Addressing the Generalizability of AI in Radiology Using a Novel Data Augmentation Framework with Synthetic Patient Image Data: Proof-of-Concept and External Validation for Classification Tasks in Multiple Sclerosis
Authors:
- Gianluca BrugnaraGBGianluca Brugnara
- Chandrakanth Jayachandran PreethaCPChandrakanth Jayachandran Preetha
- Katerina DeikeKDKaterina Deike
- Robert HaaseRHRobert Haase
- Thomas PinetzTPThomas Pinetz
- Martha Foltyn-DumitruMFMartha Foltyn-Dumitru
- Mustafa A. MahmutogluMMMustafa A. Mahmutoglu
- Brigitte WildemannBWBrigitte Wildemann
- Ricarda DiemRDRicarda Diem
- Wolfgang WickWWWolfgang Wick
- Alexander RadbruchARAlexander Radbruch
- Martin BendszusMBMartin Bendszus
- 1 November 2024
- journal article
- Published by Radiological Society of North America (RSNA) in Radiology: Artificial Intelligence
- Vol. 6 (6) , e230514
Abstract
Generative adversarial network–based synthetic data augmentation during model training substantially improved performance of artificial intelligence models when applied to external MRI data and may be extended to a variety of clinical tasks to reduce domain shift on unseen data.Keywords
This publication has 20 references indexed in Scilit:
- Systematic Review of Artificial Intelligence for Abnormality Detection in High-volume Neuroimaging and Subgroup Meta-analysis for Intracranial Hemorrhage DetectionClinical Neuroradiology, 2023
- Lessons learned from translating AI from development to deployment in healthcareNature Medicine, 2023
- The Current and Future State of AI Interpretation of Medical ImagesNew England Journal of Medicine, 2023
- Federated learning enables big data for rare cancer boundary detectionNature Communications, 2022
- 2021 MAGNIMS–CMSC–NAIMS consensus recommendations on the use of MRI in patients with multiple sclerosisThe Lancet Neurology, 2021
- Swarm Learning for decentralized and confidential clinical machine learningNature, 2021
- Blind MRI Brain Lesion Inpainting Using Deep Learning2020
- Automated volumetric assessment with artificial neural networks might enable a more accurate assessment of disease burden in patients with multiple sclerosisEuropean Radiology, 2020
- A fully convolutional neural network for new T2-w lesion detection in multiple sclerosisNeuroImage: Clinical, 2020
- Segmentation of brain MR images through a hidden Markov random field model and the expectation-maximization algorithmIEEE Transactions on Medical Imaging, 2001