Computer Vision News - February 2022

37 FMI and Deep Learning in Medical Imaging fastcore Version: 1.3.20 python Version: 3.8.10 (default, May 19, 2021, 13:12:57) [MSC v.1916 64 bits (AMD64)] torchvision: 0.8.2+cu110 torch version: 1.7.1+cu110 Cuda: True Cuda Version: 11.0 GPU: NVIDIA GeForce RTX 3070 Laptop GPU pydicom Version: 2.1.2 kornia Version: 0.2.0 Looking at the data Getting Data The data consists of 250 dicom images from the SIIM-ACR Pneumothorax Segmentation dataset which consists of chest X-rays with and without pneumothorax. If you remember, we did use a similar dataset for our Kaggle expert tutorial , a few months ago. Of course, you can use your own dataset, or anything that you feel comfortable working with! Let’s unzip (oh well, untar!) our data and have a look at the csv file coming with the dataset. pneu = untar_data(URLs.SIIM_SMALL) p_items = get_dicom_files(f'{pneu}/train') p_items (#250) [Path('C:/Users/avird/.fastai/data/siim_small/train/No Pneumothorax/000000.dcm'),Path('C:/Users/avird/.fastai/data/siim_small/train/No Pneumothorax/000002.dcm'),Path('C:/Users/avird/.fastai/data/siim_small/train/No Pneumothorax/000005.dcm'),Path('C:/Users/avird/.fastai/data/siim_small/train/No Pneumothorax/000006.dcm'),Path('C:/Users/avird/.fastai/data/siim_small/train/No Pneumothorax/000007.dcm'),Path('C:/Users/avird/.fastai/data/siim_small/train/No Pneumothorax/000008.dcm'),Path('C:/Users/avird/.fastai/data/siim_small/train/No Pneumothorax/000009.dcm'),Path('C:/Users/avird/.fastai/data/siim_small/train/No Pneumothorax/000011.dcm'),Path('C:/Users/avird/.fastai/data/siim_small/train/No Pneumothorax/000012.dcm'),Path('C:/Users/avird/.fastai/data/siim_small/train/No Pneumothorax/000014.dcm')...] df = pd.read_csv(f'{pneu}/labels.csv') df[:5] file label 0 train/No Pneumothorax/000000.dcm No Pneumothorax 1 train/Pneumothorax/000001.dcm Pneumothorax 2 train/No Pneumothorax/000002.dcm No Pneumothorax 3 train/Pneumothorax/000003.dcm Pneumothorax 4 train/Pneumothorax/000004.dcm Pneumothorax Metadata from DICOM files