将放射科医生的知识纳入MRI质量指标,使用基于排名的评级进行机器学习。

IF 3.3 2区 医学 Q1 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING Journal of Magnetic Resonance Imaging Pub Date : 2024-12-17 DOI:10.1002/jmri.29672
Chenwei Tang, Laura B Eisenmenger, Leonardo Rivera-Rivera, Eugene Huo, Jacqueline C Junn, Anthony D Kuner, Thekla H Oechtering, Anthony Peret, Jitka Starekova, Kevin M Johnson
{"title":"将放射科医生的知识纳入MRI质量指标,使用基于排名的评级进行机器学习。","authors":"Chenwei Tang, Laura B Eisenmenger, Leonardo Rivera-Rivera, Eugene Huo, Jacqueline C Junn, Anthony D Kuner, Thekla H Oechtering, Anthony Peret, Jitka Starekova, Kevin M Johnson","doi":"10.1002/jmri.29672","DOIUrl":null,"url":null,"abstract":"<p><strong>Background: </strong>Deep learning (DL) often requires an image quality metric; however, widely used metrics are not designed for medical images.</p><p><strong>Purpose: </strong>To develop an image quality metric that is specific to MRI using radiologists image rankings and DL models.</p><p><strong>Study type: </strong>Retrospective.</p><p><strong>Population: </strong>A total of 19,344 rankings on 2916 unique image pairs from the NYU fastMRI Initiative neuro database was used for the neural network-based image quality metrics training with an 80%/20% training/validation split and fivefold cross-validation.</p><p><strong>Field strength/sequence: </strong>1.5 T and 3 T T1, T1 postcontrast, T2, and FLuid Attenuated Inversion Recovery (FLAIR).</p><p><strong>Assessment: </strong>Synthetically corrupted image pairs were ranked by radiologists (N = 7), with a subset also scoring images using a Likert scale (N = 2). DL models were trained to match rankings using two architectures (EfficientNet and IQ-Net) with and without reference image subtraction and compared to ranking based on mean squared error (MSE) and structural similarity (SSIM). Image quality assessing DL models were evaluated as alternatives to MSE and SSIM as optimization targets for DL denoising and reconstruction.</p><p><strong>Statistical tests: </strong>Radiologists' agreement was assessed by a percentage metric and quadratic weighted Cohen's kappa. Ranking accuracies were compared using repeated measurements analysis of variance. Reconstruction models trained with IQ-Net score, MSE and SSIM were compared by paired t test. P < 0.05 was considered significant.</p><p><strong>Results: </strong>Compared to direct Likert scoring, ranking produced a higher level of agreement between radiologists (70.4% vs. 25%). Image ranking was subjective with a high level of intraobserver agreement ( <math> <semantics><mrow><mn>94.9</mn> <mo>%</mo> <mo>±</mo> <mn>2.4</mn> <mo>%</mo></mrow> <annotation>$$ 94.9\\%\\pm 2.4\\% $$</annotation></semantics> </math> ) and lower interobserver agreement ( <math> <semantics><mrow><mn>61.47</mn> <mo>%</mo> <mo>±</mo> <mn>5.51</mn> <mo>%</mo></mrow> <annotation>$$ 61.47\\%\\pm 5.51\\% $$</annotation></semantics> </math> ). IQ-Net and EfficientNet accurately predicted rankings with a reference image ( <math> <semantics><mrow><mn>75.2</mn> <mo>%</mo> <mo>±</mo> <mn>1.3</mn> <mo>%</mo></mrow> <annotation>$$ 75.2\\%\\pm 1.3\\% $$</annotation></semantics> </math> and <math> <semantics><mrow><mn>79.2</mn> <mo>%</mo> <mo>±</mo> <mn>1.7</mn> <mo>%</mo></mrow> <annotation>$$ 79.2\\%\\pm 1.7\\% $$</annotation></semantics> </math> ). However, EfficientNet resulted in images with artifacts and high MSE when used in denoising tasks while IQ-Net optimized networks performed well for both denoising and reconstruction tasks.</p><p><strong>Data conclusion: </strong>Image quality networks can be trained from image ranking and used to optimize DL tasks.</p><p><strong>Level of evidence: </strong>3 TECHNICAL EFFICACY: Stage 1.</p>","PeriodicalId":16140,"journal":{"name":"Journal of Magnetic Resonance Imaging","volume":" ","pages":""},"PeriodicalIF":3.3000,"publicationDate":"2024-12-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Incorporating Radiologist Knowledge Into MRI Quality Metrics for Machine Learning Using Rank-Based Ratings.\",\"authors\":\"Chenwei Tang, Laura B Eisenmenger, Leonardo Rivera-Rivera, Eugene Huo, Jacqueline C Junn, Anthony D Kuner, Thekla H Oechtering, Anthony Peret, Jitka Starekova, Kevin M Johnson\",\"doi\":\"10.1002/jmri.29672\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><strong>Background: </strong>Deep learning (DL) often requires an image quality metric; however, widely used metrics are not designed for medical images.</p><p><strong>Purpose: </strong>To develop an image quality metric that is specific to MRI using radiologists image rankings and DL models.</p><p><strong>Study type: </strong>Retrospective.</p><p><strong>Population: </strong>A total of 19,344 rankings on 2916 unique image pairs from the NYU fastMRI Initiative neuro database was used for the neural network-based image quality metrics training with an 80%/20% training/validation split and fivefold cross-validation.</p><p><strong>Field strength/sequence: </strong>1.5 T and 3 T T1, T1 postcontrast, T2, and FLuid Attenuated Inversion Recovery (FLAIR).</p><p><strong>Assessment: </strong>Synthetically corrupted image pairs were ranked by radiologists (N = 7), with a subset also scoring images using a Likert scale (N = 2). DL models were trained to match rankings using two architectures (EfficientNet and IQ-Net) with and without reference image subtraction and compared to ranking based on mean squared error (MSE) and structural similarity (SSIM). Image quality assessing DL models were evaluated as alternatives to MSE and SSIM as optimization targets for DL denoising and reconstruction.</p><p><strong>Statistical tests: </strong>Radiologists' agreement was assessed by a percentage metric and quadratic weighted Cohen's kappa. Ranking accuracies were compared using repeated measurements analysis of variance. Reconstruction models trained with IQ-Net score, MSE and SSIM were compared by paired t test. P < 0.05 was considered significant.</p><p><strong>Results: </strong>Compared to direct Likert scoring, ranking produced a higher level of agreement between radiologists (70.4% vs. 25%). Image ranking was subjective with a high level of intraobserver agreement ( <math> <semantics><mrow><mn>94.9</mn> <mo>%</mo> <mo>±</mo> <mn>2.4</mn> <mo>%</mo></mrow> <annotation>$$ 94.9\\\\%\\\\pm 2.4\\\\% $$</annotation></semantics> </math> ) and lower interobserver agreement ( <math> <semantics><mrow><mn>61.47</mn> <mo>%</mo> <mo>±</mo> <mn>5.51</mn> <mo>%</mo></mrow> <annotation>$$ 61.47\\\\%\\\\pm 5.51\\\\% $$</annotation></semantics> </math> ). IQ-Net and EfficientNet accurately predicted rankings with a reference image ( <math> <semantics><mrow><mn>75.2</mn> <mo>%</mo> <mo>±</mo> <mn>1.3</mn> <mo>%</mo></mrow> <annotation>$$ 75.2\\\\%\\\\pm 1.3\\\\% $$</annotation></semantics> </math> and <math> <semantics><mrow><mn>79.2</mn> <mo>%</mo> <mo>±</mo> <mn>1.7</mn> <mo>%</mo></mrow> <annotation>$$ 79.2\\\\%\\\\pm 1.7\\\\% $$</annotation></semantics> </math> ). However, EfficientNet resulted in images with artifacts and high MSE when used in denoising tasks while IQ-Net optimized networks performed well for both denoising and reconstruction tasks.</p><p><strong>Data conclusion: </strong>Image quality networks can be trained from image ranking and used to optimize DL tasks.</p><p><strong>Level of evidence: </strong>3 TECHNICAL EFFICACY: Stage 1.</p>\",\"PeriodicalId\":16140,\"journal\":{\"name\":\"Journal of Magnetic Resonance Imaging\",\"volume\":\" \",\"pages\":\"\"},\"PeriodicalIF\":3.3000,\"publicationDate\":\"2024-12-17\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of Magnetic Resonance Imaging\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://doi.org/10.1002/jmri.29672\",\"RegionNum\":2,\"RegionCategory\":\"医学\",\"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":"Journal of Magnetic Resonance Imaging","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1002/jmri.29672","RegionNum":2,"RegionCategory":"医学","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:深度学习(DL)通常需要一个图像质量度量;然而,广泛使用的指标并不是为医学图像设计的。目的:利用放射科医生的图像排名和深度学习模型,开发一种特定于MRI的图像质量度量。研究类型:回顾性。人口:来自NYU fastMRI Initiative神经数据库的2916对独特图像的19344个排名被用于基于神经网络的图像质量指标训练,得分为80%/20% training/validation split and fivefold cross-validation.Field strength/sequence: 1.5 T and 3 T T1, T1 postcontrast, T2, and FLuid Attenuated Inversion Recovery (FLAIR).Assessment: Synthetically corrupted image pairs were ranked by radiologists (N = 7), with a subset also scoring images using a Likert scale (N = 2). DL models were trained to match rankings using two architectures (EfficientNet and IQ-Net) with and without reference image subtraction and compared to ranking based on mean squared error (MSE) and structural similarity (SSIM). Image quality assessing DL models were evaluated as alternatives to MSE and SSIM as optimization targets for DL denoising and reconstruction.Statistical tests: Radiologists' agreement was assessed by a percentage metric and quadratic weighted Cohen's kappa. Ranking accuracies were compared using repeated measurements analysis of variance. Reconstruction models trained with IQ-Net score, MSE and SSIM were compared by paired t test. P Results: Compared to direct Likert scoring, ranking produced a higher level of agreement between radiologists (70.4% vs. 25%). Image ranking was subjective with a high level of intraobserver agreement ( 94.9 % ± 2.4 % $$ 94.9\%\pm 2.4\% $$ ) and lower interobserver agreement ( 61.47 % ± 5.51 % $$ 61.47\%\pm 5.51\% $$ ). IQ-Net and EfficientNet accurately predicted rankings with a reference image ( 75.2 % ± 1.3 % $$ 75.2\%\pm 1.3\% $$ and 79.2 % ± 1.7 % $$ 79.2\%\pm 1.7\% $$ ). However, EfficientNet resulted in images with artifacts and high MSE when used in denoising tasks while IQ-Net optimized networks performed well for both denoising and reconstruction tasks.Data conclusion: Image quality networks can be trained from image ranking and used to optimize DL tasks.Level of evidence: 3 TECHNICAL EFFICACY: Stage 1.
本文章由计算机程序翻译,如有差异,请以英文原文为准。
查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
Incorporating Radiologist Knowledge Into MRI Quality Metrics for Machine Learning Using Rank-Based Ratings.

Background: Deep learning (DL) often requires an image quality metric; however, widely used metrics are not designed for medical images.

Purpose: To develop an image quality metric that is specific to MRI using radiologists image rankings and DL models.

Study type: Retrospective.

Population: A total of 19,344 rankings on 2916 unique image pairs from the NYU fastMRI Initiative neuro database was used for the neural network-based image quality metrics training with an 80%/20% training/validation split and fivefold cross-validation.

Field strength/sequence: 1.5 T and 3 T T1, T1 postcontrast, T2, and FLuid Attenuated Inversion Recovery (FLAIR).

Assessment: Synthetically corrupted image pairs were ranked by radiologists (N = 7), with a subset also scoring images using a Likert scale (N = 2). DL models were trained to match rankings using two architectures (EfficientNet and IQ-Net) with and without reference image subtraction and compared to ranking based on mean squared error (MSE) and structural similarity (SSIM). Image quality assessing DL models were evaluated as alternatives to MSE and SSIM as optimization targets for DL denoising and reconstruction.

Statistical tests: Radiologists' agreement was assessed by a percentage metric and quadratic weighted Cohen's kappa. Ranking accuracies were compared using repeated measurements analysis of variance. Reconstruction models trained with IQ-Net score, MSE and SSIM were compared by paired t test. P < 0.05 was considered significant.

Results: Compared to direct Likert scoring, ranking produced a higher level of agreement between radiologists (70.4% vs. 25%). Image ranking was subjective with a high level of intraobserver agreement ( 94.9 % ± 2.4 % $$ 94.9\%\pm 2.4\% $$ ) and lower interobserver agreement ( 61.47 % ± 5.51 % $$ 61.47\%\pm 5.51\% $$ ). IQ-Net and EfficientNet accurately predicted rankings with a reference image ( 75.2 % ± 1.3 % $$ 75.2\%\pm 1.3\% $$ and 79.2 % ± 1.7 % $$ 79.2\%\pm 1.7\% $$ ). However, EfficientNet resulted in images with artifacts and high MSE when used in denoising tasks while IQ-Net optimized networks performed well for both denoising and reconstruction tasks.

Data conclusion: Image quality networks can be trained from image ranking and used to optimize DL tasks.

Level of evidence: 3 TECHNICAL EFFICACY: Stage 1.

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
CiteScore
9.70
自引率
6.80%
发文量
494
审稿时长
2 months
期刊介绍: The Journal of Magnetic Resonance Imaging (JMRI) is an international journal devoted to the timely publication of basic and clinical research, educational and review articles, and other information related to the diagnostic applications of magnetic resonance.
期刊最新文献
Associations of Postencephalitic Epilepsy Using Multi-Contrast Whole Brain MRI: A Large Self-Supervised Vision Foundation Model Strategy. Influence of Multiband Technique on Temporal Diffusion Spectroscopy and Its Diagnostic Value in Breast Tumors. Application of Anti-Motion Ultra-Fast Quantitative MRI in Neurological Disorder Imaging: Insights From Huntington's Disease. Editorial for "Identifying Primary Sites of Spinal Metastases: Expert-Derived Features vs. ResNet50 Model Using Non-Enhanced MRI". Effects of Static and Low-Frequency Magnetic Fields on Gene Expression.
×
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
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 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 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1