用扩散概率模型和生成对抗模型减少乳腺磁共振成像中的造影剂剂量

IF 3.7 Q1 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING
Gustav Müller-Franzes, Luisa Huck, Maike Bode, Sven Nebelung, Christiane Kuhl, Daniel Truhn, Teresa Lemainque
{"title":"用扩散概率模型和生成对抗模型减少乳腺磁共振成像中的造影剂剂量","authors":"Gustav Müller-Franzes, Luisa Huck, Maike Bode, Sven Nebelung, Christiane Kuhl, Daniel Truhn, Teresa Lemainque","doi":"10.1186/s41747-024-00451-3","DOIUrl":null,"url":null,"abstract":"<h3 data-test=\"abstract-sub-heading\">Background</h3><p>To compare denoising diffusion probabilistic models (DDPM) and generative adversarial networks (GAN) for recovering contrast-enhanced breast magnetic resonance imaging (MRI) subtraction images from virtual low-dose subtraction images.</p><h3 data-test=\"abstract-sub-heading\">Methods</h3><p>Retrospective, ethically approved study. DDPM- and GAN-reconstructed single-slice subtraction images of 50 breasts with enhancing lesions were compared to original ones at three dose levels (25%, 10%, 5%) using quantitative measures and radiologic evaluations. Two radiologists stated their preference based on the reconstruction quality and scored the lesion conspicuity as compared to the original, blinded to the model. Fifty lesion-free maximum intensity projections were evaluated for the presence of false-positives. Results were compared between models and dose levels, using generalized linear mixed models.</p><h3 data-test=\"abstract-sub-heading\">Results</h3><p>At 5% dose, both radiologists preferred the GAN-generated images, whereas at 25% dose, both radiologists preferred the DDPM-generated images. Median lesion conspicuity scores did not differ between GAN and DDPM at 25% dose (5 <i>versus</i> 5, <i>p</i> = 1.000) and 10% dose (4 <i>versus</i> 4, <i>p</i> = 1.000). At 5% dose, both readers assigned higher conspicuity to the GAN than to the DDPM (3 <i>versus</i> 2, <i>p</i> = 0.007). In the lesion-free examinations, DDPM and GAN showed no differences in the false-positive rate at 5% (15% <i>versus</i> 22%), 10% (10% <i>versus</i> 6%), and 25% (6% <i>versus</i> 4%) (<i>p</i> = 1.000).</p><h3 data-test=\"abstract-sub-heading\">Conclusions</h3><p>Both GAN and DDPM yielded promising results in low-dose image reconstruction. However, neither of them showed superior results over the other model for all dose levels and evaluation metrics. Further development is needed to counteract false-positives.</p><h3 data-test=\"abstract-sub-heading\">Relevance statement</h3><p>For MRI-based breast cancer screening, reducing the contrast agent dose is desirable. Diffusion probabilistic models and generative adversarial networks were capable of retrospectively enhancing the signal of low-dose images. Hence, they may supplement imaging with reduced doses in the future.</p><h3 data-test=\"abstract-sub-heading\">Key points</h3><p>• Deep learning may help recover signal in low-dose contrast-enhanced breast MRI.</p><p>• Two models (DDPM and GAN) were trained at different dose levels.</p><p>• Radiologists preferred DDPM at 25%, and GAN images at 5% dose.</p><p>• Lesion conspicuity between DDPM and GAN was similar, except at 5% dose.</p><p>• GAN and DDPM yield promising results in low-dose image reconstruction.</p><h3 data-test=\"abstract-sub-heading\">Graphical Abstract</h3>\n","PeriodicalId":36926,"journal":{"name":"European Radiology Experimental","volume":null,"pages":null},"PeriodicalIF":3.7000,"publicationDate":"2024-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Diffusion probabilistic versus generative adversarial models to reduce contrast agent dose in breast MRI\",\"authors\":\"Gustav Müller-Franzes, Luisa Huck, Maike Bode, Sven Nebelung, Christiane Kuhl, Daniel Truhn, Teresa Lemainque\",\"doi\":\"10.1186/s41747-024-00451-3\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<h3 data-test=\\\"abstract-sub-heading\\\">Background</h3><p>To compare denoising diffusion probabilistic models (DDPM) and generative adversarial networks (GAN) for recovering contrast-enhanced breast magnetic resonance imaging (MRI) subtraction images from virtual low-dose subtraction images.</p><h3 data-test=\\\"abstract-sub-heading\\\">Methods</h3><p>Retrospective, ethically approved study. DDPM- and GAN-reconstructed single-slice subtraction images of 50 breasts with enhancing lesions were compared to original ones at three dose levels (25%, 10%, 5%) using quantitative measures and radiologic evaluations. Two radiologists stated their preference based on the reconstruction quality and scored the lesion conspicuity as compared to the original, blinded to the model. Fifty lesion-free maximum intensity projections were evaluated for the presence of false-positives. Results were compared between models and dose levels, using generalized linear mixed models.</p><h3 data-test=\\\"abstract-sub-heading\\\">Results</h3><p>At 5% dose, both radiologists preferred the GAN-generated images, whereas at 25% dose, both radiologists preferred the DDPM-generated images. Median lesion conspicuity scores did not differ between GAN and DDPM at 25% dose (5 <i>versus</i> 5, <i>p</i> = 1.000) and 10% dose (4 <i>versus</i> 4, <i>p</i> = 1.000). At 5% dose, both readers assigned higher conspicuity to the GAN than to the DDPM (3 <i>versus</i> 2, <i>p</i> = 0.007). In the lesion-free examinations, DDPM and GAN showed no differences in the false-positive rate at 5% (15% <i>versus</i> 22%), 10% (10% <i>versus</i> 6%), and 25% (6% <i>versus</i> 4%) (<i>p</i> = 1.000).</p><h3 data-test=\\\"abstract-sub-heading\\\">Conclusions</h3><p>Both GAN and DDPM yielded promising results in low-dose image reconstruction. However, neither of them showed superior results over the other model for all dose levels and evaluation metrics. Further development is needed to counteract false-positives.</p><h3 data-test=\\\"abstract-sub-heading\\\">Relevance statement</h3><p>For MRI-based breast cancer screening, reducing the contrast agent dose is desirable. Diffusion probabilistic models and generative adversarial networks were capable of retrospectively enhancing the signal of low-dose images. Hence, they may supplement imaging with reduced doses in the future.</p><h3 data-test=\\\"abstract-sub-heading\\\">Key points</h3><p>• Deep learning may help recover signal in low-dose contrast-enhanced breast MRI.</p><p>• Two models (DDPM and GAN) were trained at different dose levels.</p><p>• Radiologists preferred DDPM at 25%, and GAN images at 5% dose.</p><p>• Lesion conspicuity between DDPM and GAN was similar, except at 5% dose.</p><p>• GAN and DDPM yield promising results in low-dose image reconstruction.</p><h3 data-test=\\\"abstract-sub-heading\\\">Graphical Abstract</h3>\\n\",\"PeriodicalId\":36926,\"journal\":{\"name\":\"European Radiology Experimental\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":3.7000,\"publicationDate\":\"2024-05-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"European Radiology Experimental\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1186/s41747-024-00451-3\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"European Radiology Experimental","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1186/s41747-024-00451-3","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING","Score":null,"Total":0}
引用次数: 0

摘要

背景比较去噪扩散概率模型(DDPM)和生成对抗网络(GAN)从虚拟低剂量减影图像中恢复对比度增强乳腺磁共振成像(MRI)减影图像的方法。在三个剂量水平(25%、10%、5%)下,采用定量测量和放射学评估方法,将 50 个有增强病灶的乳房的 DDPM 和 GAN 重建的单片减影图像与原始图像进行比较。两位放射科医生根据重建质量提出了他们的偏好,并在对模型保密的情况下,对与原始图像相比病变的清晰度进行了评分。对 50 个无病灶的最大强度投影进行了评估,以确定是否存在假阳性。使用广义线性混合模型对不同模型和剂量水平的结果进行比较。结果5%剂量时,两位放射科医生都更喜欢GAN生成的图像,而25%剂量时,两位放射科医生都更喜欢DDPM生成的图像。在 25% 剂量(5 分对 5 分,p = 1.000)和 10% 剂量(4 分对 4 分,p = 1.000)时,GAN 和 DDPM 的中位病变清晰度评分没有差异。在 5%剂量时,两位读者都认为 GAN 比 DDPM 更清晰(3 比 2,p = 0.007)。在无病灶检查中,DDPM 和 GAN 的假阳性率在 5%(15% 对 22%)、10%(10% 对 6%)和 25%(6% 对 4%)时没有差异(p = 1.000)。然而,在所有剂量水平和评估指标上,它们都没有显示出优于其他模型的结果。要消除假阳性,还需要进一步的发展。相关声明对于基于磁共振成像的乳腺癌筛查,降低造影剂剂量是可取的。扩散概率模型和生成对抗网络能够回溯性地增强低剂量图像的信号。关键点- 深度学习可帮助恢复低剂量造影剂增强乳腺 MRI 的信号- 在不同剂量水平下训练了两种模型(DDPM 和 GAN)- 放射科医生更喜欢 25% 剂量下的 DDPM 图像和 5% 剂量下的 GAN 图像- 除 5% 剂量外,DDPM 和 GAN 的病变清晰度相似- GAN 和 DDPM 在低剂量图像重建中取得了可喜的成果。
本文章由计算机程序翻译,如有差异,请以英文原文为准。

Diffusion probabilistic versus generative adversarial models to reduce contrast agent dose in breast MRI

Diffusion probabilistic versus generative adversarial models to reduce contrast agent dose in breast MRI

Background

To compare denoising diffusion probabilistic models (DDPM) and generative adversarial networks (GAN) for recovering contrast-enhanced breast magnetic resonance imaging (MRI) subtraction images from virtual low-dose subtraction images.

Methods

Retrospective, ethically approved study. DDPM- and GAN-reconstructed single-slice subtraction images of 50 breasts with enhancing lesions were compared to original ones at three dose levels (25%, 10%, 5%) using quantitative measures and radiologic evaluations. Two radiologists stated their preference based on the reconstruction quality and scored the lesion conspicuity as compared to the original, blinded to the model. Fifty lesion-free maximum intensity projections were evaluated for the presence of false-positives. Results were compared between models and dose levels, using generalized linear mixed models.

Results

At 5% dose, both radiologists preferred the GAN-generated images, whereas at 25% dose, both radiologists preferred the DDPM-generated images. Median lesion conspicuity scores did not differ between GAN and DDPM at 25% dose (5 versus 5, p = 1.000) and 10% dose (4 versus 4, p = 1.000). At 5% dose, both readers assigned higher conspicuity to the GAN than to the DDPM (3 versus 2, p = 0.007). In the lesion-free examinations, DDPM and GAN showed no differences in the false-positive rate at 5% (15% versus 22%), 10% (10% versus 6%), and 25% (6% versus 4%) (p = 1.000).

Conclusions

Both GAN and DDPM yielded promising results in low-dose image reconstruction. However, neither of them showed superior results over the other model for all dose levels and evaluation metrics. Further development is needed to counteract false-positives.

Relevance statement

For MRI-based breast cancer screening, reducing the contrast agent dose is desirable. Diffusion probabilistic models and generative adversarial networks were capable of retrospectively enhancing the signal of low-dose images. Hence, they may supplement imaging with reduced doses in the future.

Key points

• Deep learning may help recover signal in low-dose contrast-enhanced breast MRI.

• Two models (DDPM and GAN) were trained at different dose levels.

• Radiologists preferred DDPM at 25%, and GAN images at 5% dose.

• Lesion conspicuity between DDPM and GAN was similar, except at 5% dose.

• GAN and DDPM yield promising results in low-dose image reconstruction.

Graphical Abstract

求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
European Radiology Experimental
European Radiology Experimental Medicine-Radiology, Nuclear Medicine and Imaging
CiteScore
6.70
自引率
2.60%
发文量
56
审稿时长
18 weeks
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信