Deep learning to estimate impaired glucose metabolism from Magnetic Resonance Imaging of the liver: An opportunistic population screening approach

Abstract

Aim: Diabetes is a global health challenge, and many individuals are undiagnosed and not aware of their increased risk of morbidity/mortality although dedicated tests are available, which indicates the need for novel population-wide screening approaches. Here, we developed a deep learning pipeline for opportunistic screening of impaired glucose metabolism using routine magnetic resonance imaging (MRI) of the liver and tested its prognostic value in a general population setting.

Methods: In this retrospective study a fully automatic deep learning pipeline was developed to quantify liver shape features on routine MR imaging using data from a prospective population study. Subsequently, the association between liver shape features and impaired glucose metabolism was investigated in individuals with prediabetes, type 2 diabetes and healthy controls without prior cardiovascular diseases. K-medoids clustering (3 clusters) with a dissimilarity matrix based on Euclidean distance and ordinal regression was used to assess the association between liver shape features and glycaemic status.

Results: The deep learning pipeline showed a high performance for liver shape analysis with a mean Dice score of 97.0±0.01. Out of 339 included individuals (mean age 56.3±9.1 years; males 58.1%), 79 (23.3%) and 46 (13.6%) were classified as having prediabetes and type 2 diabetes, respectively. Individuals in the high risk cluster using all liver shape features (n = 14) had a 2.4 fold increased risk of impaired glucose metabolism after adjustment for cardiometabolic risk factors (age, sex, BMI, total cholesterol, alcohol consumption, hypertension, smoking and hepatic steatosis; OR 2.44 [95% CI 1.12-5.38]; p = 0.03). Based on individual shape features, the strongest association was found between liver volume and impaired glucose metabolism after adjustment for the same risk factors (OR 1.97 [1.38-2.85]; p<0.001).

Conclusions: Deep learning can estimate impaired glucose metabolism on routine liver MRI independent of cardiometabolic risk factors and hepatic steatosis.

Fig 1. Overview of the study design.

A) The deep learning framework for fully automated extraction of liver shape features was developed in participant of the KORA MRI study. B) The performance of the framework was tested in an independent random subset of 44 participants not seen during any part of model development. In addition, the predictive value of liver shape features for impaired glucose metabolism was investigated in the entire study cohort. C) Representative case of the independent testing dataset with manual (red) and automatically (green) generated liver segmentations. KORA = Cooperative Health Research in the Region of Augsburg; CVD = Cardiovascular Disease; MRI = Magnetic Resonance Imaging.

Fig 2. Participant flowchart of the KORA-MRI study.

Other reason: 1 withdrew consent; 4 missing values of liver shape; 4 missing values of hepatic steatosis; 6 missing covariates.

Fig 3. Pearson’s correlation between expert manual vs. automatic deep learning segmentations.

Pearson’s correlation between expert manual and automatic deep learning segmentations of the liver indicating a high model performance with a correlation coefficient of r = 1.0 (p<0.001) in the independent testing dataset.

Fig 4. Association between hepatic shape feature clusters and impaired glucose metabolism.

Association between hepatic shape features clusters (A = intermediate risk cluster; B = high-risk cluster) and impaired glucose metabolism (prediabetes or diabetes) from ordinal logistic regression with proportional odds assumption. Model 1 = univariate; Model 2 = adjusted for age, sex; Model 3 = adjusted age, sex and hepatic steatosis; Model 4 = adjusted for age, sex, BMI, alcohol consumption, hypertension, smoking (never, ex, current), total cholesterol and hepatic steatosis. Boxes indicate odds ratios; lines 95% confidence intervals. BMI = Body Mass Index.

Fig 5. Association between individual hepatic shape features with impaired glucose metabolism and association between liver volume and impaired glucose metabolism.

A) Association between individual hepatic shape features and impaired glucose metabolism (prediabetes or diabetes) from ordinal logistic regression with proportional odds assumption. Figure depicts univariate odds ratios with 95% confidence intervals for hepatic shape features association with impaired glucose metabolism. B) Association between liver volume and impaired glucose metabolism (prediabetes or diabetes) from ordinal logistic regression with proportional odds assumption. Model 1 = univariate; Model 2 = adjusted for age, sex; Model 3 = adjusted age, sex and hepatic steatosis; Model 4 = adjusted for age, sex, BMI, alcohol consumption, hypertension, smoking (never, ex, current), total cholesterol and hepatic steatosis. Boxes indicate odds ratios; lines 95% confidence intervals. BMI = Body Mass Index.