A Machine Learning Algorithm to Estimate Sarcopenia on Abdominal CT

Burns, Joseph E.; Yao, Jianhua; Chalhoub, Didier; Chen, Joseph J.; Summers, Ronald M.

doi:10.1016/j.acra.2019.03.011

Cited by 108 publications

(87 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…There have been several studies reporting the accuracy of automatic quantification of abdominal muscle area using deep learning (18)(19)(20)(21)(22). The algorithms used were different across studies, such as U-net neural network model, multi-atlas segmentation model, fully convolutional network (FCN), and augmented active shape model.…”

Section: Introductionmentioning

confidence: 99%

Development and Validation of a Deep Learning System for Segmentation of Abdominal Muscle and Fat on Computed Tomography

et al. 2020

View full text Add to dashboard Cite

Objective: We aimed to develop and validate a deep learning system for fully automated segmentation of abdominal muscle and fat areas on computed tomography (CT) images. Materials and Methods: A fully convolutional network-based segmentation system was developed using a training dataset of 883 CT scans from 467 subjects. Axial CT images obtained at the inferior endplate level of the 3rd lumbar vertebra were used for the analysis. Manually drawn segmentation maps of the skeletal muscle, visceral fat, and subcutaneous fat were created to serve as ground truth data. The performance of the fully convolutional network-based segmentation system was evaluated using the Dice similarity coefficient and cross-sectional area error, for both a separate internal validation dataset (426 CT scans from 308 subjects) and an external validation dataset (171 CT scans from 171 subjects from two outside hospitals). Results: The mean Dice similarity coefficients for muscle, subcutaneous fat, and visceral fat were high for both the internal (0.96, 0.97, and 0.97, respectively) and external (0.97, 0.97, and 0.97, respectively) validation datasets, while the mean cross-sectional area errors for muscle, subcutaneous fat, and visceral fat were low for both internal (2.1%, 3.8%, and 1.8%, respectively) and external (2.7%, 4.6%, and 2.3%, respectively) validation datasets. Conclusion: The fully convolutional network-based segmentation system exhibited high performance and accuracy in the automatic segmentation of abdominal muscle and fat on CT images.

show abstract

Section: Introductionmentioning

confidence: 99%

Development and Validation of a Deep Learning System for Segmentation of Abdominal Muscle and Fat on Computed Tomography

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Automation of such techniques would make them clinically viable and could further promote the use of CT as the onestop-shop imaging prior to shoulder replacement surgery. In recent years, deep learning has emerged as a very effective classification technique, which has been applied with great success to medical image segmentation, including muscle segmentation in CT datasets [16][17][18][19], and detection of large rotator cuff tears from conventional shoulder radiographs [20]. However, to the best of our knowledge, this technique has yet to be evaluated for the prediction of the premorbid muscle boundaries, which are not distinctly and readily identifiable in the images.…”

Section: Introductionmentioning

confidence: 99%

Deep learning for the rapid automatic quantification and characterization of rotator cuff muscle degeneration from shoulder CT datasets

et al. 2020

View full text Add to dashboard Cite

Objectives This study aimed at developing a convolutional neural network (CNN) able to automatically quantify and characterize the level of degeneration of rotator cuff (RC) muscles from shoulder CT images including muscle atrophy and fatty infiltration. Methods One hundred three shoulder CT scans from 95 patients with primary glenohumeral osteoarthritis undergoing anatomical total shoulder arthroplasty were retrospectively retrieved. Three independent radiologists manually segmented the premorbid boundaries of all four RC muscles on standardized sagittal-oblique CT sections. This premorbid muscle segmentation was further automatically predicted using a CNN. Automatically predicted premorbid segmentations were then used to quantify the ratio of muscle atrophy, fatty infiltration, secondary bone formation, and overall muscle degeneration. These muscle parameters were compared with measures obtained manually by human raters. Results Average Dice similarity coefficients for muscle segmentations obtained automatically with the CNN (88% ± 9%) and manually by human raters (89% ± 6%) were comparable. No significant differences were observed for the subscapularis, supraspinatus, and teres minor muscles (p > 0.120), whereas Dice coefficients of the automatic segmentation were significantly higher for the infraspinatus (p < 0.012). The automatic approach was able to provide good-very good estimates of muscle atrophy (R 2 = 0.87), fatty infiltration (R 2 = 0.91), and overall muscle degeneration (R 2 = 0.91). However, CNN-derived segmentations showed a higher variability in quantifying secondary bone formation (R 2 = 0.61) than human raters (R 2 = 0.87). Conclusions Deep learning provides a rapid and reliable automatic quantification of RC muscle atrophy, fatty infiltration, and overall muscle degeneration directly from preoperative shoulder CT scans of osteoarthritic patients, with an accuracy comparable with that of human raters. Key Points • Deep learning can not only segment RC muscles currently available in CT images but also learn their pre-existing locations and shapes from invariant anatomical structures visible on CT sections. • Our automatic method is able to provide a rapid and reliable quantification of RC muscle atrophy and fatty infiltration from conventional shoulder CT scans. • The accuracy of our automatic quantitative technique is comparable with that of human raters.

show abstract

“…There is research that studies the CT scan using automatic learning techniques, which evaluate the muscle volume of adults with sarcopenia, and the results are between 0.80 and 0.87 of precision [20][21][22][23][24]. A study on sarcopenia with similar technical characteristics to our study [20] used four classifiers, Random Forest, SVM, Gradient Boosting, and Logistic regression, with Random Forest as the best classifier, obtained an accuracy of 0.82. The article for the measurement of variables involved in the development of sarcopenia, using forecasting networks based on automatic learning approaches, where the results show an accuracy of 82%, analyzed 114 variables in this study [25].…”

Section: Discussionmentioning

confidence: 73%

Evaluation of Prevalence of the Sarcopenia Level Using Machine Learning Techniques: Case Study in Tijuana Baja California, Mexico

Cristian

Zúñiga

2020

IJERPH

View full text Add to dashboard Cite

The article presents a study based on timeline data analysis of the level of sarcopenia in older patients in Baja California, Mexico. Information was examined at the beginning of the study (first event), three months later (second event), and six months later (third event). Sarcopenia is defined as the loss of muscle mass quality and strength. The study was conducted with 166 patients. A total of 65% were women and 35% were men. The mean age of the enrolled patients was 77.24 years. The research included 99 variables that consider medical history, pharmacology, psychological tests, comorbidity (Charlson), functional capacity (Barthel and Lawton), undernourishment (mini nutritional assessment (MNA) validated test), as well as biochemical and socio-demographic data. Our aim was to evaluate the prevalence of the level of sarcopenia in a population of chronically ill patients assessed at the Tijuana General Hospital. We used machine learning techniques to assess and identify the determining variables to focus on the patients’ evolution. The following classifiers were used: Support Vector Machines, Linear Support Vector Machines, Radial Basis Function, Gaussian process, Decision Tree, Random Forest, multilayer perceptron, AdaBoost, Gaussian Naive Bayes, and Quadratic Discriminant Analysis. In order of importance, we found that the following variables determine the level of sarcopenia: Age, Systolic arterial hypertension, mini nutritional assessment (MNA), Number of chronic diseases, and Sodium. They are therefore considered relevant in the decision-making process of choosing treatment or prevention. Analysis of the relationship between the presence of the variables and the classifiers used to measure sarcopenia revealed that the Decision Tree classifier, with the Age, Systolic arterial hypertension, MNA, Number of chronic diseases, and Sodium variables, showed a precision of 0.864, accuracy of 0.831, and an F1 score of 0.900 in the first and second events. Precision of 0.867, accuracy of 0.825, and an F1 score of 0.867 were obtained in event three with the same variables. We can therefore conclude that the Decision Tree classifier yields the best results for the assessment of the determining variables and suggests that the study population’s sarcopenia did not change from moderate to severe.

show abstract

A Machine Learning Algorithm to Estimate Sarcopenia on Abdominal CT

Cited by 108 publications

References 29 publications

Development and Validation of a Deep Learning System for Segmentation of Abdominal Muscle and Fat on Computed Tomography

Development and Validation of a Deep Learning System for Segmentation of Abdominal Muscle and Fat on Computed Tomography

Deep learning for the rapid automatic quantification and characterization of rotator cuff muscle degeneration from shoulder CT datasets

Evaluation of Prevalence of the Sarcopenia Level Using Machine Learning Techniques: Case Study in Tijuana Baja California, Mexico

Contact Info

Product

Resources

About