Automatic generation of medical reports from chest X-rays in Czech
|Thesis title in Czech:||Automatické generování lékařských zpráv z rentgenových snímků hrudníku v češtině|
|Thesis title in English:||Automatic generation of medical reports from chest X-rays in Czech|
|Key words:||zpracování obrázků|generování přirozeného jazyka|lékařství|neuronové sítě|
|English key words:||image processing|natural language generation|medical|neural networks|
|Academic year of topic announcement:||2021/2022|
|Type of assignment:||diploma thesis|
|Department:||Institute of Formal and Applied Linguistics (32-UFAL)|
|Supervisor:||Mgr. Rudolf Rosa, Ph.D.|
|Author:||hidden - assigned and confirmed by the Study Dept.|
|Date of registration:||31.03.2022|
|Date of assignment:||31.03.2022|
|Confirmed by Study dept. on:||12.04.2022|
|Date and time of defence:||13.09.2022 09:00|
|Date of electronic submission:||20.07.2022|
|Date of submission of printed version:||25.07.2022|
|Date of proceeded defence:||13.09.2022|
|Reviewers:||Mgr. Jindřich Libovický, Ph.D.|
|The aim of the master thesis is the problem of automatic textual report generation for chest X-ray images. The input is one or more chest X-ray images and the output is an overall textual description (report) of the chest X-ray. This report should describe whether there are any diseases/abnormalities and, if so, they should be described in more detail.
The generation of textual descriptions will take place on the basis of neural networks and the final output will be in the Czech language. Some of openly available medical datasets will be used for training the network, but as they are not in the Czech language, they will be translated automatically using a freely available machine translation system.
The final model created as a result of the thesis will be then evaluated in order to determine the final performance of the model.
The goal is not to replace the doctor examination, as no guarantees can be made about the truthfulness of the model predictions. The thesis should rather produce an auxiliary tool that should help doctors examine X-rays.
|- POPEL, Martin, et al. Transforming machine translation: a deep learning system reaches news translation quality comparable to human professionals. Nature communications, 2020, 11.1: 1-15.
- JOHNSON, Alistair E. W., Tom POLLARD, Roger MARK, Seth BERKOWITZ and Steven HORNG, 2019. MIMIC-CXR Database (version 2.0.0). 2019. B.m.: physionet.org. Available at: doi:10.13026/C2JT1Q
- DEMNER-FUSHMAN, Dina, et al. Preparing a collection of radiology examinations for distribution and retrieval. Journal of the American Medical Informatics Association, 2016, 23.2: 304-310.
- ZENG, Xianhua, et al. Generating diagnostic report for medical image by high-middle-level visual information incorporation on double deep learning models. Computer Methods and Programs in Biomedicine, 2020, 197: 105700.
- ALFARGHALY, Omar, et al. Automated radiology report generation using conditioned transformers. Informatics in Medicine Unlocked, 2021, 24: 100557.
- AYESHA, Hareem, et al. Automatic medical image interpretation: State of the art and future directions. Pattern Recognition, 2021, 114: 107856.
- RADFORD, Alec, et al. Language models are unsupervised multitask learners. OpenAI blog, 2019, 1.8: 9.
- Tensorflow documentation: https://www.tensorflow.org/
- fastai documentation: https://docs.fast.ai/
- Hugging Face documentation: https://huggingface.co/