Document is current

Article History

Received: 15 December 2021

Revised: 13 May 2022

Accepted: 19 May 2022

First Online: 17 June 2022

Declarations

:

: The scientific guarantor of this publication is Yikai Xu.

: One of the authors (Yuting Liao) is an employee of GE Healthcare and one of the authors (Kan Deng) is an employee of Philips Healthcare. The remaining authors declare no relationships with any companies whose products or services may be related to the subject matter of the article.

: No complex statistical methods were necessary for this paper.

: Written informed consent was waived by the Institutional Review Board.

: Institutional Review Board approval was obtained.

: In fact, 98 patients in our study overlapped with a previous study (An attention-based deep learning model for predicting Microvascular invasion of hepatocellular carcinoma using an intravoxel incoherent motion model of diffusion-weighted magnetic Resonance imaging). This previous study was a methodological study aimed at proposing a new approach for processing IVIM data and exploring an attentional technique to improve the performance of convolutional neural networks in tumor characterization. However, our current study mainly evaluated the performance of CNN based on IVIM in predicting MVI of HCC lesions. Furthermore, we had included the latest cases to expand the sample size. In addition, our study also evaluated the performance of clinical characteristics, IVIM parameter values, deep learning model based on IVIM parameter maps and the fusion model combined deep features of IVIM, clinical characteristics, and ADC, which could better show the usefulness of the CNN model based on IVIM data for MVI prediction in clinical practice.

: • retrospective• diagnostic or prognostic study• performed at one institution