Self-supervised neural network improves tri-exponential intravoxel incoherent motion model fitting compared to least-squares fitting in non-alcoholic fatty liver disease
Recent literature suggests that tri-exponential models may provide additional information and fit liver intravoxel incoherent motion (IVIM) data more accurately than conventional bi-exponential models. However, voxel-wise fitting of IVIM results in noisy and unreliable parameter maps. For bi-exponen...
| Main Authors: | , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2022-09-01
|
| Series: | Frontiers in Physiology |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fphys.2022.942495/full |
| _version_ | 1798002302079991808 |
|---|---|
| author | Marian A. Troelstra Anne-Marieke Van Dijk Julia J. Witjes Anne Linde Mak Diona Zwirs Jurgen H. Runge Joanne Verheij Ulrich H. Beuers Max Nieuwdorp Adriaan G. Holleboom Aart J. Nederveen Oliver J. Gurney-Champion |
| author_facet | Marian A. Troelstra Anne-Marieke Van Dijk Julia J. Witjes Anne Linde Mak Diona Zwirs Jurgen H. Runge Joanne Verheij Ulrich H. Beuers Max Nieuwdorp Adriaan G. Holleboom Aart J. Nederveen Oliver J. Gurney-Champion |
| author_sort | Marian A. Troelstra |
| collection | DOAJ |
| description | Recent literature suggests that tri-exponential models may provide additional information and fit liver intravoxel incoherent motion (IVIM) data more accurately than conventional bi-exponential models. However, voxel-wise fitting of IVIM results in noisy and unreliable parameter maps. For bi-exponential IVIM, neural networks (NN) were able to produce superior parameter maps than conventional least-squares (LSQ) generated images. Hence, to improve parameter map quality of tri-exponential IVIM, we developed an unsupervised physics-informed deep neural network (IVIM3-NET). We assessed its performance in simulations and in patients with non-alcoholic fatty liver disease (NAFLD) and compared outcomes with bi-exponential LSQ and NN fits and tri-exponential LSQ fits. Scanning was performed using a 3.0T free-breathing multi-slice diffusion-weighted single-shot echo-planar imaging sequence with 18 b-values. Images were analysed for visual quality, comparing the bi- and tri-exponential IVIM models for LSQ fits and NN fits using parameter-map signal-to-noise ratios (SNR) and adjusted R2. IVIM parameters were compared to histological fibrosis, disease activity and steatosis grades. Parameter map quality improved with bi- and tri-exponential NN approaches, with a significant increase in average parameter-map SNR from 3.38 to 5.59 and 2.45 to 4.01 for bi- and tri-exponential LSQ and NN models respectively. In 33 out of 36 patients, the tri-exponential model exhibited higher adjusted R2 values than the bi-exponential model. Correlating IVIM data to liver histology showed that the bi- and tri-exponential NN outperformed both LSQ models for the majority of IVIM parameters (10 out of 15 significant correlations). Overall, our results support the use of a tri-exponential IVIM model in NAFLD. We show that the IVIM3-NET can be used to improve image quality compared to a tri-exponential LSQ fit and provides promising correlations with histopathology similar to the bi-exponential neural network fit, while generating potentially complementary additional parameters. |
| first_indexed | 2024-04-11T11:49:46Z |
| format | Article |
| id | doaj.art-7d3f47e3e2a64397869ab0de5bffe129 |
| institution | Directory Open Access Journal |
| issn | 1664-042X |
| language | English |
| last_indexed | 2024-04-11T11:49:46Z |
| publishDate | 2022-09-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Physiology |
| spelling | doaj.art-7d3f47e3e2a64397869ab0de5bffe1292022-12-22T04:25:24ZengFrontiers Media S.A.Frontiers in Physiology1664-042X2022-09-011310.3389/fphys.2022.942495942495Self-supervised neural network improves tri-exponential intravoxel incoherent motion model fitting compared to least-squares fitting in non-alcoholic fatty liver diseaseMarian A. Troelstra0Anne-Marieke Van Dijk1Julia J. Witjes2Anne Linde Mak3Diona Zwirs4Jurgen H. Runge5Joanne Verheij6Ulrich H. Beuers7Max Nieuwdorp8Adriaan G. Holleboom9Aart J. Nederveen10Oliver J. Gurney-Champion11Department of Radiology and Nuclear Medicine, Amsterdam UMC, Amsterdam, NetherlandsDepartment of Vascular Medicine, Amsterdam UMC, Amsterdam, NetherlandsDepartment of Vascular Medicine, Amsterdam UMC, Amsterdam, NetherlandsDepartment of Vascular Medicine, Amsterdam UMC, Amsterdam, NetherlandsDepartment of Vascular Medicine, Amsterdam UMC, Amsterdam, NetherlandsDepartment of Radiology and Nuclear Medicine, Amsterdam UMC, Amsterdam, NetherlandsDepartment of Pathology, Amsterdam UMC, Amsterdam, NetherlandsDepartment of Gastroenterology and Hepatology, Amsterdam UMC, Amsterdam, NetherlandsDepartment of Vascular Medicine, Amsterdam UMC, Amsterdam, NetherlandsDepartment of Vascular Medicine, Amsterdam UMC, Amsterdam, NetherlandsDepartment of Radiology and Nuclear Medicine, Amsterdam UMC, Amsterdam, NetherlandsDepartment of Radiology and Nuclear Medicine, Amsterdam UMC, Amsterdam, NetherlandsRecent literature suggests that tri-exponential models may provide additional information and fit liver intravoxel incoherent motion (IVIM) data more accurately than conventional bi-exponential models. However, voxel-wise fitting of IVIM results in noisy and unreliable parameter maps. For bi-exponential IVIM, neural networks (NN) were able to produce superior parameter maps than conventional least-squares (LSQ) generated images. Hence, to improve parameter map quality of tri-exponential IVIM, we developed an unsupervised physics-informed deep neural network (IVIM3-NET). We assessed its performance in simulations and in patients with non-alcoholic fatty liver disease (NAFLD) and compared outcomes with bi-exponential LSQ and NN fits and tri-exponential LSQ fits. Scanning was performed using a 3.0T free-breathing multi-slice diffusion-weighted single-shot echo-planar imaging sequence with 18 b-values. Images were analysed for visual quality, comparing the bi- and tri-exponential IVIM models for LSQ fits and NN fits using parameter-map signal-to-noise ratios (SNR) and adjusted R2. IVIM parameters were compared to histological fibrosis, disease activity and steatosis grades. Parameter map quality improved with bi- and tri-exponential NN approaches, with a significant increase in average parameter-map SNR from 3.38 to 5.59 and 2.45 to 4.01 for bi- and tri-exponential LSQ and NN models respectively. In 33 out of 36 patients, the tri-exponential model exhibited higher adjusted R2 values than the bi-exponential model. Correlating IVIM data to liver histology showed that the bi- and tri-exponential NN outperformed both LSQ models for the majority of IVIM parameters (10 out of 15 significant correlations). Overall, our results support the use of a tri-exponential IVIM model in NAFLD. We show that the IVIM3-NET can be used to improve image quality compared to a tri-exponential LSQ fit and provides promising correlations with histopathology similar to the bi-exponential neural network fit, while generating potentially complementary additional parameters.https://www.frontiersin.org/articles/10.3389/fphys.2022.942495/fullmagnetic resonance imagingdiffusion magnetic resonance imagingnon-alcoholic fatty liver diseaseintravoxel incoherent motion (IVIM)deep learningtri-exponential |
| spellingShingle | Marian A. Troelstra Anne-Marieke Van Dijk Julia J. Witjes Anne Linde Mak Diona Zwirs Jurgen H. Runge Joanne Verheij Ulrich H. Beuers Max Nieuwdorp Adriaan G. Holleboom Aart J. Nederveen Oliver J. Gurney-Champion Self-supervised neural network improves tri-exponential intravoxel incoherent motion model fitting compared to least-squares fitting in non-alcoholic fatty liver disease Frontiers in Physiology magnetic resonance imaging diffusion magnetic resonance imaging non-alcoholic fatty liver disease intravoxel incoherent motion (IVIM) deep learning tri-exponential |
| title | Self-supervised neural network improves tri-exponential intravoxel incoherent motion model fitting compared to least-squares fitting in non-alcoholic fatty liver disease |
| title_full | Self-supervised neural network improves tri-exponential intravoxel incoherent motion model fitting compared to least-squares fitting in non-alcoholic fatty liver disease |
| title_fullStr | Self-supervised neural network improves tri-exponential intravoxel incoherent motion model fitting compared to least-squares fitting in non-alcoholic fatty liver disease |
| title_full_unstemmed | Self-supervised neural network improves tri-exponential intravoxel incoherent motion model fitting compared to least-squares fitting in non-alcoholic fatty liver disease |
| title_short | Self-supervised neural network improves tri-exponential intravoxel incoherent motion model fitting compared to least-squares fitting in non-alcoholic fatty liver disease |
| title_sort | self supervised neural network improves tri exponential intravoxel incoherent motion model fitting compared to least squares fitting in non alcoholic fatty liver disease |
| topic | magnetic resonance imaging diffusion magnetic resonance imaging non-alcoholic fatty liver disease intravoxel incoherent motion (IVIM) deep learning tri-exponential |
| url | https://www.frontiersin.org/articles/10.3389/fphys.2022.942495/full |
| work_keys_str_mv | AT marianatroelstra selfsupervisedneuralnetworkimprovestriexponentialintravoxelincoherentmotionmodelfittingcomparedtoleastsquaresfittinginnonalcoholicfattyliverdisease AT annemariekevandijk selfsupervisedneuralnetworkimprovestriexponentialintravoxelincoherentmotionmodelfittingcomparedtoleastsquaresfittinginnonalcoholicfattyliverdisease AT juliajwitjes selfsupervisedneuralnetworkimprovestriexponentialintravoxelincoherentmotionmodelfittingcomparedtoleastsquaresfittinginnonalcoholicfattyliverdisease AT annelindemak selfsupervisedneuralnetworkimprovestriexponentialintravoxelincoherentmotionmodelfittingcomparedtoleastsquaresfittinginnonalcoholicfattyliverdisease AT dionazwirs selfsupervisedneuralnetworkimprovestriexponentialintravoxelincoherentmotionmodelfittingcomparedtoleastsquaresfittinginnonalcoholicfattyliverdisease AT jurgenhrunge selfsupervisedneuralnetworkimprovestriexponentialintravoxelincoherentmotionmodelfittingcomparedtoleastsquaresfittinginnonalcoholicfattyliverdisease AT joanneverheij selfsupervisedneuralnetworkimprovestriexponentialintravoxelincoherentmotionmodelfittingcomparedtoleastsquaresfittinginnonalcoholicfattyliverdisease AT ulrichhbeuers selfsupervisedneuralnetworkimprovestriexponentialintravoxelincoherentmotionmodelfittingcomparedtoleastsquaresfittinginnonalcoholicfattyliverdisease AT maxnieuwdorp selfsupervisedneuralnetworkimprovestriexponentialintravoxelincoherentmotionmodelfittingcomparedtoleastsquaresfittinginnonalcoholicfattyliverdisease AT adriaangholleboom selfsupervisedneuralnetworkimprovestriexponentialintravoxelincoherentmotionmodelfittingcomparedtoleastsquaresfittinginnonalcoholicfattyliverdisease AT aartjnederveen selfsupervisedneuralnetworkimprovestriexponentialintravoxelincoherentmotionmodelfittingcomparedtoleastsquaresfittinginnonalcoholicfattyliverdisease AT oliverjgurneychampion selfsupervisedneuralnetworkimprovestriexponentialintravoxelincoherentmotionmodelfittingcomparedtoleastsquaresfittinginnonalcoholicfattyliverdisease |