TY - JOUR
T1 - Anatomic and Molecular MR Image Synthesis Using Confidence Guided CNNs
AU - Guo, Pengfei
AU - Wang, Puyang
AU - Yasarla, Rajeev
AU - Zhou, Jinyuan
AU - Patel, Vishal M.
AU - Jiang, Shanshan
N1 - Funding Information:
Manuscript received November 1, 2020; revised December 8, 2020; accepted December 13, 2020. Date of publication December 22, 2020; date of current version September 30, 2021. This work was supported in part by the National Institutes of Health under Grant R01CA228188 and in part by the National Science Foundation under Grant 1910141. (Corresponding author: Shanshan Jiang.) Pengfei Guo, Puyang Wang, Rajeev Yasarla, and Vishal M. Patel are with the Whiting School of Engineering, Johns Hopkins University, Baltimore, MD 21218 USA (e-mail: pguo4@jhu.edu; pwang47@jhu.edu; ryasarl1@jhu.edu; vpatel36@jhu.edu).
Publisher Copyright:
© 1982-2012 IEEE.
PY - 2021/10/1
Y1 - 2021/10/1
N2 - Data-driven automatic approaches have demonstrated their great potential in resolving various clinical diagnostic dilemmas in neuro-oncology, especially with the help of standard anatomic and advanced molecular MR images. However, data quantity and quality remain a key determinant, and a significant limit of the potential applications. In our previous work, we explored the synthesis of anatomic and molecular MR image networks (SAMR) in patients with post-Treatment malignant gliomas. In this work, we extend this through a confidence-guided SAMR (CG-SAMR) that synthesizes data from lesion contour information to multi-modal MR images, including T1-weighted ( {T}-{1}\text{w} ), gadolinium enhanced {T}-{1}\text{w} (Gd-{T}-{1}\text{w} ), T2-weighted ( {T}-{2}\text{w} ), and fluid-Attenuated inversion recovery ( \textit {FLAIR} ), as well as the molecular amide proton transfer-weighted ( \textit {APT}\text{w} ) sequence. We introduce a module that guides the synthesis based on a confidence measure of the intermediate results. Furthermore, we extend the proposed architecture to allow training using unpaired data. Extensive experiments on real clinical data demonstrate that the proposed model can perform better than current the state-of-The-Art synthesis methods. Our code is available at https://github.com/guopengf/CG-SAMR.
AB - Data-driven automatic approaches have demonstrated their great potential in resolving various clinical diagnostic dilemmas in neuro-oncology, especially with the help of standard anatomic and advanced molecular MR images. However, data quantity and quality remain a key determinant, and a significant limit of the potential applications. In our previous work, we explored the synthesis of anatomic and molecular MR image networks (SAMR) in patients with post-Treatment malignant gliomas. In this work, we extend this through a confidence-guided SAMR (CG-SAMR) that synthesizes data from lesion contour information to multi-modal MR images, including T1-weighted ( {T}-{1}\text{w} ), gadolinium enhanced {T}-{1}\text{w} (Gd-{T}-{1}\text{w} ), T2-weighted ( {T}-{2}\text{w} ), and fluid-Attenuated inversion recovery ( \textit {FLAIR} ), as well as the molecular amide proton transfer-weighted ( \textit {APT}\text{w} ) sequence. We introduce a module that guides the synthesis based on a confidence measure of the intermediate results. Furthermore, we extend the proposed architecture to allow training using unpaired data. Extensive experiments on real clinical data demonstrate that the proposed model can perform better than current the state-of-The-Art synthesis methods. Our code is available at https://github.com/guopengf/CG-SAMR.
KW - Generative adversarial network
KW - confidence guidance
KW - glioma
KW - multi-modal MR image synthesis
KW - segmentation
UR - http://www.scopus.com/inward/record.url?scp=85098792747&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85098792747&partnerID=8YFLogxK
U2 - 10.1109/TMI.2020.3046460
DO - 10.1109/TMI.2020.3046460
M3 - Article
C2 - 33351754
AN - SCOPUS:85098792747
SN - 0278-0062
VL - 40
SP - 2832
EP - 2844
JO - IEEE transactions on medical imaging
JF - IEEE transactions on medical imaging
IS - 10
ER -