Detection of diabetes from whole-body MRI using deep learning
Obesity is one of the main drivers of type 2 diabetes, but it is not uniformly associated with the disease. The location of fat accumulation is critical for metabolic health. Specific patterns of body fat distribution, such as visceral fat, are closely related to insulin resistance. There might be f...
| Main Authors: | , , , , , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
American Society for Clinical investigation
2021-11-01
|
| Series: | JCI Insight |
| Subjects: | |
| Online Access: | https://doi.org/10.1172/jci.insight.146999 |
| _version_ | 1811334800962748416 |
|---|---|
| author | Benedikt Dietz Jürgen Machann Vaibhav Agrawal Martin Heni Patrick Schwab Julia Dienes Steffen Reichert Andreas L. Birkenfeld Hans-Ulrich Häring Fritz Schick Norbert Stefan Andreas Fritsche Hubert Preissl Bernhard Schölkopf Stefan Bauer Robert Wagner |
| author_facet | Benedikt Dietz Jürgen Machann Vaibhav Agrawal Martin Heni Patrick Schwab Julia Dienes Steffen Reichert Andreas L. Birkenfeld Hans-Ulrich Häring Fritz Schick Norbert Stefan Andreas Fritsche Hubert Preissl Bernhard Schölkopf Stefan Bauer Robert Wagner |
| author_sort | Benedikt Dietz |
| collection | DOAJ |
| description | Obesity is one of the main drivers of type 2 diabetes, but it is not uniformly associated with the disease. The location of fat accumulation is critical for metabolic health. Specific patterns of body fat distribution, such as visceral fat, are closely related to insulin resistance. There might be further, hitherto unknown, features of body fat distribution that could additionally contribute to the disease. We used machine learning with dense convolutional neural networks to detect diabetes-related variables from 2371 T1-weighted whole-body MRI data sets. MRI was performed in participants undergoing metabolic screening with oral glucose tolerance tests. Models were trained for sex, age, BMI, insulin sensitivity, HbA1c, and prediabetes or incident diabetes. The results were compared with those of conventional models. The area under the receiver operating characteristic curve was 87% for the type 2 diabetes discrimination and 68% for prediabetes, both superior to conventional models. Mean absolute regression errors were comparable to those of conventional models. Heatmaps showed that lower visceral abdominal regions were critical in diabetes classification. Subphenotyping revealed a group with high future diabetes and microalbuminuria risk.Our results show that diabetes is detectable from whole-body MRI without additional data. Our technique of heatmap visualization identifies plausible anatomical regions and highlights the leading role of fat accumulation in the lower abdomen in diabetes pathogenesis. |
| first_indexed | 2024-04-13T17:14:26Z |
| format | Article |
| id | doaj.art-31e0f5d9a2fd48a4930b2d7b97746b0e |
| institution | Directory Open Access Journal |
| issn | 2379-3708 |
| language | English |
| last_indexed | 2024-04-13T17:14:26Z |
| publishDate | 2021-11-01 |
| publisher | American Society for Clinical investigation |
| record_format | Article |
| series | JCI Insight |
| spelling | doaj.art-31e0f5d9a2fd48a4930b2d7b97746b0e2022-12-22T02:38:10ZengAmerican Society for Clinical investigationJCI Insight2379-37082021-11-01621Detection of diabetes from whole-body MRI using deep learningBenedikt DietzJürgen MachannVaibhav AgrawalMartin HeniPatrick SchwabJulia DienesSteffen ReichertAndreas L. BirkenfeldHans-Ulrich HäringFritz SchickNorbert StefanAndreas FritscheHubert PreisslBernhard SchölkopfStefan BauerRobert WagnerObesity is one of the main drivers of type 2 diabetes, but it is not uniformly associated with the disease. The location of fat accumulation is critical for metabolic health. Specific patterns of body fat distribution, such as visceral fat, are closely related to insulin resistance. There might be further, hitherto unknown, features of body fat distribution that could additionally contribute to the disease. We used machine learning with dense convolutional neural networks to detect diabetes-related variables from 2371 T1-weighted whole-body MRI data sets. MRI was performed in participants undergoing metabolic screening with oral glucose tolerance tests. Models were trained for sex, age, BMI, insulin sensitivity, HbA1c, and prediabetes or incident diabetes. The results were compared with those of conventional models. The area under the receiver operating characteristic curve was 87% for the type 2 diabetes discrimination and 68% for prediabetes, both superior to conventional models. Mean absolute regression errors were comparable to those of conventional models. Heatmaps showed that lower visceral abdominal regions were critical in diabetes classification. Subphenotyping revealed a group with high future diabetes and microalbuminuria risk.Our results show that diabetes is detectable from whole-body MRI without additional data. Our technique of heatmap visualization identifies plausible anatomical regions and highlights the leading role of fat accumulation in the lower abdomen in diabetes pathogenesis.https://doi.org/10.1172/jci.insight.146999EndocrinologyMetabolism |
| spellingShingle | Benedikt Dietz Jürgen Machann Vaibhav Agrawal Martin Heni Patrick Schwab Julia Dienes Steffen Reichert Andreas L. Birkenfeld Hans-Ulrich Häring Fritz Schick Norbert Stefan Andreas Fritsche Hubert Preissl Bernhard Schölkopf Stefan Bauer Robert Wagner Detection of diabetes from whole-body MRI using deep learning JCI Insight Endocrinology Metabolism |
| title | Detection of diabetes from whole-body MRI using deep learning |
| title_full | Detection of diabetes from whole-body MRI using deep learning |
| title_fullStr | Detection of diabetes from whole-body MRI using deep learning |
| title_full_unstemmed | Detection of diabetes from whole-body MRI using deep learning |
| title_short | Detection of diabetes from whole-body MRI using deep learning |
| title_sort | detection of diabetes from whole body mri using deep learning |
| topic | Endocrinology Metabolism |
| url | https://doi.org/10.1172/jci.insight.146999 |
| work_keys_str_mv | AT benediktdietz detectionofdiabetesfromwholebodymriusingdeeplearning AT jurgenmachann detectionofdiabetesfromwholebodymriusingdeeplearning AT vaibhavagrawal detectionofdiabetesfromwholebodymriusingdeeplearning AT martinheni detectionofdiabetesfromwholebodymriusingdeeplearning AT patrickschwab detectionofdiabetesfromwholebodymriusingdeeplearning AT juliadienes detectionofdiabetesfromwholebodymriusingdeeplearning AT steffenreichert detectionofdiabetesfromwholebodymriusingdeeplearning AT andreaslbirkenfeld detectionofdiabetesfromwholebodymriusingdeeplearning AT hansulrichharing detectionofdiabetesfromwholebodymriusingdeeplearning AT fritzschick detectionofdiabetesfromwholebodymriusingdeeplearning AT norbertstefan detectionofdiabetesfromwholebodymriusingdeeplearning AT andreasfritsche detectionofdiabetesfromwholebodymriusingdeeplearning AT hubertpreissl detectionofdiabetesfromwholebodymriusingdeeplearning AT bernhardscholkopf detectionofdiabetesfromwholebodymriusingdeeplearning AT stefanbauer detectionofdiabetesfromwholebodymriusingdeeplearning AT robertwagner detectionofdiabetesfromwholebodymriusingdeeplearning |