{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"## Definitions\n\nThe goal of this competition is to identify fractures in CT scans of the cervical spine (neck) at both the level of a single vertebrae and the entire patient. Figure 1 presents a set of slice related of vertebrae with fractures of different grades: G0 (`normal`), G1(`mild`), G2(`moderate`) and G3(`severe`). \n\n![Screenshot from 2022-09-04 11-15-10.png](attachment:0d39aa27-c36d-44f1-a1ef-f8a3eb2d4476.png)\n\n[1] https://arxiv.org/pdf/2208.10698.pdf\n\n\nIn this sense, the dataset is composed by:\n\n- train.csv Metadata for the train test set.\n\nStudyInstanceUID - The study ID. There is one unique study ID for each patient scan.\npatient_overall - One of the target columns. The patient level outcome, i.e. if any of the vertebrae are fractured.\nC[1-7] - The other target columns. Whether the given vertebrae is fractured. See this diagram for the real location of each vertbrae in the spine.\n\n- test.csv Metadata for the test set prediction structure. Only the first few rows of the test set are available for download.\n\nrow_id - The row ID. This will match the same column in the sample submission file.\nStudyInstanceUID - The study ID.\nprediction_type - Which one of the eight target columns needs a prediction in this row.\n[train/test]_images/[StudyInstanceUID]/[slice_number].dcm The image data, organized with one folder per scan. Expect to see roughly 1,500 scans in the hidden test set. Each image is in the dicom file format. The DICOM image files are ≤ 1 mm slice thickness, axial orientation, and bone kernel. Note that some of the DICOM files are JPEG compressed. You may require additional resources to read the pixel array of these files, such as GDCM and pylibjpeg.\n\n- sample_submission.csv A valid sample submission.\n\nrow_id - The row ID. See the test.csv for what prediction needs to be filed in that row.\nfractured - The target column.\ntrain_bounding_boxes.csv Bounding boxes for a subset of the training set.\n\n\n__________________________________________","metadata":{},"attachments":{"0d39aa27-c36d-44f1-a1ef-f8a3eb2d4476.png":{"image/png":"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"}}},{"cell_type":"markdown","source":"## EDA / Pre processing\n\nIn order to faster up the process of eda and image processing, I'll use the library TorchIO [2]\n\n[2] https://torchio.readthedocs.io/\n","metadata":{}},{"cell_type":"code","source":"%pip install --quiet torchio","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:47:28.107407Z","iopub.execute_input":"2022-10-06T00:47:28.108136Z","iopub.status.idle":"2022-10-06T00:47:40.230347Z","shell.execute_reply.started":"2022-10-06T00:47:28.108043Z","shell.execute_reply":"2022-10-06T00:47:40.229091Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import copy\nimport torchio as tio\nimport torch\nfrom tqdm import tqdm\nimport matplotlib as mpl\nimport matplotlib.pyplot as plt\nfrom torchio.transforms.preprocessing.spatial.to_canonical import ToCanonical\nimport numpy as np\n\ntorch.manual_seed(20220802)\nmpl.rcParams['figure.figsize'] = 12, 8","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:53:33.404678Z","iopub.execute_input":"2022-10-06T00:53:33.405051Z","iopub.status.idle":"2022-10-06T00:53:34.719144Z","shell.execute_reply.started":"2022-10-06T00:53:33.405013Z","shell.execute_reply":"2022-10-06T00:53:34.717795Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"root_dir = '../input/rsna-2022-cervical-spine-fracture-detection'\ndataset = tio.datasets.RSNACervicalSpineFracture(root_dir, add_segmentations=True, add_bounding_boxes=True)\nprint('Number of subjects:', len(dataset))","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:53:34.726272Z","iopub.execute_input":"2022-10-06T00:53:34.729894Z","iopub.status.idle":"2022-10-06T00:53:36.949696Z","shell.execute_reply.started":"2022-10-06T00:53:34.729838Z","shell.execute_reply":"2022-10-06T00:53:36.948714Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### Dataset general evaluation (EDA)","metadata":{}},{"cell_type":"code","source":"first_subject = dataset[10]\nfirst_subject","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:53:38.408801Z","iopub.execute_input":"2022-10-06T00:53:38.409149Z","iopub.status.idle":"2022-10-06T00:53:43.687118Z","shell.execute_reply.started":"2022-10-06T00:53:38.409119Z","shell.execute_reply":"2022-10-06T00:53:43.686039Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"for key, value in first_subject.items():\n    print(f'{key}: {value}')","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:53:43.691074Z","iopub.execute_input":"2022-10-06T00:53:43.69144Z","iopub.status.idle":"2022-10-06T00:53:43.708478Z","shell.execute_reply.started":"2022-10-06T00:53:43.691407Z","shell.execute_reply":"2022-10-06T00:53:43.707427Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"According to those informations, we have:\n\n- patient's C1 and C2 vertebrae are fractured, according to the labels. \n- CT scan is an instance of ScalarImage with  512×512×243 voxels\n- in-plane pixel spacing of 0.44 mm\n- slice thickness of 0.80 mm\n- LPS+ orientation data type \"short tensor\" (or signed 16-bit integers)\n- takes 121.5 MiB of memory.","metadata":{}},{"cell_type":"markdown","source":"Let's visualize some data","metadata":{}},{"cell_type":"code","source":"first_subject.plot()","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:53:50.677749Z","iopub.execute_input":"2022-10-06T00:53:50.678495Z","iopub.status.idle":"2022-10-06T00:53:52.063233Z","shell.execute_reply.started":"2022-10-06T00:53:50.678441Z","shell.execute_reply":"2022-10-06T00:53:52.061216Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### Now, let's normalize this!","metadata":{}},{"cell_type":"code","source":"# Hounsfield scale to keep only intensity values within a sensible range using the Clamp transform\nHOUNSFIELD_AIR, HOUNSFIELD_BONE = -1000, 1900\nclamp = tio.Clamp(out_min=HOUNSFIELD_AIR, out_max=HOUNSFIELD_BONE)\nfirst_subject_clamped = clamp(first_subject)\nfirst_subject_clamped.ct.hist()\nfirst_subject_clamped.plot()\n\nrescale = tio.RescaleIntensity(percentiles=(0.5, 99.5))\npreprocess_intensity = tio.Compose([\n    clamp,\n    rescale,\n])","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:53:52.708907Z","iopub.execute_input":"2022-10-06T00:53:52.709318Z","iopub.status.idle":"2022-10-06T00:53:57.641237Z","shell.execute_reply.started":"2022-10-06T00:53:52.709286Z","shell.execute_reply":"2022-10-06T00:53:57.640292Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"first_subject_preprocessed = preprocess_intensity(first_subject)\nfirst_subject_preprocessed.ct.hist(show=False), plt.ylim(0, 1e6)\nfirst_subject_preprocessed.plot()","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:53:57.642955Z","iopub.execute_input":"2022-10-06T00:53:57.644709Z","iopub.status.idle":"2022-10-06T00:54:04.057418Z","shell.execute_reply.started":"2022-10-06T00:53:57.644662Z","shell.execute_reply":"2022-10-06T00:54:04.056358Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"After processing a single image, would be interesting to save those set of images as .png considering the fact that this would allow us to use different frameworks. Let's explore how to make this..","metadata":{}},{"cell_type":"code","source":"images_names = first_subject_preprocessed.get_images_names()\nimages_list = first_subject_preprocessed.get_images()\nimages_array = images_list[0].numpy()\niterable_dict = enumerate(first_subject_preprocessed.get_images_dict(intensity_only=False).items())","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:54:04.059193Z","iopub.execute_input":"2022-10-06T00:54:04.059598Z","iopub.status.idle":"2022-10-06T00:54:04.065239Z","shell.execute_reply.started":"2022-10-06T00:54:04.059561Z","shell.execute_reply":"2022-10-06T00:54:04.064469Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def color_labels(arrays, cmap_dict):\n    results = []\n    for array in arrays:\n        si, sj = array.shape\n        rgb = np.zeros((si, sj, 3), dtype=np.uint8)\n        for label, color in cmap_dict.items():\n            if isinstance(color, str):\n                mpl, _ = import_mpl_plt()\n                color = mpl.colors.to_rgb(color)\n                color = [255 * n for n in color]\n            rgb[array == label] = color\n        results.append(rgb)\n    return results\n\ndef rotate(image, radiological=True):\n    # Rotate for visualization purposes\n    image = np.rot90(image, -1)\n    if radiological:\n#         image = np.fliplr(image)\n        image=image\n    return image","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:54:04.067299Z","iopub.execute_input":"2022-10-06T00:54:04.068055Z","iopub.status.idle":"2022-10-06T00:54:04.078076Z","shell.execute_reply.started":"2022-10-06T00:54:04.06802Z","shell.execute_reply":"2022-10-06T00:54:04.076974Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"kwargs = {}\nchannel = -1\npercentiles=(0.5, 99.5)\n\nfor image_index, (name, image) in iterable_dict:\n    image = ToCanonical()(image)\n\n    sr, sa, ss = image.spacing\n    sag_aspect = ss / sa\n    cor_aspect = ss / sr\n    axi_aspect = sa / sr\n    data = image.data[channel]\n\n    p1, p2 = np.percentile(data, percentiles)\n    kwargs['vmin'] = p1\n    kwargs['vmax'] = p2\n    kwargs['origin'] = 'lower'\n    kwargs['cmap'] = 'gray'\n\n    \n    indices = np.array(data.shape) // 2\n    i, j, k = indices\n    \n    slice_x = rotate(data[i, :, :], radiological=True)\n    slice_y = rotate(data[:, j, :], radiological=True)\n    slice_z = rotate(data[:, :, k], radiological=True)\n    \n#     slices = slice_x, slice_y, slice_z\n#     slice_x, slice_y, slice_z = color_labels(slices, 'gray')\n    \n    slice_x_arr = slice_x\n    slice_y_arr = slice_y\n    slice_z_arr = slice_z\n\n    plt.imshow(slice_x_arr, aspect=sag_aspect, **kwargs) # Sagittal\n    plt.imshow(slice_y_arr, aspect=cor_aspect, **kwargs) # Coronal\n    plt.imshow(slice_z_arr, aspect=axi_aspect, **kwargs) # Axial\n","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:54:04.080211Z","iopub.execute_input":"2022-10-06T00:54:04.080864Z","iopub.status.idle":"2022-10-06T00:54:05.121118Z","shell.execute_reply.started":"2022-10-06T00:54:04.080825Z","shell.execute_reply":"2022-10-06T00:54:05.120309Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Ok. Checked that we can save the processed image. Let's check the occurrences of bounding box","metadata":{}},{"cell_type":"markdown","source":"Spatial preprocessing: we have to segmentation with raw images.. But how to use those segmentations?","metadata":{}},{"cell_type":"markdown","source":"Do we have bbox in which plane?\n\n","metadata":{}},{"cell_type":"code","source":"import os\nimport glob\nimport pydicom\nimport nibabel as nib\nimport pandas as pd\nimport numpy as np\nfrom pydicom.pixel_data_handlers.util import apply_voi_lut\nimport matplotlib.pyplot as plt\nimport matplotlib.patches as patches\nimport seaborn as sns","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:54:05.846078Z","iopub.execute_input":"2022-10-06T00:54:05.846631Z","iopub.status.idle":"2022-10-06T00:54:06.092672Z","shell.execute_reply.started":"2022-10-06T00:54:05.846596Z","shell.execute_reply":"2022-10-06T00:54:06.091711Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"DATA_DIR = root_dir\nTRAIN_DIR = os.path.join(DATA_DIR, \"train_images\")\nSEGM_DIR = os.path.join(DATA_DIR, \"segmentations\")\ntrain_df = pd.read_csv(root_dir + \"/train.csv\")","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:54:06.094582Z","iopub.execute_input":"2022-10-06T00:54:06.094934Z","iopub.status.idle":"2022-10-06T00:54:06.106841Z","shell.execute_reply.started":"2022-10-06T00:54:06.094899Z","shell.execute_reply":"2022-10-06T00:54:06.105804Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"bboxes_df = pd.read_csv(os.path.join(DATA_DIR, \"train_bounding_boxes.csv\"))\nprint(bboxes_df.shape)\nbboxes_df.head()","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:54:06.285599Z","iopub.execute_input":"2022-10-06T00:54:06.285868Z","iopub.status.idle":"2022-10-06T00:54:06.310758Z","shell.execute_reply.started":"2022-10-06T00:54:06.285843Z","shell.execute_reply":"2022-10-06T00:54:06.309815Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def rescale_img_to_hu(dcm_ds):\n    \"\"\"Rescales the image to Hounsfield unit.\n    \"\"\"\n    return dcm_ds.pixel_array * dcm_ds.RescaleSlope + dcm_ds.RescaleIntercept","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:54:08.12695Z","iopub.execute_input":"2022-10-06T00:54:08.127304Z","iopub.status.idle":"2022-10-06T00:54:08.132613Z","shell.execute_reply.started":"2022-10-06T00:54:08.127273Z","shell.execute_reply":"2022-10-06T00:54:08.131648Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def show_bboxes_for_patient(patient_id, nrows, ncols):\n    patient_dir = os.path.join(TRAIN_DIR, patient_id)\n    num_slices = len(glob.glob(f\"{patient_dir}/*\"))\n    print(f\"Number of slices for patient: {num_slices}\")\n    \n    bboxes_patient_df = bboxes_df[bboxes_df[\"StudyInstanceUID\"] == patient_id]\n    num_bboxes = bboxes_patient_df.shape[0]\n    print(f\"Number of bounding boxes for patient: {num_bboxes}\")\n    \n    print(f\"Fracture information for patient: \")\n    print(f\"{train_df[train_df['StudyInstanceUID'] == patient_id].iloc[:, 1:-1]}\")\n    \n    fig, axs = plt.subplots(nrows, ncols, figsize=(24,15))\n    axs = axs.flatten()\n    \n    for i in range(min(num_bboxes, nrows*ncols)):\n        slice_number = bboxes_patient_df.iloc[i][\"slice_number\"]\n        ds = pydicom.dcmread(os.path.join(patient_dir, f\"{slice_number}.dcm\"))\n        print(str(slice_number))\n\n        if i==0:\n            # TODO: do we have bbox in oder planes?\n            pixel_spacing = ds.PixelSpacing\n            slice_thickness = ds.SliceThickness\n            axial_aspect = pixel_spacing[1] / pixel_spacing[0]\n            sagittal_aspect = pixel_spacing[1] / slice_thickness\n            coronal_aspect = slice_thickness / pixel_spacing[0]\n            img_shape = list(ds.pixel_array.shape)\n            img_shape.append(num_slices)\n            img3d = np.zeros(img_shape)\n        img2d = rescale_img_to_hu(ds)\n        img3d[:, :, i] = img2d\n        \n        axs[i].imshow(rescale_img_to_hu(ds), cmap=\"bone\")\n#         axs[i].imshow(img3d[:, :, img_shape[2]//2], cmap=\"bone\") # axial\n#         axs[i].imshow(img3d[:, img_shape[1]//2, :], cmap=\"bone\") # sagittal\n#         axs[i].set_aspect(sagittal_aspect)\n#         axs[i].imshow(img3d[img_shape[0]//2, :, :].T, cmap=\"bone\") # coronal\n#         axs[i].set_aspect(coronal_aspect)\n        axs[i].axis(\"off\")\n        \n        x = bboxes_patient_df.iloc[i][\"x\"]\n        y = bboxes_patient_df.iloc[i][\"y\"]\n        width = bboxes_patient_df.iloc[i][\"width\"]\n        height = bboxes_patient_df.iloc[i][\"height\"]\n        rect = patches.Rectangle((x, y), width, height, fill=False, edgecolor=\"blue\", linewidth=1.5)\n        axs[i].add_patch(rect)\n        \n","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:54:08.346309Z","iopub.execute_input":"2022-10-06T00:54:08.347044Z","iopub.status.idle":"2022-10-06T00:54:08.360703Z","shell.execute_reply.started":"2022-10-06T00:54:08.347005Z","shell.execute_reply":"2022-10-06T00:54:08.359433Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# show_bboxes_for_patient(\"1.2.826.0.1.3680043.10051\", 3, 5)\n# show_bboxes_for_patient(\"1.2.826.0.1.3680043.6200\", 3, 5)\nshow_bboxes_for_patient(\"1.2.826.0.1.3680043.4744\", 5, 5)","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:54:10.258099Z","iopub.execute_input":"2022-10-06T00:54:10.258481Z","iopub.status.idle":"2022-10-06T00:54:12.638731Z","shell.execute_reply.started":"2022-10-06T00:54:10.258449Z","shell.execute_reply":"2022-10-06T00:54:12.637455Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Based on previously function, bounding box is associated with axial vision","metadata":{}},{"cell_type":"code","source":"def calc_num_slices_per_patient(train_df):\n    num_slices = []\n    for patient_id in train_df[\"StudyInstanceUID\"]:\n        slice_paths = glob.glob(f\"{TRAIN_DIR}/{patient_id}/*\")\n        num_slices.append(len(slice_paths))\n    train_df[\"num_slices\"] = num_slices\n    return train_df","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:54:17.1658Z","iopub.execute_input":"2022-10-06T00:54:17.166161Z","iopub.status.idle":"2022-10-06T00:54:17.172463Z","shell.execute_reply.started":"2022-10-06T00:54:17.166128Z","shell.execute_reply":"2022-10-06T00:54:17.171455Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_df = calc_num_slices_per_patient(train_df)\ntrain_df.head()","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:54:22.30006Z","iopub.execute_input":"2022-10-06T00:54:22.300456Z","iopub.status.idle":"2022-10-06T00:56:09.400953Z","shell.execute_reply.started":"2022-10-06T00:54:22.300417Z","shell.execute_reply":"2022-10-06T00:56:09.399967Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def show_segm_for_patient(patient_id, nrows, ncols, step=1):\n    \"\"\"Shows slices with segmentation masks for the patient, skips slices that have empty segmentation masks.\"\"\"\n    num_slices = train_df.loc[train_df[\"StudyInstanceUID\"] == patient_id, \"num_slices\"].values[0]\n    print(f\"Number of slices for patient: {num_slices}\")\n    \n    print(f\"Fracture information for patient: \")\n    print(f\"{train_df[train_df['StudyInstanceUID'] == patient_id].iloc[:, 1:-1]}\")\n    patient_dir = os.path.join(TRAIN_DIR, patient_id)\n    \n    segm_path = os.path.join(SEGM_DIR, f\"{patient_id}.nii\")\n    # source: https://www.kaggle.com/code/kretes/segmentations-fracture-zoom-in\n    segm_mask = nib.load(segm_path).get_fdata()\n        \n    # flip in z axis\n    segm_mask = np.flip(segm_mask, axis=-1)\n    # rotate 90 degrees in xy\n    segm_mask = np.rot90(segm_mask, axes=(0, 1))\n        \n    fig, axs = plt.subplots(nrows, ncols, figsize=(24,15))\n    axs = axs.flatten()\n    \n    count = 0\n    idx = 0\n    while count < nrows*ncols:\n        if segm_mask[:, :, idx].max() > 0:\n            # dicom filenames start from 1\n            ds = pydicom.dcmread(os.path.join(patient_dir, f\"{idx+1}.dcm\"))\n            axs[count].imshow(rescale_img_to_hu(ds), cmap=\"bone\")\n            axs[count].axis(\"off\")\n            segm_im = axs[count].imshow(segm_mask[:, :, idx], alpha=0.4)\n            segm_max = segm_mask[:, :, idx].max()\n\n            segm_labels = np.unique(segm_mask[:, :, idx][segm_mask[:, :, idx].nonzero()])\n            segm_colors = [segm_im.cmap(label/segm_max) for label in segm_labels]\n            ptch = [patches.Patch(color=segm_colors[i], label=f\"C{int(segm_labels[i])}\") for i in range(len(segm_labels))]\n            axs[count].legend(handles=ptch)\n            count += 1\n            print('Cs: ', segm_labels)\n        idx += step","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:56:09.402856Z","iopub.execute_input":"2022-10-06T00:56:09.403927Z","iopub.status.idle":"2022-10-06T00:56:09.415799Z","shell.execute_reply.started":"2022-10-06T00:56:09.403888Z","shell.execute_reply":"2022-10-06T00:56:09.414877Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# show_segm_for_patient(\"1.2.826.0.1.3680043.4744\", 5, 5)\nshow_segm_for_patient(\"1.2.826.0.1.3680043.10921\", 5, 5)","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:56:09.417321Z","iopub.execute_input":"2022-10-06T00:56:09.417697Z","iopub.status.idle":"2022-10-06T00:56:13.685447Z","shell.execute_reply.started":"2022-10-06T00:56:09.417664Z","shell.execute_reply":"2022-10-06T00:56:13.684428Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Besides that, we have some pixel segmentations ","metadata":{}},{"cell_type":"markdown","source":"_________________________________________________","metadata":{}},{"cell_type":"markdown","source":"### Data preparatation and augmentation","metadata":{}},{"cell_type":"markdown","source":"0. Use the set of codes presented above and save segmentations images and masks as .png files ","metadata":{}},{"cell_type":"code","source":"%mkdir /kaggle/working/images_test\n%mkdir /kaggle/working/masks_test\n\n%mkdir /kaggle/working/images\n%mkdir /kaggle/working/masks","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:56:30.748959Z","iopub.execute_input":"2022-10-06T00:56:30.749315Z","iopub.status.idle":"2022-10-06T00:56:34.631431Z","shell.execute_reply.started":"2022-10-06T00:56:30.749284Z","shell.execute_reply":"2022-10-06T00:56:34.630082Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import cv2","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:56:34.633915Z","iopub.execute_input":"2022-10-06T00:56:34.634871Z","iopub.status.idle":"2022-10-06T00:56:34.829453Z","shell.execute_reply.started":"2022-10-06T00:56:34.634823Z","shell.execute_reply":"2022-10-06T00:56:34.828542Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def show_segm_for_patient(patient_id, step=1):\n    \"\"\"Shows slices with segmentation masks for the patient, skips slices that have empty segmentation masks.\"\"\"\n    num_slices = train_df.loc[train_df[\"StudyInstanceUID\"] == patient_id, \"num_slices\"].values[0]\n    print(f\"Number of slices for patient: {num_slices}\")\n    \n    print(f\"Fracture information for patient: \")\n    print(f\"{train_df[train_df['StudyInstanceUID'] == patient_id].iloc[:, 1:-1]}\")\n    patient_dir = os.path.join(TRAIN_DIR, patient_id)\n    \n    segm_path = os.path.join(SEGM_DIR, f\"{patient_id}.nii\")\n    # source: https://www.kaggle.com/code/kretes/segmentations-fracture-zoom-in\n    segm_mask = nib.load(segm_path).get_fdata()\n\n    # flip in z axis\n    segm_mask = np.flip(segm_mask, axis=-1)\n\n    # rotate 90 degrees in xy\n    segm_mask = np.rot90(segm_mask, axes=(0, 1))\n\n    count = 0\n    idx = 0\n    while count < num_slices:\n        if segm_mask[:, :, idx].max() > 0:\n            # dicom filenames start from 1\n            ds = pydicom.dcmread(os.path.join(patient_dir, f\"{idx+1}.dcm\"))\n\n            # Save raw image as .png\n            img = rescale_img_to_hu(ds)\n\n            segm_im_plt = segm_mask[:, :, idx]\n            segm_labels = np.unique(segm_mask[:, :, idx][segm_mask[:, :, idx].nonzero()])\n\n            segm_im_plt_filter = segm_im_plt.copy()\n            for lbl in segm_labels:\n                for i in range(segm_im_plt.shape[0]):\n                    for j in range(segm_im_plt.shape[1]):\n                        if segm_im_plt[i][j] != lbl:\n                            segm_im_plt_filter[i][j] = 0\n                        else:\n                            segm_im_plt_filter[i][j] = 255\n\n                cv2.imwrite('images/' + str(patient_id) + '_slc_' + str(count) + '_' + str(lbl) + '.png', img)\n                cv2.imwrite('masks/' + str(patient_id) + '_slc_' + str(count) + '_' + str(lbl) + '.png', segm_im_plt_filter)\n\n            count += 1\n        idx += step\n","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:56:38.46395Z","iopub.execute_input":"2022-10-06T00:56:38.464327Z","iopub.status.idle":"2022-10-06T00:56:38.477143Z","shell.execute_reply.started":"2022-10-06T00:56:38.464294Z","shell.execute_reply":"2022-10-06T00:56:38.476191Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"for inst_id in train_df['StudyInstanceUID']:\n    print(inst_id)\n    try:\n        show_segm_for_patient(inst_id)\n        # print(str(inst_id))\n    except Exception as e:\n        pass","metadata":{"execution":{"iopub.status.busy":"2022-10-04T11:06:19.949314Z","iopub.execute_input":"2022-10-04T11:06:19.949757Z","iopub.status.idle":"2022-10-04T12:38:56.673286Z","shell.execute_reply.started":"2022-10-04T11:06:19.949718Z","shell.execute_reply":"2022-10-04T12:38:56.671847Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### UNET for semantic segmentation","metadata":{}},{"cell_type":"code","source":"!pip install -q imutils","metadata":{"execution":{"iopub.status.busy":"2022-10-04T12:38:56.675621Z","iopub.execute_input":"2022-10-04T12:38:56.676081Z","iopub.status.idle":"2022-10-04T12:39:13.66097Z","shell.execute_reply.started":"2022-10-04T12:38:56.676036Z","shell.execute_reply":"2022-10-04T12:39:13.659144Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# import the necessary packages\nimport os\nimport time\n\nimport cv2\nimport zipfile\n\nimport torch\nfrom torch.nn import ReLU\nfrom torch.nn import Conv2d\nfrom torch.nn import Module\nfrom torch.optim import Adam\nfrom torch.nn import MaxPool2d\nfrom torch.nn import ModuleList\nfrom torchvision import transforms\nfrom torch.utils.data import Dataset\nfrom torch.nn import ConvTranspose2d\nfrom torch.nn import functional as F\nfrom torch.nn import BCEWithLogitsLoss\nfrom torch.utils.data import DataLoader\nfrom torchvision.transforms import CenterCrop\n\nfrom sklearn.model_selection import train_test_split\n\n\nfrom imutils import paths\nfrom tqdm import tqdm\nimport matplotlib.pyplot as plt","metadata":{"execution":{"iopub.status.busy":"2022-10-04T12:39:13.663356Z","iopub.execute_input":"2022-10-04T12:39:13.663872Z","iopub.status.idle":"2022-10-04T12:39:14.327383Z","shell.execute_reply.started":"2022-10-04T12:39:13.663821Z","shell.execute_reply":"2022-10-04T12:39:14.3257Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# define the path to the images and masks dataset\nIMAGE_DATASET_PATH = os.path.join(\"/kaggle/working/images/\")\nMASK_DATASET_PATH = os.path.join(\"/kaggle/working/masks/\")\n\n# define the test split\nTEST_SPLIT = 0.15\n\n# determine the device to be used for training and evaluation\nDEVICE = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n\n# determine if we will be pinning memory during data loading\nPIN_MEMORY = True if DEVICE == \"cuda\" else False","metadata":{"execution":{"iopub.status.busy":"2022-10-04T12:39:14.348317Z","iopub.execute_input":"2022-10-04T12:39:14.349029Z","iopub.status.idle":"2022-10-04T12:39:14.492842Z","shell.execute_reply.started":"2022-10-04T12:39:14.348932Z","shell.execute_reply":"2022-10-04T12:39:14.489831Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"class SegmentationDataset(Dataset):\n\tdef __init__(self, imagePaths, maskPaths, transforms):\n\t\t# store the image and mask filepaths, and augmentation transforms\n\t\tself.imagePaths = imagePaths\n\t\tself.maskPaths = maskPaths\n\t\tself.transforms = transforms\n        \n\tdef __len__(self):\n\t\t# return the number of total samples contained in the dataset\n\t\treturn len(self.imagePaths)\n    \n\tdef __getitem__(self, idx):\n\t\t# grab the image path from the current index\n\t\timagePath = self.imagePaths[idx]\n        \n\t\t# load the image from disk, swap its channels from BGR to RGB,\n\t\t# and read the associated mask from disk in grayscale mode\n\t\timage = cv2.imread(imagePath)\n\t\timage = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)\n\t\tmask = cv2.imread(self.maskPaths[idx], 0)\n        \n\t\t# check to see if we are applying any transformations\n\t\tif self.transforms is not None:\n\t\t\t# apply the transformations to both image and its mask\n\t\t\timage = self.transforms(image)\n\t\t\tmask = self.transforms(mask)\n            \n\t\t# return a tuple of the image and its mask\n\t\treturn (image, mask)","metadata":{"execution":{"iopub.status.busy":"2022-10-04T12:39:14.497825Z","iopub.execute_input":"2022-10-04T12:39:14.499488Z","iopub.status.idle":"2022-10-04T12:39:14.514496Z","shell.execute_reply.started":"2022-10-04T12:39:14.499432Z","shell.execute_reply":"2022-10-04T12:39:14.512382Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"class Block(Module):\n\tdef __init__(self, inChannels, outChannels):\n\t\tsuper().__init__()\n\t\t# store the convolution and RELU layers\n\t\tself.conv1 = Conv2d(inChannels, outChannels, 3)\n\t\tself.relu = ReLU()\n\t\tself.conv2 = Conv2d(outChannels, outChannels, 3)\n        \n\tdef forward(self, x):\n\t\t# apply CONV => RELU => CONV block to the inputs and return it\n\t\treturn self.conv2(self.relu(self.conv1(x)))","metadata":{"execution":{"iopub.status.busy":"2022-10-04T12:39:14.520378Z","iopub.execute_input":"2022-10-04T12:39:14.523461Z","iopub.status.idle":"2022-10-04T12:39:14.547038Z","shell.execute_reply.started":"2022-10-04T12:39:14.523097Z","shell.execute_reply":"2022-10-04T12:39:14.544987Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"class Encoder(Module):\n\tdef __init__(self, channels=(3, 16, 32, 64)):\n\t\tsuper().__init__()\n\t\t# store the encoder blocks and maxpooling layer\n\t\tself.encBlocks = ModuleList([Block(channels[i], channels[i + 1]) for i in range(len(channels) - 1)])\n\t\tself.pool = MaxPool2d(2)\n        \n\tdef forward(self, x):\n\t\t# initialize an empty list to store the intermediate outputs\n\t\tblockOutputs = []\n        \n\t\t# loop through the encoder blocks\n\t\tfor block in self.encBlocks:\n\t\t\t# pass the inputs through the current encoder block, store the outputs, and then apply maxpooling on the output\n\t\t\tx = block(x)\n            # add blocked x to the end\n\t\t\tblockOutputs.append(x)\n            # pooling\n\t\t\tx = self.pool(x)\n            \n        # return the list containing the intermediate outputs\n\t\treturn blockOutputs","metadata":{"execution":{"iopub.status.busy":"2022-10-04T12:39:14.549803Z","iopub.execute_input":"2022-10-04T12:39:14.550805Z","iopub.status.idle":"2022-10-04T12:39:14.572331Z","shell.execute_reply.started":"2022-10-04T12:39:14.55073Z","shell.execute_reply":"2022-10-04T12:39:14.567697Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"class Decoder(Module):\n\tdef __init__(self, channels=(64, 32, 16)):\n\t\tsuper().__init__()\n        \n\t\t# initialize the number of channels, upsampler blocks, and decoder blocks\n\t\tself.channels = channels\n        # upconvolution 64 to 32, 32 to 16 respectively\n\t\tself.upconvs = ModuleList([ConvTranspose2d(channels[i], channels[i + 1], 2, 2) for i in range(len(channels) - 1)])\n        # define 3*3 conv and RELU block\n\t\tself.dec_blocks = ModuleList([Block(channels[i], channels[i + 1]) for i in range(len(channels) - 1)])\n        \n\tdef forward(self, x, encFeatures):\n\t\t# loop through the number of channels\n\t\tfor i in range(len(self.channels) - 1):\n\t\t\t# pass the inputs through the upsampler blocks\n\t\t\tx = self.upconvs[i](x)\n            \n\t\t\t# crop the current features from the encoder blocks, concatenate them with the current upsampled features,\n\t\t\t# and pass the concatenated output through the current decoder block\n\t\t\tencFeat = self.crop(encFeatures[i], x)\n            # Concatenate!\n\t\t\tx = torch.cat([x, encFeat], dim=1)\n            # 3*3 conv and RELU for each upscending layer.\n\t\t\tx = self.dec_blocks[i](x)\n            \n\t\t# return the final decoder output\n\t\treturn x\n    \n\tdef crop(self, encFeatures, x):\n\t\t# grab the dimensions of the inputs, and crop the encoder features to match the dimensions\n\t\t(_, _, H, W) = x.shape\n\t\tencFeatures = CenterCrop([H, W])(encFeatures)\n        \n\t\t# return the cropped features\n\t\treturn encFeatures","metadata":{"execution":{"iopub.status.busy":"2022-10-04T12:39:14.578974Z","iopub.execute_input":"2022-10-04T12:39:14.581161Z","iopub.status.idle":"2022-10-04T12:39:14.608339Z","shell.execute_reply.started":"2022-10-04T12:39:14.5811Z","shell.execute_reply":"2022-10-04T12:39:14.606483Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# define the number of channels in the input, number of classes,\n# and number of levels in the U-Net model\nNUM_CHANNELS = 1\nNUM_CLASSES = 1\nNUM_LEVELS = 100\n\n# initialize learning rate, number of epochs to train for, and the batch size\nINIT_LR = 0.001\nNUM_EPOCHS = 50\nBATCH_SIZE = 32\n\n# define the input image dimensions\n# Crop\nINPUT_IMAGE_WIDTH = 512\nINPUT_IMAGE_HEIGHT = 512\n\n# define threshold to filter weak predictions\nTHRESHOLD = 0.2\n\n# define the path to the base output directory\nBASE_OUTPUT = \"output\"\n\n# define the path to the output serialized model, model training plot, and testing image paths\nMODEL_PATH = \"unet_tgs_salt.pth\"\nPLOT_PATH = \"plot.png\"\nTEST_PATHS = \"test_paths.txt\"","metadata":{"execution":{"iopub.status.busy":"2022-10-04T12:39:14.610563Z","iopub.execute_input":"2022-10-04T12:39:14.611954Z","iopub.status.idle":"2022-10-04T12:39:14.627275Z","shell.execute_reply.started":"2022-10-04T12:39:14.611896Z","shell.execute_reply":"2022-10-04T12:39:14.625804Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"class UNet(Module):\n\tdef __init__(self, encChannels=(3, 16, 32, 64), decChannels=(64, 32, 16), nbClasses=1, retainDim=True, outSize=(INPUT_IMAGE_HEIGHT,  INPUT_IMAGE_WIDTH)):\n\t\tsuper().__init__()\n\t\t\n\t\t# initialize the encoder and decoder\n\t\tself.encoder = Encoder(encChannels)\n\t\tself.decoder = Decoder(decChannels)\n\t\t\n\t\t# initialize the regression head and store the class variables\n\t\tself.head = Conv2d(decChannels[-1], nbClasses, 1)\n\t\tself.retainDim = retainDim\n\t\tself.outSize = outSize\n\t\t\n\tdef forward(self, x):\n\t\t# grab the features from the encoder\n\t\tencFeatures = self.encoder(x)\n\t\t\n\t\t# pass the encoder features through decoder making sure that their dimensions are suited for concatenation\n        # last output as 1st decfeature and every\n\t\tdecFeatures = self.decoder(encFeatures[::-1][0],encFeatures[::-1][1:])\n\t\t\n\t\t# pass the decoder features through the regression head to obtain the segmentation mask\n\t\tmap_ = self.head(decFeatures)\n\t\t\n\t\t# check to see if we are retaining the original output dimensions and if so, then resize the output to match them\n\t\tif self.retainDim:\n\t\t\tmap_ = F.interpolate(map_, self.outSize)\n\t\t\t\n\t\t# return the segmentation map\n\t\treturn map_","metadata":{"execution":{"iopub.status.busy":"2022-10-04T12:39:14.628555Z","iopub.execute_input":"2022-10-04T12:39:14.629062Z","iopub.status.idle":"2022-10-04T12:39:14.6558Z","shell.execute_reply.started":"2022-10-04T12:39:14.629022Z","shell.execute_reply":"2022-10-04T12:39:14.654316Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# load the image and mask filepaths in a sorted manner\nimagePaths = sorted(list(paths.list_images(IMAGE_DATASET_PATH)))\nmaskPaths = sorted(list(paths.list_images(MASK_DATASET_PATH)))\n\n# partition the data into training and testing splits using 85% of the data for training and the remaining 15% for testing\nsplit = train_test_split(imagePaths, maskPaths, test_size=TEST_SPLIT, random_state=42)\n\n# unpack the data split\n(trainImages, testImages) = split[:2]\n(trainMasks, testMasks) = split[2:]\n\n# write the testing image paths to disk so that we can use then when evaluating/testing our model\nprint(\"[INFO] saving testing image paths...\")\nf = open(\"test_paths.txt\", \"w\")\nf.write(\"\\n\".join(testImages))\nf.close()","metadata":{"execution":{"iopub.status.busy":"2022-10-04T12:39:14.657819Z","iopub.execute_input":"2022-10-04T12:39:14.658945Z","iopub.status.idle":"2022-10-04T12:39:15.211938Z","shell.execute_reply.started":"2022-10-04T12:39:14.658797Z","shell.execute_reply":"2022-10-04T12:39:15.209923Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# define transformations\ntransforms = transforms.Compose([transforms.ToPILImage(), transforms.Resize((INPUT_IMAGE_HEIGHT, INPUT_IMAGE_WIDTH)), transforms.ToTensor()])\n\n# create the train and test datasets\ntrainDS = SegmentationDataset(imagePaths=trainImages, maskPaths=trainMasks,\ttransforms=transforms)\ntestDS = SegmentationDataset(imagePaths=testImages, maskPaths=testMasks, transforms=transforms)\nprint(f\"[INFO] found {len(trainDS)} examples in the training set...\")\nprint(f\"[INFO] found {len(testDS)} examples in the test set...\")\n\n# create the training and test data loaders\ntrainLoader = DataLoader(trainDS, shuffle=True,\tbatch_size=BATCH_SIZE, pin_memory=PIN_MEMORY, num_workers=os.cpu_count())\ntestLoader = DataLoader(testDS, shuffle=False, batch_size=BATCH_SIZE, pin_memory=PIN_MEMORY, num_workers=os.cpu_count())","metadata":{"execution":{"iopub.status.busy":"2022-10-04T12:39:15.217305Z","iopub.execute_input":"2022-10-04T12:39:15.220189Z","iopub.status.idle":"2022-10-04T12:39:15.237296Z","shell.execute_reply.started":"2022-10-04T12:39:15.220126Z","shell.execute_reply":"2022-10-04T12:39:15.23606Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# initialize our UNet model\nunet = UNet().to(DEVICE)\n\n# initialize loss function and optimizer\nlossFunc = BCEWithLogitsLoss()\nopt = Adam(unet.parameters(), lr=INIT_LR)\n\n# calculate steps per epoch for training and test set\ntrainSteps = len(trainDS) // BATCH_SIZE\ntestSteps = len(testDS) // BATCH_SIZE\n\n# initialize a dictionary to store training history\nH = {\"train_loss\": [], \"test_loss\": []}","metadata":{"execution":{"iopub.status.busy":"2022-10-04T12:39:15.24341Z","iopub.execute_input":"2022-10-04T12:39:15.246206Z","iopub.status.idle":"2022-10-04T12:39:19.55809Z","shell.execute_reply.started":"2022-10-04T12:39:15.246145Z","shell.execute_reply":"2022-10-04T12:39:19.556558Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# loop over epochs\nprint(\"[INFO] training the network...\")\nstartTime = time.time()\nfor e in tqdm(range(NUM_EPOCHS)):\n\t# set the model in training mode\n\tunet.train()\n    \n\t# initialize the total training and validation loss\n\ttotalTrainLoss = 0\n\ttotalTestLoss = 0\n    \n\t# loop over the training set\n\tfor (i, (x, y)) in enumerate(trainLoader):\n\t\t# send the input to the device\n\t\t(x, y) = (x.to(DEVICE), y.to(DEVICE))\n        \n\t\t# perform a forward pass and calculate the training loss\n\t\tpred = unet(x)\n\t\tloss = lossFunc(pred, y)\n        \n\t\t# first, zero out any previously accumulated gradients, then perform backpropagation, and then update model parameters\n\t\topt.zero_grad()\n\t\tloss.backward()\n\t\topt.step()\n        \n\t\t# add the loss to the total training loss so far\n\t\ttotalTrainLoss += loss\n        \n\t# switch off autograd\n\twith torch.no_grad():\n\t\t# set the model in evaluation mode\n\t\tunet.eval()\n        \n\t\t# loop over the validation set\n\t\tfor (x, y) in testLoader:\n\t\t\t# send the input to the device\n\t\t\t(x, y) = (x.to(DEVICE), y.to(DEVICE))\n            \n\t\t\t# make the predictions and calculate the validation loss\n\t\t\tpred = unet(x)\n\t\t\ttotalTestLoss += lossFunc(pred, y)\n            \n\t# calculate the average training and validation loss\n\tavgTrainLoss = totalTrainLoss / trainSteps\n\tavgTestLoss = totalTestLoss / testSteps\n    \n\t# update our training history\n\tH[\"train_loss\"].append(avgTrainLoss.cpu().detach().numpy())\n\tH[\"test_loss\"].append(avgTestLoss.cpu().detach().numpy())\n    \n\t# print the model training and validation information\n\tprint(\"[INFO] EPOCH: {}/{}\".format(e + 1, NUM_EPOCHS))\n\tprint(\"Train loss: {:.6f}, Test loss: {:.4f}\".format(avgTrainLoss, avgTestLoss))\n    \n# display the total time needed to perform the training\nendTime = time.time()\nprint(\"[INFO] total time taken to train the model: {:.2f}s\".format(endTime - startTime))","metadata":{"execution":{"iopub.status.busy":"2022-10-04T12:39:19.560157Z","iopub.execute_input":"2022-10-04T12:39:19.560631Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# plot the training loss\nplt.style.use(\"ggplot\")\nplt.figure()\nplt.plot(H[\"train_loss\"], label=\"train_loss\")\nplt.plot(H[\"test_loss\"], label=\"test_loss\")\nplt.title(\"Training Loss on Dataset\")\nplt.xlabel(\"Epoch #\")\nplt.ylabel(\"Loss\")\nplt.legend(loc=\"lower left\")\nplt.savefig(PLOT_PATH)\n\n# serialize the model to disk\ntorch.save(unet, MODEL_PATH)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def prepare_plot(origImage, origMask, predMask):\n\t# initialize our figure\n\tfigure, ax = plt.subplots(nrows=1, ncols=3, figsize=(10, 10))\n\n\t# plot the original image, its mask, and the predicted mask\n\tax[0].imshow(origImage)\n\tax[1].imshow(origMask)\n\tax[2].imshow(predMask)\n\n\t# set the titles of the subplots\n\tax[0].set_title(\"Image\")\n\tax[1].set_title(\"Original Mask\")\n\tax[2].set_title(\"Predicted Mask\")\n\n\t# set the layout of the figure and display it\n\tfigure.tight_layout()\n\tfigure.show()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def make_predictions(model, imagePath):\n\t# set model to evaluation mode\n\tmodel.eval()\n\n\t# turn off gradient tracking\n\twith torch.no_grad():\n\t\t# load the image from disk, swap its color channels, cast it to float data type, and scale its pixel values\n\t\timage = cv2.imread(imagePath)\n\t\timage = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)\n\t\timage = image.astype(\"float32\") / 255.0\n\n\t\t# resize the image and make a copy of it for visualization\n\t\timage = cv2.resize(image, (128, 128))\n\t\torig = image.copy()\n\n\t\t# find the filename and generate the path to ground truth mask\n\t\tfilename = imagePath.split(os.path.sep)[-1]\n\t\tgroundTruthPath = os.path.join(MASK_DATASET_PATH, filename)\n\n\t\t# load the ground-truth segmentation mask in grayscale mode and resize it\n\t\tgtMask = cv2.imread(groundTruthPath, 0)\n\t\tgtMask = cv2.resize(gtMask, (INPUT_IMAGE_HEIGHT, INPUT_IMAGE_HEIGHT))\n        \n        # make the channel axis to be the leading one, add a batch dimension, create a PyTorch tensor, and flash it to the\n\t\t# current device\n\t\timage = np.transpose(image, (2, 0, 1))\n\t\timage = np.expand_dims(image, 0)\n\t\timage = torch.from_numpy(image).to(DEVICE)\n\n\t\t# make the prediction, pass the results through the sigmoid\n\t\t# function, and convert the result to a NumPy array\n\t\tpredMask = model(image).squeeze()\n\t\tpredMask = torch.sigmoid(predMask)\n\t\tpredMask = predMask.cpu().numpy()\n\n\t\t# filter out the weak predictions and convert them to integers\n\t\tpredMask = (predMask > THRESHOLD) * 255\n\t\tpredMask = predMask.astype(np.uint8)\n\n\t\t# prepare a plot for visualization\n\t\tprepare_plot(orig, gtMask, predMask)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# load the image paths in our testing file and randomly select 10 image paths\nprint(\"[INFO] loading up test image paths...\")\nimagePaths = open(TEST_PATHS).read().strip().split(\"\\n\")\nimagePaths = np.random.choice(imagePaths, size=10)\n\n# load our model from disk and flash it to the current device\nprint(\"[INFO] load up model...\")\nunet = torch.load(MODEL_PATH).to(DEVICE)\n\n# iterate over the randomly selected test image paths\nfor path in imagePaths:\n\t# make predictions and visualize the results\n\tmake_predictions(unet, path)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!zip -r images_test.zip /kaggle/working/images_test/\n!zip -r masks_test.zip /kaggle/working/masks_test/","metadata":{"execution":{"iopub.status.busy":"2022-09-18T13:25:20.990915Z","iopub.status.idle":"2022-09-18T13:25:20.991691Z","shell.execute_reply.started":"2022-09-18T13:25:20.99142Z","shell.execute_reply":"2022-09-18T13:25:20.991462Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"_________________________________________________________________________________________________","metadata":{}},{"cell_type":"markdown","source":"### let's try some detector","metadata":{}},{"cell_type":"code","source":"import os\nimport numpy as np\nimport pandas as pd\nimport cv2\nimport glob\n\nimport matplotlib.pyplot as plt\nimport seaborn as sns\n\nimport os\nimport nibabel as nib\nfrom tqdm import tqdm\nimport squarify as sq\n\n# installing pydicom dependencies\n!pip install pylibjpeg\n!pip install python_gdcm\n!pip install pylibjpeg_libjpeg\n!pip install glob2\n\nfrom glob import glob\nimport pydicom\nfrom pydicom.pixel_data_handlers.util import apply_voi_lut","metadata":{"execution":{"iopub.status.busy":"2022-10-06T00:59:14.283074Z","iopub.execute_input":"2022-10-06T00:59:14.283441Z","iopub.status.idle":"2022-10-06T00:59:59.334091Z","shell.execute_reply.started":"2022-10-06T00:59:14.283404Z","shell.execute_reply":"2022-10-06T00:59:59.332938Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!pip install pylibjpeg pylibjpeg-libjpeg","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:00:03.87166Z","iopub.execute_input":"2022-10-06T01:00:03.872252Z","iopub.status.idle":"2022-10-06T01:00:13.409201Z","shell.execute_reply.started":"2022-10-06T01:00:03.872203Z","shell.execute_reply":"2022-10-06T01:00:13.40792Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"root_dir = '../input/rsna-2022-cervical-spine-fracture-detection'\ndicom_path_var = os.path.join(root_dir, \"train_images/\")\ndir_nii = os.path.join(root_dir, \"segmentations/\")\nall_nii = glob(dir_nii + '*.nii')\n\ntrain_df = pd.read_csv(root_dir + \"/train.csv\")\nbounding_box_df = pd.read_csv(root_dir + \"/train_bounding_boxes.csv\")","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:00:13.411911Z","iopub.execute_input":"2022-10-06T01:00:13.412308Z","iopub.status.idle":"2022-10-06T01:00:13.439773Z","shell.execute_reply.started":"2022-10-06T01:00:13.412255Z","shell.execute_reply":"2022-10-06T01:00:13.438928Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"bounding_box_df","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:00:13.442471Z","iopub.execute_input":"2022-10-06T01:00:13.442859Z","iopub.status.idle":"2022-10-06T01:00:13.561935Z","shell.execute_reply.started":"2022-10-06T01:00:13.442818Z","shell.execute_reply":"2022-10-06T01:00:13.560612Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Rearrange data in order to add bbox infos","metadata":{}},{"cell_type":"code","source":"bounding_box_df['StudyInstanceUID'] = bounding_box_df['StudyInstanceUID'].astype('str')\nbounding_box_df['slice_number'] = bounding_box_df['slice_number'].astype('str')\nbounding_box_df['x_min'] = bounding_box_df['x'].astype('int64')\nbounding_box_df['y_min'] = bounding_box_df['y'].astype('int64')\nbounding_box_df['x_max'] = bounding_box_df['x_min'] + bounding_box_df['width'].astype('int64')\nbounding_box_df['y_max'] = bounding_box_df['y_min'] + bounding_box_df['height'].astype('int64')\nbounding_box_df.drop(['x','y','width','height'], axis=1, inplace=True)\nbounding_box_df","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:00:20.692918Z","iopub.execute_input":"2022-10-06T01:00:20.693284Z","iopub.status.idle":"2022-10-06T01:00:20.724634Z","shell.execute_reply.started":"2022-10-06T01:00:20.693253Z","shell.execute_reply":"2022-10-06T01:00:20.723702Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"nii_ids = [x.split('/')[-1][:-4] for x in all_nii]\nisin_nii_ids = bounding_box_df['StudyInstanceUID'].apply(lambda x : x if x in nii_ids else None)\nisin_nii_ids","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:00:21.167828Z","iopub.execute_input":"2022-10-06T01:00:21.168525Z","iopub.status.idle":"2022-10-06T01:00:21.189047Z","shell.execute_reply.started":"2022-10-06T01:00:21.168486Z","shell.execute_reply":"2022-10-06T01:00:21.187796Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"intersecting_dfs = bounding_box_df.iloc[isin_nii_ids.dropna().index]\nintersecting_dfs.reset_index(drop=True, inplace=True)\nintersecting_dfs","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:00:22.628421Z","iopub.execute_input":"2022-10-06T01:00:22.628774Z","iopub.status.idle":"2022-10-06T01:00:22.645716Z","shell.execute_reply.started":"2022-10-06T01:00:22.628744Z","shell.execute_reply":"2022-10-06T01:00:22.644773Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Example of draw some img+bbox: form train dataset from here...","metadata":{}},{"cell_type":"code","source":"intersecting_dfs.describe()","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:00:32.818101Z","iopub.execute_input":"2022-10-06T01:00:32.818484Z","iopub.status.idle":"2022-10-06T01:00:32.846319Z","shell.execute_reply.started":"2022-10-06T01:00:32.818451Z","shell.execute_reply":"2022-10-06T01:00:32.845339Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#refered to https://www.kaggle.com/code/songseungwon/eda-let-s-draw-the-fracture-more-clearly\n\nfig = plt.figure(figsize=(12,12))\n\nfor i, row in intersecting_dfs.iterrows():\n    s_uid = row['StudyInstanceUID']\n    s_num = row['slice_number']\n    x_min = row['x_min']\n    x_max = row['x_max']\n    y_min = row['y_min']\n    y_max = row['y_max']\n    \n    nii = nib.load(dir_nii + s_uid + '.nii')\n    # (ref) discussion : Explaining Data and Submission in detail\n    seg = nii.get_fdata()[:, ::-1, ::-1].transpose(2,1,0)\n    nii = nii.get_fdata()[:, ::-1, ::-1].transpose(2,1,0)\n    nii = nii[int(s_num)]\n    \n    dicom = pydicom.read_file(dicom_path_var + s_uid + f'/{s_num}.dcm')\n    dicom = dicom.pixel_array\n    dicom = (dicom - np.min(dicom)) / np.max(dicom)\n    dicom = (dicom * 255).astype(np.uint8)\n    dicom = cv2.cvtColor(dicom, cv2.COLOR_GRAY2RGB)\n    print(type(dicom))\n    print(dicom.shape)\n    cv2.rectangle(dicom, pt1=(x_min, y_min), pt2=(x_max, y_max), color=(255,0,0), thickness=3)\n    if i <= 8:\n        ax = fig.add_subplot(int(f'33{i+1}'))\n        ax.imshow(nii, alpha=1, cmap='gray')\n        ax.imshow(dicom, alpha=0.45)\n        ax.set_title(f'sample({i+1})\\n')\n        ax.axis('off')\n    else:\n        break;\n        \nplt.tight_layout()\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:01:00.817553Z","iopub.execute_input":"2022-10-06T01:01:00.818104Z","iopub.status.idle":"2022-10-06T01:01:04.959872Z","shell.execute_reply.started":"2022-10-06T01:01:00.818063Z","shell.execute_reply":"2022-10-06T01:01:04.958907Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"So, here we go prepare this dataset (tfrecord, e.g) and train some huge detector (eff?)","metadata":{}},{"cell_type":"markdown","source":"#### Convert annotations to xml (Pascal voc) and Create associated images (jpg) with the same names ","metadata":{}},{"cell_type":"code","source":"train_df","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:01:38.119007Z","iopub.execute_input":"2022-10-06T01:01:38.119365Z","iopub.status.idle":"2022-10-06T01:01:38.137014Z","shell.execute_reply.started":"2022-10-06T01:01:38.119335Z","shell.execute_reply":"2022-10-06T01:01:38.136073Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"### But, how we can associate the vertebrae with slice?\n#### Let's find out","metadata":{}},{"cell_type":"code","source":"seg.shape","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:01:40.278638Z","iopub.execute_input":"2022-10-06T01:01:40.279322Z","iopub.status.idle":"2022-10-06T01:01:40.286523Z","shell.execute_reply.started":"2022-10-06T01:01:40.279284Z","shell.execute_reply":"2022-10-06T01:01:40.285466Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"np.unique(nii[150])","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:01:41.473881Z","iopub.execute_input":"2022-10-06T01:01:41.474237Z","iopub.status.idle":"2022-10-06T01:01:41.483466Z","shell.execute_reply.started":"2022-10-06T01:01:41.474207Z","shell.execute_reply":"2022-10-06T01:01:41.48237Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Store segmentation paths in a dataframe\nroot_dir = \"../input/rsna-2022-cervical-spine-fracture-detection\"\nseg_paths = glob(f\"{root_dir}/segmentations/*\")\nseg_df = pd.DataFrame({'path': seg_paths})\nseg_df['StudyInstanceUID'] = seg_df['path'].apply(lambda x:x.split('/')[-1][:-4])\nseg_df = seg_df[['StudyInstanceUID','path']]\nprint('seg_df shape:', seg_df.shape)\nseg_df.head(3)","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:01:43.826896Z","iopub.execute_input":"2022-10-06T01:01:43.827256Z","iopub.status.idle":"2022-10-06T01:01:43.846766Z","shell.execute_reply.started":"2022-10-06T01:01:43.827227Z","shell.execute_reply":"2022-10-06T01:01:43.845886Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# https://www.kaggle.com/code/samuelcortinhas/extracting-vertebrae-c1-c7/notebook\n# Metadata was extracted previously (check out my RSNA dataset)\n# https://www.kaggle.com/datasets/samuelcortinhas/rsna-2022-spine-fracture-detection-metadata\nmeta_train = pd.read_csv(\"../input/rsna-2022-spine-fracture-detection-metadata/meta_train_clean.csv\")\n\n# Only select patients with segmentations\nmeta_seg = meta_train[meta_train['StudyInstanceUID'].isin(seg_df['StudyInstanceUID'])].reset_index(drop=True)","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:01:44.205486Z","iopub.execute_input":"2022-10-06T01:01:44.206201Z","iopub.status.idle":"2022-10-06T01:01:45.286118Z","shell.execute_reply.started":"2022-10-06T01:01:44.206154Z","shell.execute_reply":"2022-10-06T01:01:45.285065Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Initialise targets\ntargets = ['C1','C2','C3','C4','C5','C6','C7']\nmeta_seg[targets]=0","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:01:46.619078Z","iopub.execute_input":"2022-10-06T01:01:46.619624Z","iopub.status.idle":"2022-10-06T01:01:46.629018Z","shell.execute_reply.started":"2022-10-06T01:01:46.619589Z","shell.execute_reply":"2022-10-06T01:01:46.627819Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"k=0\n# Loop over 87 patients with segmentations\nfor path, UID in zip(seg_df['path'], seg_df['StudyInstanceUID']):\n    # Get segmentations for patient\n    seg_nib = nib.load(path)\n    seg = seg_nib.get_fdata()\n    seg = seg[:, ::-1, ::-1].transpose(2, 1, 0) # Align orientation with train images\n    num_slices, _, _ = seg.shape\n    \n    # Loop over slices\n    for i in range(num_slices):\n        mask = seg[i]\n        unique_vals = np.unique(mask)\n        \n        # Loop over unique values (except 0)\n        for j in unique_vals[1:]:\n            \n            # Ignore thoratic spine etc\n            if j <= 7:   \n                meta_seg.loc[(meta_seg['StudyInstanceUID']==UID)&(meta_seg['Slice']==i),f'C{int(j)}'] = 1\n                \n    # Iteration tracker\n    if (k%10)==0:\n        print(f'Iteration:{k}')\n    k+=1\n\n# Save extracted targets\nmeta_seg.to_csv(\"meta_segmentation.csv\", index=False)\n","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:01:49.729208Z","iopub.execute_input":"2022-10-06T01:01:49.729583Z","iopub.status.idle":"2022-10-06T01:07:39.861068Z","shell.execute_reply.started":"2022-10-06T01:01:49.729549Z","shell.execute_reply":"2022-10-06T01:07:39.85995Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"meta_seg.describe()","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:07:39.863267Z","iopub.execute_input":"2022-10-06T01:07:39.863757Z","iopub.status.idle":"2022-10-06T01:07:39.922638Z","shell.execute_reply.started":"2022-10-06T01:07:39.863714Z","shell.execute_reply":"2022-10-06T01:07:39.921378Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Example\nmeta_seg[['StudyInstanceUID','Slice']+targets].iloc[199:204,:]","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:07:39.926281Z","iopub.execute_input":"2022-10-06T01:07:39.926595Z","iopub.status.idle":"2022-10-06T01:07:39.940183Z","shell.execute_reply.started":"2022-10-06T01:07:39.926567Z","shell.execute_reply":"2022-10-06T01:07:39.939099Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"%mkdir /kaggle/working/labels_\n%mkdir /kaggle/working/images_","metadata":{"execution":{"iopub.status.busy":"2022-10-06T02:03:57.24911Z","iopub.execute_input":"2022-10-06T02:03:57.2495Z","iopub.status.idle":"2022-10-06T02:03:59.220397Z","shell.execute_reply.started":"2022-10-06T02:03:57.249466Z","shell.execute_reply":"2022-10-06T02:03:59.219059Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"img_counter = 0\n\nfor i, row in intersecting_dfs.iterrows():\n    s_uid = row['StudyInstanceUID']\n    s_num = row['slice_number']\n    x_min = row['x_min']\n    x_max = row['x_max']\n    y_min = row['y_min']\n    y_max = row['y_max']\n    \n    if s_uid in train_df['StudyInstanceUID'].values:\n        try:\n            s_uid_train = train_df[train_df['StudyInstanceUID']==s_uid]['StudyInstanceUID'].values[0]\n            patient_overall_class = train_df[train_df['StudyInstanceUID']==s_uid]['patient_overall'].values[0]\n            vertebrae_index = meta_seg[(meta_seg['StudyInstanceUID']==s_uid) & (meta_seg['Slice']==int(s_num))]\n            vertebrae_index_name = ''\n            vertebrae_index_list = []\n            if vertebrae_index['C1'].values[0]==1:\n                vertebrae_index_name = vertebrae_index_name + '_C1_'\n                vertebrae_index_list.append(1)\n            if vertebrae_index['C2'].values[0]==1:\n                vertebrae_index_name = vertebrae_index_name + '_C2_'\n                vertebrae_index_list.append(2)\n            if vertebrae_index['C3'].values[0]==1:\n                vertebrae_index_name = vertebrae_index_name + '_C3_'\n                vertebrae_index_list.append(3)\n            if vertebrae_index['C4'].values[0]==1:\n                vertebrae_index_name = vertebrae_index_name + '_C4_'\n                vertebrae_index_list.append(4)\n            if vertebrae_index['C5'].values[0]==1:\n                vertebrae_index_name = vertebrae_index_name + '_C5_'\n                vertebrae_index_list.append(5)\n            if vertebrae_index['C6'].values[0]==1:\n                vertebrae_index_name = vertebrae_index_name + '_C6_'\n                vertebrae_index_list.append(6)\n            if vertebrae_index['C7'].values[0]==1:\n                vertebrae_index_name = vertebrae_index_name + '_C7_'\n                vertebrae_index_list.append(7)\n                \n            # print(vertebrae_index_name)\n            \n            for ver_idx in vertebrae_index_list:\n                list_label = str(ver_idx) + ' ' + str(x_min) + ' ' + str(y_min) + ' ' + str(x_max) + ' ' + str(y_max)\n    \n                # Saves as a yolo annotation type\n                with open('labels_/' + str(s_uid_train) + '_' + str(s_num) + '_' + str(ver_idx) + '.txt', 'w') as out:\n                    out.write(str(list_label))\n\n                # Here we have to save associated image\n                dicom = pydicom.read_file(dicom_path_var + s_uid + f'/{s_num}.dcm')\n                dicom = dicom.pixel_array\n                dicom = (dicom - np.min(dicom)) / np.max(dicom)\n                dicom = (dicom * 255).astype(np.uint8)\n                dicom = cv2.cvtColor(dicom, cv2.COLOR_GRAY2RGB)\n        \n                ### TODO: SAVE WHICH IS THE SPINAL VERTEBRAE IS THIS\n                ### get from df `meta_seg`\n                cv2.imwrite('images_/' + str(s_uid_train)  + '_' + str(s_num) + '_' + str(ver_idx) + '.jpg', dicom)\n\n                # print('Saved ' + str(s_uid_train)  + '_' + str(s_num))\n                img_counter += 1\n                # print(img_counter)\n                # print(patient_overall_class)\n                # if patient_overall_class==0:\n                    # print('Saved ' + str(s_uid_train)  + '_' + str(s_num))\n            \n            # break\n\n                \n                \n            # list_label = str(patient_overall_class) + ' ' + str(x_min) + ' ' + str(y_min) + ' ' + str(x_max) + ' ' + str(y_max)\n    \n            # Saves as a yolo annotation type\n            # with open('labels/' + str(s_uid_train) + '_' + str(s_num) + '_' + vertebrae_index_name + '.txt', 'w') as out:\n                # out.write(str(list_label))\n        \n            # Here we have to save associated image\n            # dicom = pydicom.read_file(dicom_path_var + s_uid + f'/{s_num}.dcm')\n            # dicom = dicom.pixel_array\n            # dicom = (dicom - np.min(dicom)) / np.max(dicom)\n            # dicom = (dicom * 255).astype(np.uint8)\n            # dicom = cv2.cvtColor(dicom, cv2.COLOR_GRAY2RGB)\n        \n            ### TODO: SAVE WHICH IS THE SPINAL VERTEBRAE IS THIS\n            ### get from df `meta_seg`\n            # cv2.imwrite('images/' + str(s_uid_train)  + '_' + str(s_num) + '_' + vertebrae_index_name + '.jpg', dicom)\n        \n            # print('Saved ' + str(s_uid_train)  + '_' + str(s_num))\n            # img_counter += 1\n            # print(img_counter)\n            # print(patient_overall_class)\n            # if patient_overall_class==0:\n                # print('Saved ' + str(s_uid_train)  + '_' + str(s_num))\n            \n            # break\n\n        \n        except Exception as e:\n#             print(e)\n#             print(patient_overall_class)\n            pass\n\n        # cv2.rectangle(dicom, pt1=(x_min, y_min), pt2=(x_max, y_max), color=(255,0,0), thickness=3)\n        # print(type(dicom))\n        # print(dicom.shape)\n        # plt.imshow(dicom)\n        # plt.show()\n        # break","metadata":{"execution":{"iopub.status.busy":"2022-10-06T02:04:35.418081Z","iopub.execute_input":"2022-10-06T02:04:35.418471Z","iopub.status.idle":"2022-10-06T02:04:58.897063Z","shell.execute_reply.started":"2022-10-06T02:04:35.418431Z","shell.execute_reply":"2022-10-06T02:04:58.896059Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Move to appropriate folders","metadata":{}},{"cell_type":"code","source":"!ls /kaggle/working/images_ | wc -l\n# !ls /kaggle/working/images","metadata":{"execution":{"iopub.status.busy":"2022-10-06T02:04:58.898823Z","iopub.execute_input":"2022-10-06T02:04:58.899172Z","iopub.status.idle":"2022-10-06T02:04:59.895085Z","shell.execute_reply.started":"2022-10-06T02:04:58.899138Z","shell.execute_reply":"2022-10-06T02:04:59.893822Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!ls /kaggle/working/labels_ | wc -l","metadata":{"execution":{"iopub.status.busy":"2022-10-06T02:04:59.896964Z","iopub.execute_input":"2022-10-06T02:04:59.897641Z","iopub.status.idle":"2022-10-06T02:05:00.891128Z","shell.execute_reply.started":"2022-10-06T02:04:59.897592Z","shell.execute_reply":"2022-10-06T02:05:00.889973Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"\"\"\"\n@author: Caroline Pacheco do E. Silva\n\"\"\"\nimport os\nimport cv2\nfrom xml.dom.minidom import parseString\nfrom lxml.etree import Element, SubElement, tostring\nimport numpy as np\nfrom os.path import join\n\n## coco classes\nCLASSES = ('1', '0')\n\n## converts the normalized positions  into integer positions\ndef unconvert(class_id, width, height, x, y, w, h):\n\n#     xmax = int((x*width) + (w * width)/2.0)\n#     xmin = int((x*width) - (w * width)/2.0)\n#     ymax = int((y*height) + (h * height)/2.0)\n#     ymin = int((y*height) - (h * height)/2.0)\n\n    xmin = int(x)\n    ymin = int(y)\n    xmax = int(w)\n    ymax = int(h)\n    \n    class_id = int(class_id)\n    return (class_id, xmin, xmax, ymin, ymax)\n\n## path root folder\nROOT = '/kaggle/working/'\n\n## converts coco into xml \ndef xml_transform(root, classes):  \n    class_path  = join(root, 'labels')\n    ids = list()\n    l=os.listdir(class_path)\n    \n    check = '.DS_Store' in l\n    if check == True:\n        l.remove('.DS_Store')\n        \n    # ids=[x.split('.')[0] for x in l]   \n    ids=[x[:-4] for x in l]\n\n    annopath = join(root, 'labels', '%s.txt')\n    imgpath = join(root, 'images', '%s.jpg')\n    \n    os.makedirs(join(root, 'outputs'), exist_ok=True)\n    outpath = join(root, 'outputs', '%s.xml')\n\n    for i in range(len(ids)):\n        try:        \n            img_id = ids[i]\n            if img_id == \"classes\":\n                continue\n            if os.path.exists(outpath % img_id):\n                continue\n            # print(imgpath % img_id)\n            img= cv2.imread(imgpath % img_id)\n            #print(img)\n            height, width, channels = img.shape # pega tamanhos e canais das images\n\n            node_root = Element('annotation')\n            node_folder = SubElement(node_root, 'folder')\n            node_folder.text = 'VOC2007'\n            img_name = img_id + '.jpg'\n    \n            node_filename = SubElement(node_root, 'filename')\n            node_filename.text = img_name\n        \n            node_source= SubElement(node_root, 'source')\n            node_database = SubElement(node_source, 'database')\n            node_database.text = 'Coco database'\n        \n            node_size = SubElement(node_root, 'size')\n            node_width = SubElement(node_size, 'width')\n            node_width.text = str(width)\n    \n            node_height = SubElement(node_size, 'height')\n            node_height.text = str(height)\n\n            node_depth = SubElement(node_size, 'depth')\n            node_depth.text = str(channels)\n\n            node_segmented = SubElement(node_root, 'segmented')\n            node_segmented.text = '0'\n\n            target = (annopath % img_id)\n            if os.path.exists(target):\n                label_norm= np.loadtxt(target).reshape(-1, 5)\n\n                for i in range(len(label_norm)):\n                    labels_conv = label_norm[i]\n                    new_label = unconvert(labels_conv[0], width, height, labels_conv[1], labels_conv[2], labels_conv[3], labels_conv[4])\n                    node_object = SubElement(node_root, 'object')\n                    node_name = SubElement(node_object, 'name')\n                    node_name.text = classes[new_label[0]]\n                \n                    node_pose = SubElement(node_object, 'pose')\n                    node_pose.text = 'Unspecified'\n                \n                \n                    node_truncated = SubElement(node_object, 'truncated')\n                    node_truncated.text = '0'\n                    node_difficult = SubElement(node_object, 'difficult')\n                    node_difficult.text = '0'\n                    node_bndbox = SubElement(node_object, 'bndbox')\n                    node_xmin = SubElement(node_bndbox, 'xmin')\n                    node_xmin.text = str(new_label[1])\n                    node_ymin = SubElement(node_bndbox, 'ymin')\n                    node_ymin.text = str(new_label[3])\n                    node_xmax = SubElement(node_bndbox, 'xmax')\n                    node_xmax.text =  str(new_label[2])\n                    node_ymax = SubElement(node_bndbox, 'ymax')\n                    node_ymax.text = str(new_label[4])\n                    xml = tostring(node_root, pretty_print=True)  \n                    dom = parseString(xml)\n            # print(xml)  \n            f =  open(outpath % img_id, \"wb\")\n            #f = open(os.path.join(outpath, img_id), \"w\")\n            #os.remove(target)\n            f.write(xml)\n            f.close()   \n\n        except Exception as e:\n            # print(e)\n            pass\n\nxml_transform(ROOT, CLASSES)","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:17:56.007006Z","iopub.execute_input":"2022-10-06T01:17:56.007382Z","iopub.status.idle":"2022-10-06T01:18:02.598618Z","shell.execute_reply.started":"2022-10-06T01:17:56.007343Z","shell.execute_reply":"2022-10-06T01:18:02.59757Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!ls /kaggle/working/outputs | wc -l","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:18:02.600545Z","iopub.execute_input":"2022-10-06T01:18:02.601017Z","iopub.status.idle":"2022-10-06T01:18:03.704366Z","shell.execute_reply.started":"2022-10-06T01:18:02.600978Z","shell.execute_reply":"2022-10-06T01:18:03.703144Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!zip -r labels.zip /kaggle/working/labels_/ ","metadata":{"execution":{"iopub.status.busy":"2022-10-06T02:05:40.497287Z","iopub.execute_input":"2022-10-06T02:05:40.497751Z","iopub.status.idle":"2022-10-06T02:05:41.705356Z","shell.execute_reply.started":"2022-10-06T02:05:40.497707Z","shell.execute_reply":"2022-10-06T02:05:41.704087Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!zip -r images.zip /kaggle/working/images_/ ","metadata":{"execution":{"iopub.status.busy":"2022-10-06T02:05:57.991454Z","iopub.execute_input":"2022-10-06T02:05:57.992515Z","iopub.status.idle":"2022-10-06T02:06:05.108119Z","shell.execute_reply.started":"2022-10-06T02:05:57.992439Z","shell.execute_reply":"2022-10-06T02:06:05.106976Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!zip -r output.zip /kaggle/working/outputs/ ","metadata":{"execution":{"iopub.status.busy":"2022-10-06T01:18:11.772739Z","iopub.execute_input":"2022-10-06T01:18:11.773682Z","iopub.status.idle":"2022-10-06T01:18:12.770263Z","shell.execute_reply.started":"2022-10-06T01:18:11.773634Z","shell.execute_reply":"2022-10-06T01:18:12.769151Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"### Prepare training env","metadata":{}},{"cell_type":"markdown","source":"## ","metadata":{}},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}