Back

Feasibility to virtually generate T2 fat-saturated breast MRI by convolutional neural networks

Liebert, A.; Hadler, D.; Schreiter, H.; Ehring, C.; Brock, L.; Kapsner, L. A.; Eberle, J.; Erber, R.; Emons, J.; Laun, F. B.; Uder, M.; Wenkel, E.; Ohlmeyer, S.; Bickelhaupt, S.

2024-06-25 radiology and imaging
10.1101/2024.06.25.24309404 medRxiv
Show abstract

BackgroundBreast magnetic resonance imaging (MRI) protocols often include T2-weighted fat-saturated (T2w-FS) sequences, which are vital for tissue characterization but significantly increase scan time. PurposeThis study aims to evaluate whether a 2D-U-Net neural network can generate virtual T2w-FS images from routine multiparametric breast MRI sequences. Materials and MethodsThis IRB approved, retrospective study included n=914 breast MRI examinations performed between January 2017 and June 2020. The dataset was divided into training (n=665), validation (n=74), and test sets (n=175). The U-Net was trained on T1-weighted (T1w), diffusion-weighted imaging (DWI), and dynamic contrast-enhanced (DCE) sequences to generate virtual T2w-FS images (VirtuT2). Quantitative metrics and a qualitative multi-reader assessment by two radiologists were used to evaluate the VirtuT2 images. ResultsVirtuT2 images demonstrated high structural similarity (SSIM=0.87) and peak signal-to-noise ratio (PSNR=24.90) compared to original T2w-FS images. High level of the frequency error norm (HFNE=0.87) indicates strong blurring presence in the VirtuT2 images, which was also confirmed in qualitative reading. Radiologists correctly identified VirtuT2 images with 92.3% and 94.2% accuracy, respectively. No significant difference in diagnostic image quality (DIQ) was noted for one reader (p=0.21), while the other reported significantly lower DIQ for VirtuT2 (p<=0.001). Moderate inter-reader agreement was observed for edema detection on T2w-FS images (=0.43), decreasing to fair on VirtuT2 images (=0.36). ConclusionThe 2D-U-Net can technically generate virtual T2w-FS images with high similarity to real T2w-FS images, though blurring remains a limitation. Further investigation of other architectures and using larger datasets are needed to improve clinical applicability. Summary StatementVirtual T2-weighted fat-saturated images can be generated from routine breast MRI sequences using convolutional neural networks, showing high structural similarity but with notable blurring, necessitating further refinement for clinical use. Key ResultsO_LIImages with T2w-FS characteristics can be virtually generated from T1w and DWI images using deep learning C_LIO_LIImage blurring occurring in the VirtuT2 image limit clinical use for the current moment C_LIO_LIFurther investigation of different architectures and with larger datasets are necessary in the future to improve the VirtuT2 performance. C_LI

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.