Synthesising Brain Iron Maps from Quantitative Magnetic Resonance Images Using Interpretable Generative Adversarial Networks

TY - CHAP

T1 - Synthesising Brain Iron Maps from Quantitative Magnetic Resonance Images Using Interpretable Generative Adversarial Networks

AU - Munroe, Lindsay

AU - Deprez, Maria

AU - Michaelides, Christos

AU - Parkes, Harry G.

AU - Geraki, Kalotina

AU - Herlihy, Amy H.

AU - So, Po Wah

N1 - Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.

PY - 2023

Y1 - 2023

N2 - Accurate spatial estimation of brain iron concentration in-vivo is vital to elucidate the role of iron in neurodegenerative diseases, among other applications. However, ground truth quantitative iron maps of the brain can only be acquired post-mortem from ex-vivo samples. Quantitative magnetic resonance imaging (QMRI) methods are iron-sensitive and hold potential to quantitatively measure brain iron. We hypothesise interpretability methods can identify the most salient QMRI parameter(s) for iron prediction. In this study, a generative adversarial network with spatially adaptive normalisation layers (SPADE) was trained to synthesise maps of brain iron content from QMRI parameters, including those from relaxometry, diffusion and magnetisation transfer MRI. Ground truth maps of iron content were obtained by synchrotron radiation X-ray fluorescence (SRXRF). QMRI and SRXRF datasets were registered, and a distribution-based loss was proposed to address misalignment from multi-modal QMRI-to-SRXRF registration. To enable interpretation, channel attention was incorporated to learn feature importance for QMRI parameters. Attention weights were compared against occlusion and local interpretable model-agnostic explanations. Our model achieved dice scores of 0.97 and 0.95 for grey and white matter, respectively, when comparing tissue boundaries of synthesised vs. MRI images. Examining the contrast in predicted vs. ground truth iron maps, our model achieved 15.2% and 17.8% normalised absolute error for grey and white matter, respectively. All three interpretable methods ranked fractional anisotropy as the most salient, followed by myelin water fraction and magnetisation transfer ratio. The co-location of iron and myelin may explain the finding that myelin-related QMRI parameters are strong predictors of iron.

AB - Accurate spatial estimation of brain iron concentration in-vivo is vital to elucidate the role of iron in neurodegenerative diseases, among other applications. However, ground truth quantitative iron maps of the brain can only be acquired post-mortem from ex-vivo samples. Quantitative magnetic resonance imaging (QMRI) methods are iron-sensitive and hold potential to quantitatively measure brain iron. We hypothesise interpretability methods can identify the most salient QMRI parameter(s) for iron prediction. In this study, a generative adversarial network with spatially adaptive normalisation layers (SPADE) was trained to synthesise maps of brain iron content from QMRI parameters, including those from relaxometry, diffusion and magnetisation transfer MRI. Ground truth maps of iron content were obtained by synchrotron radiation X-ray fluorescence (SRXRF). QMRI and SRXRF datasets were registered, and a distribution-based loss was proposed to address misalignment from multi-modal QMRI-to-SRXRF registration. To enable interpretation, channel attention was incorporated to learn feature importance for QMRI parameters. Attention weights were compared against occlusion and local interpretable model-agnostic explanations. Our model achieved dice scores of 0.97 and 0.95 for grey and white matter, respectively, when comparing tissue boundaries of synthesised vs. MRI images. Examining the contrast in predicted vs. ground truth iron maps, our model achieved 15.2% and 17.8% normalised absolute error for grey and white matter, respectively. All three interpretable methods ranked fractional anisotropy as the most salient, followed by myelin water fraction and magnetisation transfer ratio. The co-location of iron and myelin may explain the finding that myelin-related QMRI parameters are strong predictors of iron.

KW - Interpretable deep learning

KW - Iron prediction

KW - Supervised image-to-image translation

UR - http://www.scopus.com/inward/record.url?scp=85185708180&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-47425-5_20

DO - 10.1007/978-3-031-47425-5_20

M3 - Conference paper

AN - SCOPUS:85185708180

SN - 9783031474248

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 214

EP - 226

BT - Medical Image Computing and Computer Assisted Intervention – MICCAI 2023 Workshops - MTSAIL 2023, LEAF 2023, AI4Treat 2023, MMMI 2023, REMIA 2023, Held in Conjunction with MICCAI 2023, Proceedings

A2 - Woo, Jonghye

A2 - Hering, Alessa

A2 - Silva, Wilson

A2 - Li, Xiang

A2 - Fu, Huazhu

A2 - Liu, Xiaofeng

A2 - Xing, Fangxu

A2 - Purushotham, Sanjay

A2 - Mathai, T.S.

A2 - Mukherjee, Pritam

A2 - De Grauw, Max

A2 - Beets Tan, Regina

A2 - Corbetta, Valentina

A2 - Kotter, Elmar

A2 - Reyes, Mauricio

A2 - Baumgartner, C.F.

A2 - Li, Quanzheng

A2 - Leahy, Richard

A2 - Dong, Bin

A2 - Chen, Hao

A2 - Huo, Yuankai

A2 - Lv, Jinglei

A2 - Xu, Xinxing

A2 - Li, Xiaomeng

A2 - Mahapatra, Dwarikanath

A2 - Cheng, Li

A2 - Petitjean, Caroline

A2 - Presles, Benoît

PB - Springer Science and Business Media Deutschland GmbH

T2 - 26th International Conference on Medical Image Computing and Computer-Assisted Intervention , MICCAI 2023

Y2 - 8 October 2023 through 12 October 2023

ER -