Comments on "DIRECT-Net: A unified mutual-domain material decomposition network for quantitative dual-energy CT Imaging'', Med. Phys. 2022, Vol. 49, pgs. 917-934

TL;DR

This paper comments on the DIRECT-Net deep learning approach for direct quantitative image reconstruction in dual-energy CT, discussing its methodology, implications, and providing critical insights within the context of ongoing research.

Contribution

The authors provide a critical commentary on the DIRECT-Net method, highlighting its strengths and limitations in the context of dual-energy CT image reconstruction.

Findings

Highlights the potential of deep learning in dual-energy CT reconstruction

Identifies limitations and areas for improvement in DIRECT-Net approach

Provides insights that guide future research in quantitative CT imaging

Abstract

Quantitative image reconstruction in dual-energy computed tomography (CT) remains a topic of active research. We read with interest ``DIRECT-Net: A unified mutual-domain material decomposition network for quantitative dual-energy CT imaging,'' which appears in the 2022 February Issue of Med Phys. In the paper the authors propose a deep-learning (DL) method, referred to as the Direct-Net method, to address the problem of quantitative image reconstruction directly from data in full-scan dual-energy CT (DECT). We comment on the study and conclusion in the paper. The Reply to this comment appears under Communications on medphys.org: https://www.medphys.org/Communications/Reply-Pan_Response-Su.pdf In order to have context for the Reply, we provide the full text of our Comments.