ABSTRACT
OBJECTIVES: To develop, test, and validate a body composition profiling algorithm for automated segmentation of body compartments in whole-body magnetic resonance imaging (wbMRI) and to investigate the influence of different acquisition parameters on performance and robustness. MATERIALS AND METHODS: A segmentation algorithm for subcutaneous and visceral adipose tissue (SCAT and VAT) and total muscle mass (TMM) was designed using a deep learning U-net architecture convolutional neuronal network. Twenty clinical wbMRI scans were manually segmented and used as training, validation, and test datasets. Segmentation performance was then tested on different data, including different magnetic resonance imaging protocols and scanners with and without use of contrast media. Test-retest reliability on 2 consecutive scans of 16 healthy volunteers each as well as impact of parameters slice thickness, matrix resolution, and different coil settings were investigated. Sorensen-Dice coefficient (DSC) was used to measure the algorithms' performance with manual segmentations as reference standards. Test-retest reliability and parameter effects were investigated comparing respective compartment volumes. Abdominal volumes were compared with published normative values. RESULTS: Algorithm performance measured by DSC was 0.93 (SCAT) to 0.77 (VAT) using the test dataset. Dependent from the respective compartment, similar or slightly reduced performance was seen for other scanners and scan protocols (DSC ranging from 0.69-0.72 for VAT to 0.83-0.91 for SCAT). No significant differences in body composition profiling was seen on repetitive volunteer scans (P = 0.88-1) or after variation of protocol parameters (P = 0.07-1). CONCLUSIONS: Body composition profiling from wbMRI by using a deep learning-based convolutional neuronal network algorithm for automated segmentation of body compartments is generally possible. First results indicate that robust and reproducible segmentations equally accurate to a manual expert may be expected also for a range of different acquisition parameters.
Subject(s)
Deep Learning , Magnetic Resonance Imaging , Algorithms , Body Composition , Humans , Reproducibility of Results , Whole Body ImagingABSTRACT
OBJECTIVE: We compared patients with neovascular age-related macular degeneration (nvAMD), diabetic macular oedema (DMO) and other macular pathologies testing their vision with the hyperacuity home-monitoring app Alleye to patients not performing home-monitoring regarding clinical outcomes and clinical management. DESIGN: Matched-pair analysis. SETTING: Retina Referral Centre, Switzerland. PARTICIPANTS: For each eye using Alleye, we matched 2-4 controls not using home-monitoring based on age, gender, number of previous intravitreal injections (IVI), best corrected visual acuity (BCVA) (Early Treatment Diabetic Retinopathy Study letters), central macular thickness (CRT) and time point of enrolment, using the Mahalanobis distance matching algorithm. We included 514 eyes (288 patients); 107 eyes with nvAMD using home monitoring and 218 controls not using home monitoring, 25 eyes with DMO (n=52 controls) and 40 eyes with miscellaneous conditions (n=72 controls). 173 eyes (33.7%) received no IVI during follow-up. MAIN OUTCOME MEASURES: Improvement of ≥5 letters, number of injection visits and treatment retention after correcting for differences in baseline characteristics with multivariate analyses. RESULTS: The mean follow-up duration was 809 days (range 147-1353) and the mean number of IVI/year among treated eyes was 6.7 (SD 3.1). Mean age at baseline was 70.4 years (SD 10.9), BCVA was 77.6 letters (SD 11.6) and CRT was 263.6 µm (SD 86.7) and was similar between patients using and not using home monitoring. In multivariate analyses, patients using home monitoring had a higher chance to improve visual acuity by ≥5 letters (OR 1.67 (95% CI 1.01 to 2.76; p=0.044)) than controls. Treated eyes using home monitoring had less injection visits/year (-0.99 (95% CI -1.59 to -0.40; p=0.001)) and a longer treatment retention +69.2 days (95% CI 2.4 to 136.0; p=0.042). These effects were similar across retinal pathologies. CONCLUSIONS: This data suggest that patients capable of performing mobile hyperacuity home monitoring benefit in terms of visual acuity and discontinue treatment less often than patients not using home monitoring.