Quantitative Evaluation And Treatment Of Masseter Muscle Involvement In Hemifacial Microsomia

ObjectiveMasseter hypoplasia is common in hemifacial microsomia(HFM)while studies of masseter functional involvement are limited.We aimed to quantitatively describe both the morphology and functional involvement of the masseter muscles based on a database of HFM patients.We proposed a coarse-to-fine learning framework based on a U-shaped network to perform automatic segmentation of the masseter muscle and evaluate the learning-based convolutional neural networks algorithm and verify the efficiency and accuracy of the U-net neural convolutional network algorithm clinically.For the comprehensive treatment of facial asymmetry in HFM,we explored the feasibility of neuromuscular electrical stimulation therapy(MMES)therapy.Method1.Ninety-eight patients with HFM underwent 3DCT and surface electromyography from 2012 to 2019 diagnosed in the ninth people’s hospital were included.The mean action potential during maximum voluntary contraction in the intercuspal position was recorded.The Asymmetry of Compound Muscle Action Potential(ACMAP)was calculated as an indicator of functional involvement.We assessed the affected-unaffected differences,morphology-function correlation and the relationship between ACMAP and OMENS+classification using paired T-test,Pearson correlation analysis and Spearman’s correlation analysis.2.To validate the feasibility and accuracy of the approach and assist clinical applications,Dice’s similarity coefficient(DSC),the volume of masseter muscle,average time were used to evaluate the efficiency and robustness compared with manual segmentation in test set of 20 HFM patients randomly selected for non-inferiority test(p<15%).3.The sample size of the prospective,randomized controlled study was calculated with PASS software.An extra NMES intervention in 51 patients called group A and 51 Patients accepted routine masseter muscle care in group B The change of ACMAP was taken as the main reference index with masseter volume as secondary parameter.Paired t-test was used in all parameters.Result1.The masseter muscle was absent on the affected side in 11 patients.In the remaining 87 patients,the masseter had a moderate correlation with Kaban-Pruzansky classification both morphologically and functionally.No significant correlation was found between ACMAP and the soft tissue gradings in OMENS.A masseter function classification was proposed as Type Ⅰ(ACMAP<0.2),Type Ⅱ(0.2<ACMAP<0.35),Type Ⅲ(0.35≤ACMAP<0.55),and Type Ⅳ(ACMAP≥0.55).2.For the CT images,we achieved for segmentation a mean DSC of 0.794±0.028.When comparing the bilateral differences,there was no significant difference of masseter volume.The average segmentation time of the algorithm is 6.4s,much less than the manual time(20min).3.102 patients with mean age 5.5 years were enrolled.The ACMAP and AMV were significantly reduced in six-months following up in group A.Significant difference was shown in ACMAP((p=0.046<0.05)between Group A(0.09±0.05)and Group B(0.07±0.05).There’s no significant difference with 12-month in Group A.No complications were found in all patients.Conclusion1.The masseter muscle of the affected side were involved both morphology and function.CT and EMG were useful to evaluate masseter involvement in HFM.Masseter function classification may be a beneficial tool in patients with HFM.2.The accuracy of the Algorithm is not inferior to manual segmentation with more efficiency.3.These results demonstrated positive effect of NMES compared with traditional masseter exercise treatment.The results were limited by the number of samples and further experiment was needed.