Automatic reporting for Multiple Sclerosis based on medical imaging and Deep Learning
SONAR|HES-SO
- Morel, Benjamin
- Depeursinge, Adrien (Degree supervisor)
- Sierre : Haute Ecole de Gestion Valais, 2023
Bachelor of Science HES-SO (BSc) in Business Information Technology: Haute Ecole de Gestion Valais, 2023
English
Context: The field of medical imaging has progressed considerably with the integration of artificial intelligence. Today, many algorithms are on par with human experts, particularly for tasks such as organ analysis and segmentation. Despite these remarkable achievements, a crucial challenge persists in effectively translating the outputs generated by algorithms into easily comprehensible radiology reports.
Objectives: Firstly, it aims to explore the possibilities offered by the DICOM SR standard format, and to understand how a structured report can be generated and visualized. Secondly, it seeks to determine the most appropriate format and technology for generating structured reports that summarize the outputs of AI tools. Finally, the project aims to implement an end-to-end pipeline for report generation, seamlessly integrating this mechanism into existing medical systems.
Result: The outcome of this project is a fully functional report generation mechanism tailored to MRI images of Multiple Sclerosis. The generated reports adhere to the DICOM SR standard format, ensuring compatibility with various medical systems including the PACS, and are presented in a clear, user-friendly format.
Conclusion: This project represents an important step towards bridging the gap between AI-generated outputs and radiology reports. By leveraging the DICOM SR standard and using an end-to-end pipeline, it enables the production of Structured Reports that improve the quality and accessibility of crucial medical information. Also, future research challenges are identified to encourage further improvements of the methods.
Objectives: Firstly, it aims to explore the possibilities offered by the DICOM SR standard format, and to understand how a structured report can be generated and visualized. Secondly, it seeks to determine the most appropriate format and technology for generating structured reports that summarize the outputs of AI tools. Finally, the project aims to implement an end-to-end pipeline for report generation, seamlessly integrating this mechanism into existing medical systems.
Result: The outcome of this project is a fully functional report generation mechanism tailored to MRI images of Multiple Sclerosis. The generated reports adhere to the DICOM SR standard format, ensuring compatibility with various medical systems including the PACS, and are presented in a clear, user-friendly format.
Conclusion: This project represents an important step towards bridging the gap between AI-generated outputs and radiology reports. By leveraging the DICOM SR standard and using an end-to-end pipeline, it enables the production of Structured Reports that improve the quality and accessibility of crucial medical information. Also, future research challenges are identified to encourage further improvements of the methods.
- Language
-
- English
- Classification
- Computer science and technology
- Notes
-
- Haute Ecole de Gestion Valais
- Informatique de gestion - Wirtschaftsinformatik
- hesso:hegvs
- Persistent URL
- https://folia.unifr.ch/global/documents/326903
Statistics
Document views: 120
File downloads:
- Morel_Benjamin_2023.pdf: 167