Within the scope of the Kaggle competition (https://www.kaggle.com/c/rsna-intracranial-hemorrhage-detection), I applied the pretrained model resnext50 using Pytorch to detect acute intracranial haemorrhage and its subtypes. The project is written in Python and can be run with Anaconda notebook. Following files are included: Notebooks: Exploration.ipynb: exploration of pydicom format; Preparation_data.ipynb: create a balanced dataset to use for training; Training_validation.ipynb: preprocessing and application of resnext50 model for training and vaildation. Examples of images: Folder ‘png_examples2’; Folder ‘pydicom_examples’. Files with data: rsna_train.csv, train_50a.csv, train_all.csv, train_any1.csv