results_masks/abus23_25/raw_stack_train10/50/MASK_099.nii.gz'], 0.6: ['results_masks/abus23_25/raw_stack_train10/60/MASK_007.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_009.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_010.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_014.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_022.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_025.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_031.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_039.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_053.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_061.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_062.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_066.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_067.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_074.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_075.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_080.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_087.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_089.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_090.nii.gz', 'results_masks/abus23_25/raw_stack_train10/60/MASK_099.nii.gz'], 0.7000000000000001: ['results_masks/abus23_25/raw_stack_train10/70/MASK_007.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_009.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_010.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_014.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_022.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_025.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_031.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_039.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_053.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_061.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_062.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_066.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_067.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_074.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_075.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_080.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_087.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_089.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_090.nii.gz', 'results_masks/abus23_25/raw_stack_train10/70/MASK_099.nii.gz'], 0.8: ['results_masks/abus23_25/raw_stack_train10/80/MASK_007.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_009.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_010.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_014.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_022.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_025.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_031.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_039.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_053.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_061.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_062.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_066.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_067.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_074.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_075.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_080.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_087.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_089.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_090.nii.gz', 'results_masks/abus23_25/raw_stack_train10/80/MASK_099.nii.gz'], 0.9: ['results_masks/abus23_25/raw_stack_train10/90/MASK_007.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_009.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_010.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_014.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_022.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_025.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_031.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_039.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_053.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_061.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_062.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_066.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_067.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_074.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_075.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_080.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_087.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_089.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_090.nii.gz', 'results_masks/abus23_25/raw_stack_train10/90/MASK_099.nii.gz']}\n",
      "['datasets/Train/MASK/MASK_007.nrrd', 'datasets/Train/MASK/MASK_009.nrrd', 'datasets/Train/MASK/MASK_010.nrrd', 'datasets/Train/MASK/MASK_014.nrrd', 'datasets/Train/MASK/MASK_022.nrrd', 'datasets/Train/MASK/MASK_025.nrrd', 'datasets/Train/MASK/MASK_031.nrrd', 'datasets/Train/MASK/MASK_039.nrrd', 'datasets/Train/MASK/MASK_053.nrrd', 'datasets/Train/MASK/MASK_061.nrrd', 'datasets/Train/MASK/MASK_062.nrrd', 'datasets/Train/MASK/MASK_066.nrrd', 'datasets/Train/MASK/MASK_067.nrrd', 'datasets/Train/MASK/MASK_074.nrrd', 'datasets/Train/MASK/MASK_075.nrrd', 'datasets/Train/MASK/MASK_080.nrrd', 'datasets/Train/MASK/MASK_087.nrrd', 'datasets/Train/MASK/MASK_089.nrrd', 'datasets/Train/MASK/MASK_090.nrrd', 'datasets/Train/MASK/MASK_099.nrrd']\n"
     ]
    }
   ],
   "source": [
    "gt_files = [g.replace(\"DATA\", \"MASK\") for g in gt_files]\n",
    "print(pred_files)\n",
    "print(gt_files)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Validating..\n",
      "Confidance: 0.1\n",
      "Segmentation:\n",
      "------------------------------------------\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Case: MASK_007.nii.gz   Results: {'DiceCoefficient': 0.243705660618637, 'HDCoefficient': 293.7567020512043, 'score': -293.51299639058567}\n",
      "Case: MASK_009.nii.gz   Results: {'DiceCoefficient': 0.012826633431208939, 'HDCoefficient': 291.79787524928963, 'score': -291.78504861585844}\n",
      "Case: MASK_010.nii.gz   Results: {'DiceCoefficient': 0.4042952485307301, 'HDCoefficient': 308.4185467834255, 'score': -308.01425153489475}\n",
      "Case: MASK_014.nii.gz   Results: {'DiceCoefficient': 0.07309371907086526, 'HDCoefficient': 379.4970355615443, 'score': -379.4239418424734}\n",
      "Case: MASK_022.nii.gz   Results: {'DiceCoefficient': 0.7425240719675729, 'HDCoefficient': 206.4654934849889, 'score': -205.72296941302133}\n",
      "Case: MASK_025.nii.gz   Results: {'DiceCoefficient': 0.5590613494821595, 'HDCoefficient': 173.44163283364233, 'score': -172.88257148416017}\n",
      "Case: MASK_031.nii.gz   Results: {'DiceCoefficient': 0.3248270119304658, 'HDCoefficient': 260.679880313, 'score': -260.3550533010696}\n",
      "Case: MASK_039.nii.gz   Results: {'DiceCoefficient': 0.2428690547475279, 'HDCoefficient': 343.9316792620302, 'score': -343.68881020728264}\n",
      "Case: MASK_053.nii.gz   Results: {'DiceCoefficient': 0.11805596639649767, 'HDCoefficient': 200.90047287151916, 'score': -200.78241690512266}\n",
      "Case: MASK_061.nii.gz   Results: {'DiceCoefficient': 0.3153384209549356, 'HDCoefficient': 309.3121400785944, 'score': -308.9968016576394}\n",
      "Case: MASK_062.nii.gz   Results: {'DiceCoefficient': 0.37746949485217623, 'HDCoefficient': 196.72823894906395, 'score': -196.3507694542118}\n",
      "Case: MASK_066.nii.gz   Results: {'DiceCoefficient': 0.7237506887574379, 'HDCoefficient': 156.16977940690063, 'score': -155.4460287181432}\n",
      "Case: MASK_067.nii.gz   Results: {'DiceCoefficient': 0.3428009198890443, 'HDCoefficient': 272.9725260900811, 'score': -272.62972517019205}\n",
      "Case: MASK_074.nii.gz   Results: {'DiceCoefficient': 0.09836239712740355, 'HDCoefficient': 236.73825208444873, 'score': -236.63988968732133}\n",
      "Case: MASK_075.nii.gz   Results: {'DiceCoefficient': 0.8835480382066622, 'HDCoefficient': 187.47533171060132, 'score': -186.59178367239466}\n",
      "Case: MASK_080.nii.gz   Results: {'DiceCoefficient': 0.4952417398742761, 'HDCoefficient': 391.32084023215526, 'score': -390.82559849228096}\n",
      "Case: MASK_087.nii.gz   Results: {'DiceCoefficient': 0.570543748745125, 'HDCoefficient': 277.8506793225455, 'score': -277.28013557380035}\n",
      "Case: MASK_089.nii.gz   Results: {'DiceCoefficient': 0.8558869015599785, 'HDCoefficient': 135.2553141285029, 'score': -134.39942722694292}\n",
      "Case: MASK_090.nii.gz   Results: {'DiceCoefficient': 0.042419744594582365, 'HDCoefficient': 335.1492204973779, 'score': -335.1068007527833}\n",
      "Case: MASK_099.nii.gz   Results: {'DiceCoefficient': 0.6131719339843384, 'HDCoefficient': 254.21447637772323, 'score': -253.60130444373888}\n",
      "\n",
      "DiceCoefficient:\n",
      "   - Min: 0.0128\n",
      "   - Max: 0.8835\n",
      "   - Mean: 0.4020\n",
      "\n",
      "HDCoefficient:\n",
      "   - Min: 135.2553\n",
      "   - Max: 391.3208\n",
      "   - Mean: 260.6038\n",
      "\n",
      "score:\n",
      "   - Min: -390.8256\n",
      "   - Max: -134.3994\n",
      "   - Mean: -260.2018\n",
      "Confidance: 0.2\n",
      "Segmentation:\n",
      "------------------------------------------\n",
      "Case: MASK_007.nii.gz   Results: {'DiceCoefficient': 0.243705660618637, 'HDCoefficient': 293.7567020512043, 'score': -293.51299639058567}\n",
      "Case: MASK_009.nii.gz   Results: {'DiceCoefficient': 0.012826633431208939, 'HDCoefficient': 291.79787524928963, 'score': -291.78504861585844}\n",
      "Case: MASK_010.nii.gz   Results: {'DiceCoefficient': 0.4042952485307301, 'HDCoefficient': 308.4185467834255, 'score': -308.01425153489475}\n",
      "Case: MASK_014.nii.gz   Results: {'DiceCoefficient': 0.07309371907086526, 'HDCoefficient': 379.4970355615443, 'score': -379.4239418424734}\n",
      "Case: MASK_022.nii.gz   Results: {'DiceCoefficient': 0.7425240719675729, 'HDCoefficient': 206.4654934849889, 'score': -205.72296941302133}\n",
      "Case: MASK_025.nii.gz   Results: {'DiceCoefficient': 0.5590613494821595, 'HDCoefficient': 173.44163283364233, 'score': -172.88257148416017}\n",
      "Case: MASK_031.nii.gz   Results: {'DiceCoefficient': 0.3248270119304658, 'HDCoefficient': 260.679880313, 'score': -260.3550533010696}\n",
      "Case: MASK_039.nii.gz   Results: {'DiceCoefficient': 0.2428690547475279, 'HDCoefficient': 343.9316792620302, 'score': -343.68881020728264}\n",
      "Case: MASK_053.nii.gz   Results: {'DiceCoefficient': 0.11805596639649767, 'HDCoefficient': 200.90047287151916, 'score': -200.78241690512266}\n",
      "Case: MASK_061.nii.gz   Results: {'DiceCoefficient': 0.3153384209549356, 'HDCoefficient': 309.3121400785944, 'score': -308.9968016576394}\n",
      "Case: MASK_062.nii.gz   Results: {'DiceCoefficient': 0.37746949485217623, 'HDCoefficient': 196.72823894906395, 'score': -196.3507694542118}\n",
      "Case: MASK_066.nii.gz   Results: {'DiceCoefficient': 0.7237506887574379, 'HDCoefficient': 156.16977940690063, 'score': -155.4460287181432}\n",
      "Case: MASK_067.nii.gz   Results: {'DiceCoefficient': 0.3428009198890443, 'HDCoefficient': 272.9725260900811, 'score': -272.62972517019205}\n",
      "Case: MASK_074.nii.gz   Results: {'DiceCoefficient': 0.09836239712740355, 'HDCoefficient': 236.73825208444873, 'score': -236.63988968732133}\n",
      "Case: MASK_075.nii.gz   Results: {'DiceCoefficient': 0.8835480382066622, 'HDCoefficient': 187.47533171060132, 'score': -186.59178367239466}\n",
      "Case: MASK_080.nii.gz   Results: {'DiceCoefficient': 0.4952417398742761, 'HDCoefficient': 391.32084023215526, 'score': -390.82559849228096}\n",
      "Case: MASK_087.nii.gz   Results: {'DiceCoefficient': 0.570543748745125, 'HDCoefficient': 277.8506793225455, 'score': -277.28013557380035}\n",
      "Case: MASK_089.nii.gz   Results: {'DiceCoefficient': 0.8558869015599785, 'HDCoefficient': 135.2553141285029, 'score': -134.39942722694292}\n",
      "Case: MASK_090.nii.gz   Results: {'DiceCoefficient': 0.042419744594582365, 'HDCoefficient': 335.1492204973779, 'score': -335.1068007527833}\n",
      "Case: MASK_099.nii.gz   Results: {'DiceCoefficient': 0.6131719339843384, 'HDCoefficient': 254.21447637772323, 'score': -253.60130444373888}\n",
      "\n",
      "DiceCoefficient:\n",
      "   - Min: 0.0128\n",
      "   - Max: 0.8835\n",
      "   - Mean: 0.4020\n",
      "\n",
      "HDCoefficient:\n",
      "   - Min: 135.2553\n",
      "   - Max: 391.3208\n",
      "   - Mean: 260.6038\n",
      "\n",
      "score:\n",
      "   - Min: -390.8256\n",
      "   - Max: -134.3994\n",
      "   - Mean: -260.2018\n",
      "Confidance: 0.30000000000000004\n",
      "Segmentation:\n",
      "------------------------------------------\n",
      "Case: MASK_007.nii.gz   Results: {'DiceCoefficient': 0.27349462821508014, 'HDCoefficient': 293.7567020512043, 'score': -293.4832074229892}\n",
      "Case: MASK_009.nii.gz   Results: {'DiceCoefficient': 0.010400755933522316, 'HDCoefficient': 291.79787524928963, 'score': -291.7874744933561}\n",
      "Case: MASK_010.nii.gz   Results: {'DiceCoefficient': 0.41681249435174456, 'HDCoefficient': 226.57890457851542, 'score': -226.16209208416367}\n",
      "Case: MASK_014.nii.gz   Results: {'DiceCoefficient': 0.0869253624378518, 'HDCoefficient': 379.4970355615443, 'score': -379.41011019910644}\n",
      "Case: MASK_022.nii.gz   Results: {'DiceCoefficient': 0.7126998067802565, 'HDCoefficient': 206.4654934849889, 'score': -205.75279367820866}\n",
      "Case: MASK_025.nii.gz   Results: {'DiceCoefficient': 0.5750062172785437, 'HDCoefficient': 143.7810835958611, 'score': -143.20607737858256}\n",
      "Case: MASK_031.nii.gz   Results: {'DiceCoefficient': 0.35027865482825116, 'HDCoefficient': 169.57004452437937, 'score': -169.2197658695511}\n",
      "Case: MASK_039.nii.gz   Results: {'DiceCoefficient': 0.2632324353503832, 'HDCoefficient': 236.02754076590298, 'score': -235.7643083305526}\n",
      "Case: MASK_053.nii.gz   Results: {'DiceCoefficient': 0.16188534980175026, 'HDCoefficient': 200.90047287151916, 'score': -200.7385875217174}\n",
      "Case: MASK_061.nii.gz   Results: {'DiceCoefficient': 0.3136725478329686, 'HDCoefficient': 309.3121400785944, 'score': -308.9984675307614}\n",
      "Case: MASK_062.nii.gz   Results: {'DiceCoefficient': 0.4198628155546758, 'HDCoefficient': 196.72823894906395, 'score': -196.30837613350928}\n",
      "Case: MASK_066.nii.gz   Results: {'DiceCoefficient': 0.7295690084593717, 'HDCoefficient': 156.16977940690063, 'score': -155.44021039844125}\n",
      "Case: MASK_067.nii.gz   Results: {'DiceCoefficient': 0.36156539134783694, 'HDCoefficient': 272.9725260900811, 'score': -272.61096069873327}\n",
      "Case: MASK_074.nii.gz   Results: {'DiceCoefficient': 0.10025932348696905, 'HDCoefficient': 236.73825208444873, 'score': -236.63799276096177}\n",
      "Case: MASK_075.nii.gz   Results: {'DiceCoefficient': 0.8945303303427992, 'HDCoefficient': 187.47533171060132, 'score': -186.58080138025852}\n",
      "Case: MASK_080.nii.gz   Results: {'DiceCoefficient': 0.5184155336145345, 'HDCoefficient': 306.5175362030695, 'score': -305.99912066945495}\n",
      "Case: MASK_087.nii.gz   Results: {'DiceCoefficient': 0.5830994424376675, 'HDCoefficient': 251.60683615514105, 'score': -251.0237367127034}\n",
      "Case: MASK_089.nii.gz   Results: {'DiceCoefficient': 0.8570930734063659, 'HDCoefficient': 122.00819644597652, 'score': -121.15110337257016}\n",
      "Case: MASK_090.nii.gz   Results: {'DiceCoefficient': 0.0446648941746, 'HDCoefficient': 335.1492204973779, 'score': -335.10455560320327}\n",
      "Case: MASK_099.nii.gz   Results: {'DiceCoefficient': 0.6576583755478399, 'HDCoefficient': 234.448288541418, 'score': -233.79063016587017}\n",
      "\n",
      "DiceCoefficient:\n",
      "   - Min: 0.0104\n",
      "   - Max: 0.8945\n",
      "   - Mean: 0.4166\n",
      "\n",
      "HDCoefficient:\n",
      "   - Min: 122.0082\n",
      "   - Max: 379.4970\n",
      "   - Mean: 237.8751\n",
      "\n",
      "score:\n",
      "   - Min: -379.4101\n",
      "   - Max: -121.1511\n",
      "   - Mean: -237.4585\n",
      "Confidance: 0.4\n",
      "Segmentation:\n",
      "------------------------------------------\n",
      "Case: MASK_007.nii.gz   Results: {'DiceCoefficient': 0.34625231236786475, 'HDCoefficient': 293.4177908716511, 'score': -293.0715385592832}\n",
      "Case: MASK_009.nii.gz   Results: {'DiceCoefficient': 0.007096413166743425, 'HDCoefficient': 230.59488285736091, 'score': -230.58778644419417}\n",
      "Case: MASK_010.nii.gz   Results: {'DiceCoefficient': 0.43784330479362477, 'HDCoefficient': 226.57890457851542, 'score': -226.1410612737218}\n",
      "Case: MASK_014.nii.gz   Results: {'DiceCoefficient': 0.10323189732887554, 'HDCoefficient': 379.4970355615443, 'score': -379.3938036642154}\n",
      "Case: MASK_022.nii.gz   Results: {'DiceCoefficient': 0.6892366724638553, 'HDCoefficient': 32.01562118716424, 'score': -31.326384514700386}\n",
      "Case: MASK_025.nii.gz   Results: {'DiceCoefficient': 0.5925841437681781, 'HDCoefficient': 143.7810835958611, 'score': -143.18849945209294}\n",
      "Case: MASK_031.nii.gz   Results: {'DiceCoefficient': 0.2640094635388723, 'HDCoefficient': 169.57004452437937, 'score': -169.3060350608405}\n",
      "Case: MASK_039.nii.gz   Results: {'DiceCoefficient': 0.33295485058882196, 'HDCoefficient': 232.13358223230003, 'score': -231.8006273817112}\n",
      "Case: MASK_053.nii.gz   Results: {'DiceCoefficient': 0.1923837069173295, 'HDCoefficient': 200.90047287151916, 'score': -200.70808916460183}\n",
      "Case: MASK_061.nii.gz   Results: {'DiceCoefficient': 0.349368137850615, 'HDCoefficient': 309.3121400785944, 'score': -308.9627719407438}\n",
      "Case: MASK_062.nii.gz   Results: {'DiceCoefficient': 0.43544613144758737, 'HDCoefficient': 180.49376720540795, 'score': -180.05832107396037}\n",
      "Case: MASK_066.nii.gz   Results: {'DiceCoefficient': 0.740627041734165, 'HDCoefficient': 139.61017154921055, 'score': -138.86954450747638}\n",
      "Case: MASK_067.nii.gz   Results: {'DiceCoefficient': 0.4419313675298577, 'HDCoefficient': 178.1488141975691, 'score': -177.70688283003923}\n",
      "Case: MASK_074.nii.gz   Results: {'DiceCoefficient': 0.10843804863346174, 'HDCoefficient': 236.73825208444873, 'score': -236.62981403581526}\n",
      "Case: MASK_075.nii.gz   Results: {'DiceCoefficient': 0.9040503446080744, 'HDCoefficient': 171.75272923595713, 'score': -170.84867889134907}\n",
      "Case: MASK_080.nii.gz   Results: {'DiceCoefficient': 0.5490802905133901, 'HDCoefficient': 306.5175362030695, 'score': -305.96845591255607}\n",
      "Case: MASK_087.nii.gz   Results: {'DiceCoefficient': 0.6159665485584369, 'HDCoefficient': 226.29184695874486, 'score': -225.6758804101864}\n",
      "Case: MASK_089.nii.gz   Results: {'DiceCoefficient': 0.866087896379199, 'HDCoefficient': 122.00819644597652, 'score': -121.14210854959732}\n",
      "Case: MASK_090.nii.gz   Results: {'DiceCoefficient': 0.051428916283994715, 'HDCoefficient': 335.1492204973779, 'score': -335.0977915810939}\n",
      "Case: MASK_099.nii.gz   Results: {'DiceCoefficient': 0.6840961207720238, 'HDCoefficient': 234.448288541418, 'score': -233.764192420646}\n",
      "\n",
      "DiceCoefficient:\n",
      "   - Min: 0.0071\n",
      "   - Max: 0.9041\n",
      "   - Mean: 0.4356\n",
      "\n",
      "HDCoefficient:\n",
      "   - Min: 32.0156\n",
      "   - Max: 379.4970\n",
      "   - Mean: 217.4480\n",
      "\n",
      "score:\n",
      "   - Min: -379.3938\n",
      "   - Max: -31.3264\n",
      "   - Mean: -217.0124\n",
      "Confidance: 0.5\n",
      "Segmentation:\n",
      "------------------------------------------\n",
      "Case: MASK_007.nii.gz   Results: {'DiceCoefficient': 0.366561433568254, 'HDCoefficient': 293.4177908716511, 'score': -293.0512294380828}\n",
      "Case: MASK_009.nii.gz   Results: {'DiceCoefficient': 0.0034205575508807935, 'HDCoefficient': 230.59488285736091, 'score': -230.59146229981005}\n",
      "Case: MASK_010.nii.gz   Results: {'DiceCoefficient': 0.44985422321800456, 'HDCoefficient': 226.57890457851542, 'score': -226.1290503552974}\n",
      "Case: MASK_014.nii.gz   Results: {'DiceCoefficient': 0.11639609976323628, 'HDCoefficient': 379.4970355615443, 'score': -379.3806394617811}\n",
      "Case: MASK_022.nii.gz   Results: {'DiceCoefficient': 0.6731964307273545, 'HDCoefficient': 32.01562118716424, 'score': -31.342424756436888}\n",
      "Case: MASK_025.nii.gz   Results: {'DiceCoefficient': 0.6082384127007057, 'HDCoefficient': 86.61408661412992, 'score': -86.00584820142922}\n",
      "Case: MASK_031.nii.gz   Results: {'DiceCoefficient': 0.27897159874865096, 'HDCoefficient': 105.80170130957252, 'score': -105.52272971082387}\n",
      "Case: MASK_039.nii.gz   Results: {'DiceCoefficient': 0.3889147773097948, 'HDCoefficient': 232.13358223230003, 'score': -231.74466745499024}\n",
      "Case: MASK_053.nii.gz   Results: {'DiceCoefficient': 0.22254440014498006, 'HDCoefficient': 200.90047287151916, 'score': -200.67792847137417}\n",
      "Case: MASK_061.nii.gz   Results: {'DiceCoefficient': 0.3133425390523863, 'HDCoefficient': 309.3121400785944, 'score': -308.998797539542}\n",
      "Case: MASK_062.nii.gz   Results: {'DiceCoefficient': 0.48205959667670817, 'HDCoefficient': 170.2410056361275, 'score': -169.7589460394508}\n",
      "Case: MASK_066.nii.gz   Results: {'DiceCoefficient': 0.7284200434203429, 'HDCoefficient': 106.38608931622592, 'score': -105.65766927280558}\n",
      "Case: MASK_067.nii.gz   Results: {'DiceCoefficient': 0.42698615497111264, 'HDCoefficient': 178.1488141975691, 'score': -177.721828042598}\n",
      "Case: MASK_074.nii.gz   Results: {'DiceCoefficient': 0.11550527072223213, 'HDCoefficient': 236.73825208444873, 'score': -236.6227468137265}\n",
      "Case: MASK_075.nii.gz   Results: {'DiceCoefficient': 0.9140215479976795, 'HDCoefficient': 109.07795377618706, 'score': -108.16393222818938}\n",
      "Case: MASK_080.nii.gz   Results: {'DiceCoefficient': 0.5843310426285837, 'HDCoefficient': 305.85944484354246, 'score': -305.27511380091386}\n",
      "Case: MASK_087.nii.gz   Results: {'DiceCoefficient': 0.5996311994275335, 'HDCoefficient': 63.25345840347388, 'score': -62.653827204046344}\n",
      "Case: MASK_089.nii.gz   Results: {'DiceCoefficient': 0.8728668687134888, 'HDCoefficient': 20.024984394500787, 'score': -19.152117525787297}\n",
      "Case: MASK_090.nii.gz   Results: {'DiceCoefficient': 0.08089053803339519, 'HDCoefficient': 325.9279061387656, 'score': -325.8470156007322}\n",
      "Case: MASK_099.nii.gz   Results: {'DiceCoefficient': 0.6894229799253505, 'HDCoefficient': 234.448288541418, 'score': -233.75886556149266}\n",
      "\n",
      "DiceCoefficient:\n",
      "   - Min: 0.0034\n",
      "   - Max: 0.9140\n",
      "   - Mean: 0.4458\n",
      "\n",
      "HDCoefficient:\n",
      "   - Min: 20.0250\n",
      "   - Max: 379.4970\n",
      "   - Mean: 192.3486\n",
      "\n",
      "score:\n",
      "   - Min: -379.3806\n",
      "   - Max: -19.1521\n",
      "   - Mean: -191.9028\n",
      "Confidance: 0.6\n",
      "Segmentation:\n",
      "------------------------------------------\n",
      "Case: MASK_007.nii.gz   Results: {'DiceCoefficient': 0.4108192408071463, 'HDCoefficient': 293.4177908716511, 'score': -293.00697163084396}\n",
      "Case: MASK_009.nii.gz   Results: {'DiceCoefficient': 0.0, 'HDCoefficient': 230.59488285736091, 'score': -230.59488285736091}\n",
      "Case: MASK_010.nii.gz   Results: {'DiceCoefficient': 0.5383357066362257, 'HDCoefficient': 226.57890457851542, 'score': -226.0405688718792}\n",
      "Case: MASK_014.nii.gz   Results: {'DiceCoefficient': 0.13659776114277866, 'HDCoefficient': 340.74917461382057, 'score': -340.6125768526778}\n",
      "Case: MASK_022.nii.gz   Results: {'DiceCoefficient': 0.5968216750111011, 'HDCoefficient': 57.706152185014034, 'score': -57.109330510002934}\n",
      "Case: MASK_025.nii.gz   Results: {'DiceCoefficient': 0.6108740813545691, 'HDCoefficient': 45.221676218380054, 'score': -44.61080213702549}\n",
      "Case: MASK_031.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_039.nii.gz   Results: {'DiceCoefficient': 0.5201721162144022, 'HDCoefficient': 232.13358223230003, 'score': -231.61341011608562}\n",
      "Case: MASK_053.nii.gz   Results: {'DiceCoefficient': 0.2100306919642857, 'HDCoefficient': 200.90047287151916, 'score': -200.69044217955488}\n",
      "Case: MASK_061.nii.gz   Results: {'DiceCoefficient': 0.2605907619648637, 'HDCoefficient': 309.3121400785944, 'score': -309.0515493166295}\n",
      "Case: MASK_062.nii.gz   Results: {'DiceCoefficient': 0.5090851957617121, 'HDCoefficient': 109.47145746723207, 'score': -108.96237227147036}\n",
      "Case: MASK_066.nii.gz   Results: {'DiceCoefficient': 0.7194820219879716, 'HDCoefficient': 50.00999900019995, 'score': -49.290516978211976}\n",
      "Case: MASK_067.nii.gz   Results: {'DiceCoefficient': 0.3836228621350089, 'HDCoefficient': 150.70500987027606, 'score': -150.32138700814104}\n",
      "Case: MASK_074.nii.gz   Results: {'DiceCoefficient': 0.1378053104190706, 'HDCoefficient': 236.73825208444873, 'score': -236.60044677402965}\n",
      "Case: MASK_075.nii.gz   Results: {'DiceCoefficient': 0.9201153132664924, 'HDCoefficient': 109.07795377618706, 'score': -108.15783846292057}\n",
      "Case: MASK_080.nii.gz   Results: {'DiceCoefficient': 0.5843310426285837, 'HDCoefficient': 305.85944484354246, 'score': -305.27511380091386}\n",
      "Case: MASK_087.nii.gz   Results: {'DiceCoefficient': 0.539117290096585, 'HDCoefficient': 35.4400902933387, 'score': -34.900973003242115}\n",
      "Case: MASK_089.nii.gz   Results: {'DiceCoefficient': 0.8744031935424175, 'HDCoefficient': 20.024984394500787, 'score': -19.15058120095837}\n",
      "Case: MASK_090.nii.gz   Results: {'DiceCoefficient': 0.08434102764909673, 'HDCoefficient': 325.9279061387656, 'score': -325.8435651111165}\n",
      "Case: MASK_099.nii.gz   Results: {'DiceCoefficient': 0.6350133486732451, 'HDCoefficient': 51.82663407939975, 'score': -51.191620730726505}\n",
      "\n",
      "DiceCoefficient:\n",
      "   - Min: 0.0000\n",
      "   - Max: 0.9201\n",
      "   - Mean: 0.4336\n",
      "\n",
      "HDCoefficient:\n",
      "   - Min: 0.0000\n",
      "   - Max: 340.7492\n",
      "   - Mean: 166.5848\n",
      "\n",
      "score:\n",
      "   - Min: -340.6126\n",
      "   - Max: 0.0000\n",
      "   - Mean: -166.1512\n",
      "Confidance: 0.7000000000000001\n",
      "Segmentation:\n",
      "------------------------------------------\n",
      "Case: MASK_007.nii.gz   Results: {'DiceCoefficient': 0.3913304643336158, 'HDCoefficient': 281.9734030010632, 'score': -281.5820725367296}\n",
      "Case: MASK_009.nii.gz   Results: {'DiceCoefficient': 0.0, 'HDCoefficient': 230.59488285736091, 'score': -230.59488285736091}\n",
      "Case: MASK_010.nii.gz   Results: {'DiceCoefficient': 0.5186201264522551, 'HDCoefficient': 206.9541011915444, 'score': -206.43548106509215}\n",
      "Case: MASK_014.nii.gz   Results: {'DiceCoefficient': 0.11436818386047082, 'HDCoefficient': 340.74917461382057, 'score': -340.6348064299601}\n",
      "Case: MASK_022.nii.gz   Results: {'DiceCoefficient': 0.5578079042891886, 'HDCoefficient': 99.82985525382675, 'score': -99.27204734953756}\n",
      "Case: MASK_025.nii.gz   Results: {'DiceCoefficient': 0.5532529406682384, 'HDCoefficient': 45.552167895721496, 'score': -44.99891495505326}\n",
      "Case: MASK_031.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_039.nii.gz   Results: {'DiceCoefficient': 0.6576752009819893, 'HDCoefficient': 157.38170160472913, 'score': -156.72402640374713}\n",
      "Case: MASK_053.nii.gz   Results: {'DiceCoefficient': 0.4258086453861521, 'HDCoefficient': 173.26280616450836, 'score': -172.8369975191222}\n",
      "Case: MASK_061.nii.gz   Results: {'DiceCoefficient': 0.2554219772271477, 'HDCoefficient': 309.3121400785944, 'score': -309.0567181013672}\n",
      "Case: MASK_062.nii.gz   Results: {'DiceCoefficient': 0.5176954211609691, 'HDCoefficient': 109.07795377618706, 'score': -108.5602583550261}\n",
      "Case: MASK_066.nii.gz   Results: {'DiceCoefficient': 0.6995418471401929, 'HDCoefficient': 50.00999900019995, 'score': -49.31045715305975}\n",
      "Case: MASK_067.nii.gz   Results: {'DiceCoefficient': 0.33676909907234326, 'HDCoefficient': 150.70500987027606, 'score': -150.36824077120372}\n",
      "Case: MASK_074.nii.gz   Results: {'DiceCoefficient': 0.1593447529342695, 'HDCoefficient': 236.73825208444873, 'score': -236.57890733151447}\n",
      "Case: MASK_075.nii.gz   Results: {'DiceCoefficient': 0.9263714086522585, 'HDCoefficient': 73.95944834840239, 'score': -73.03307693975013}\n",
      "Case: MASK_080.nii.gz   Results: {'DiceCoefficient': 0.5800836502930468, 'HDCoefficient': 305.85944484354246, 'score': -305.27936119324943}\n",
      "Case: MASK_087.nii.gz   Results: {'DiceCoefficient': 0.4550725093091583, 'HDCoefficient': 35.4400902933387, 'score': -34.98501778402954}\n",
      "Case: MASK_089.nii.gz   Results: {'DiceCoefficient': 0.8744031935424175, 'HDCoefficient': 20.024984394500787, 'score': -19.15058120095837}\n",
      "Case: MASK_090.nii.gz   Results: {'DiceCoefficient': 0.06875654479424281, 'HDCoefficient': 278.12587078515367, 'score': -278.0571142403594}\n",
      "Case: MASK_099.nii.gz   Results: {'DiceCoefficient': 0.6180825370156463, 'HDCoefficient': 51.82663407939975, 'score': -51.20855154238411}\n",
      "\n",
      "DiceCoefficient:\n",
      "   - Min: 0.0000\n",
      "   - Max: 0.9264\n",
      "   - Mean: 0.4355\n",
      "\n",
      "HDCoefficient:\n",
      "   - Min: 0.0000\n",
      "   - Max: 340.7492\n",
      "   - Mean: 157.8689\n",
      "\n",
      "score:\n",
      "   - Min: -340.6348\n",
      "   - Max: 0.0000\n",
      "   - Mean: -157.4334\n",
      "Confidance: 0.8\n",
      "Segmentation:\n",
      "------------------------------------------\n",
      "Case: MASK_007.nii.gz   Results: {'DiceCoefficient': 0.34800000000000003, 'HDCoefficient': 47.042533945356304, 'score': -46.694533945356305}\n",
      "Case: MASK_009.nii.gz   Results: {'DiceCoefficient': 0.0, 'HDCoefficient': 230.59488285736091, 'score': -230.59488285736091}\n",
      "Case: MASK_010.nii.gz   Results: {'DiceCoefficient': 0.33924337425575635, 'HDCoefficient': 31.080540535840107, 'score': -30.741297161584352}\n",
      "Case: MASK_014.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_022.nii.gz   Results: {'DiceCoefficient': 0.2986232003387597, 'HDCoefficient': 101.11874208078342, 'score': -100.82011888044467}\n",
      "Case: MASK_025.nii.gz   Results: {'DiceCoefficient': 0.14156174571074437, 'HDCoefficient': 57.89645930452051, 'score': -57.754897558809766}\n",
      "Case: MASK_031.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_039.nii.gz   Results: {'DiceCoefficient': 0.7555427584221134, 'HDCoefficient': 16.73320053068151, 'score': -15.977657772259398}\n",
      "Case: MASK_053.nii.gz   Results: {'DiceCoefficient': 0.16678930586215354, 'HDCoefficient': 13.638181696985855, 'score': -13.471392391123702}\n",
      "Case: MASK_061.nii.gz   Results: {'DiceCoefficient': 0.25346659529378984, 'HDCoefficient': 294.84911395491764, 'score': -294.59564735962385}\n",
      "Case: MASK_062.nii.gz   Results: {'DiceCoefficient': 0.5415204518039258, 'HDCoefficient': 105.78279633286313, 'score': -105.24127588105921}\n",
      "Case: MASK_066.nii.gz   Results: {'DiceCoefficient': 0.4545487818268327, 'HDCoefficient': 57.436921923097515, 'score': -56.982373141270685}\n",
      "Case: MASK_067.nii.gz   Results: {'DiceCoefficient': 0.10940023689027674, 'HDCoefficient': 75.21303078589507, 'score': -75.10363054900479}\n",
      "Case: MASK_074.nii.gz   Results: {'DiceCoefficient': 0.23082684990062619, 'HDCoefficient': 226.76198976018887, 'score': -226.53116291028823}\n",
      "Case: MASK_075.nii.gz   Results: {'DiceCoefficient': 0.9246590271065458, 'HDCoefficient': 15.0, 'score': -14.075340972893454}\n",
      "Case: MASK_080.nii.gz   Results: {'DiceCoefficient': 0.4498556825973943, 'HDCoefficient': 305.85944484354246, 'score': -305.4095891609451}\n",
      "Case: MASK_087.nii.gz   Results: {'DiceCoefficient': 0.20431923465199875, 'HDCoefficient': 35.4400902933387, 'score': -35.2357710586867}\n",
      "Case: MASK_089.nii.gz   Results: {'DiceCoefficient': 0.8744031935424175, 'HDCoefficient': 20.024984394500787, 'score': -19.15058120095837}\n",
      "Case: MASK_090.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_099.nii.gz   Results: {'DiceCoefficient': 0.3827797964344149, 'HDCoefficient': 28.635642126552707, 'score': -28.25286233011829}\n",
      "\n",
      "DiceCoefficient:\n",
      "   - Min: 0.0000\n",
      "   - Max: 0.9247\n",
      "   - Mean: 0.3238\n",
      "\n",
      "HDCoefficient:\n",
      "   - Min: 0.0000\n",
      "   - Max: 305.8594\n",
      "   - Mean: 83.1554\n",
      "\n",
      "score:\n",
      "   - Min: -305.4096\n",
      "   - Max: 0.0000\n",
      "   - Mean: -82.8317\n",
      "Confidance: 0.9\n",
      "Segmentation:\n",
      "------------------------------------------\n",
      "Case: MASK_007.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_009.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_010.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_014.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_022.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_025.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_031.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_039.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_053.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_061.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_062.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_066.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_067.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_074.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_075.nii.gz   Results: {'DiceCoefficient': 0.24094221895973153, 'HDCoefficient': 29.5296461204668, 'score': -29.28870390150707}\n",
      "Case: MASK_080.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_087.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_089.nii.gz   Results: {'DiceCoefficient': 0.21784582988104173, 'HDCoefficient': 22.58317958127243, 'score': -22.365333751391386}\n",
      "Case: MASK_090.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "Case: MASK_099.nii.gz   Results: {'DiceCoefficient': 0, 'HDCoefficient': 0, 'score': 0}\n",
      "\n",
      "DiceCoefficient:\n",
      "   - Min: 0.0000\n",
      "   - Max: 0.2409\n",
      "   - Mean: 0.0229\n",
      "\n",
      "HDCoefficient:\n",
      "   - Min: 0.0000\n",
      "   - Max: 29.5296\n",
      "   - Mean: 2.6056\n",
      "\n",
      "score:\n",
      "   - Min: -29.2887\n",
      "   - Max: 0.0000\n",
      "   - Mean: -2.5827\n"
     ]
    }
   ],
   "source": [
    "if \"gt_files\" not in locals() or \"pred_files\" not in locals():\n",
    "    if not os.path.exists(os.path.join(output_folder,\"cases.json\")):\n",
    "        print(f\"Data not found in {output_folder}\")\n",
    "    print(f\"Loading data from {output_folder}..\")\n",
    "    with open(os.path.join(output_folder,\"cases.json\")) as fp:\n",
    "        json_data = json.load(fp)\n",
    "        gt_files, pred_files = json_data[\"gt_files\"], json_data[\"pred_files\"]\n",
    "        confidance_th = json_data[\"confidance_th\"]\n",
    "\n",
    "print(\"Validating..\")\n",
    "scores = []\n",
    "for conf in confidance_th:\n",
    "    print(f\"Confidance: {conf}\")\n",
    "    scores.append(Validate(pred_files[conf], gt_files))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>DiceCoefficient</th>\n",
       "      <th>HDCoefficient</th>\n",
       "      <th>score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0.1</th>\n",
       "      <td>0.613172</td>\n",
       "      <td>254.214476</td>\n",
       "      <td>-253.601304</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.2</th>\n",
       "      <td>0.613172</td>\n",
       "      <td>254.214476</td>\n",
       "      <td>-253.601304</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.3</th>\n",
       "      <td>0.657658</td>\n",
       "      <td>234.448289</td>\n",
       "      <td>-233.790630</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.4</th>\n",
       "      <td>0.684096</td>\n",
       "      <td>234.448289</td>\n",
       "      <td>-233.764192</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.5</th>\n",
       "      <td>0.689423</td>\n",
       "      <td>234.448289</td>\n",
       "      <td>-233.758866</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.6</th>\n",
       "      <td>0.635013</td>\n",
       "      <td>51.826634</td>\n",
       "      <td>-51.191621</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.7</th>\n",
       "      <td>0.618083</td>\n",
       "      <td>51.826634</td>\n",
       "      <td>-51.208552</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.8</th>\n",
       "      <td>0.382780</td>\n",
       "      <td>28.635642</td>\n",
       "      <td>-28.252862</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.9</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     DiceCoefficient  HDCoefficient       score\n",
       "0.1         0.613172     254.214476 -253.601304\n",
       "0.2         0.613172     254.214476 -253.601304\n",
       "0.3         0.657658     234.448289 -233.790630\n",
       "0.4         0.684096     234.448289 -233.764192\n",
       "0.5         0.689423     234.448289 -233.758866\n",
       "0.6         0.635013      51.826634  -51.191621\n",
       "0.7         0.618083      51.826634  -51.208552\n",
       "0.8         0.382780      28.635642  -28.252862\n",
       "0.9         0.000000       0.000000    0.000000"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "df_scores = pd.DataFrame(scores, index=confidance_th)\n",
    "display(df_scores)\n",
    "\n",
    "for df_col in df_scores.columns:\n",
    "    plt.plot(df_scores.index, df_scores[df_col])\n",
    "    plt.xlabel('Confidance')\n",
    "    plt.ylabel(df_col)\n",
    "    plt.title(f'{df_col} vs Confidance')\n",
    "    plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "py310_1",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.6"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}