Mask Distillation Network for Conjunctival Hyperemia Severity Classification

To achieve automatic,fast and accurate severity classification of bulbar conjunctival hyperemia severity,we proposed a novel prior knowledge-based framework called mask distillation network(MDN).The proposed MDN consists of a segmentation network and a classification network with teacher-student branches.The segmentation network is used to generate a bulbar conjunctival mask and the classification network divides the severity of bulbar conjunctival hyperemia into four grades.In the classification network,we feed the original image and the image with the bulbar conjunctival mask into the student and teacher branches respectively,and an attention consistency loss and a classification consistency loss are used to keep a similar learning mode for these two branches.This design of“different input but same output”,named mask distillation(MD),aims to introduce the regional prior knowledge that“bulbar conjunctival hyperemia severity classification is only related to the bulbar conjunctiva region”.Extensive experiments on 5117 anterior segment images have proven the effectiveness of mask distillation technology:1)The accuracy of the MDN student branch is 3.5%higher than that of a single optimal baseline network and 2%higher than that of the baseline network combination.2)In the test phase,only the student branch is needed,and no additional segmentation network is required.The framework only takes 0.003 s to classify a single image,achieving the fastest speed in all the methods we compared.3)Compared with a single baseline network,the attention of both teacher and student branches in the MDN has been intuitively improved.

Assessment of load-sharing thoracolumbar injury: A modified scoring system

BACKGROUND Many classification systems of thoracolumbar spinal fractures have been proposed to enhance treatment protocols,but none have achieved universal adoption.AIM To develop a new patient scoring system for cases with thoracolumbar injury classification and severity score(TLICS)=4,namely the load-sharing thoracolumbar injury score(LSTLIS).METHODS Based on thoracolumbar injury classification and severity score,this study proposes the use of the established load-sharing classification(LSC)to develop an improved classification system(LSTLIS).To prove the reliability and reproducibility of LSTLIS,a retrospective analysis for patients with thoracolumbar vertebral fractures has been conducted.RESULTS A total of 102 cases were enrolled in the study.The scoring trend of LSTLIS is roughly similar as the LSC scoring,however,the average deviation based on the former method is relatively smaller than that of the latter.Thus,the robustness of the LSTLIS scoring method is better than that of LSC.LSTLIS can further classify patients with TLICS=4,so as to assess more accurately this particular circumstance,and the majority of LSTLIS recommendations are consistent with actual clinical decisions.LSTLIS is a scoring system that combines LSC and TLICS to compensate for the lack of appropriate inclusion of anterior and middle column compression fractures with TLICS.Following preliminary clinical verification,LSTLIS has greater feasibility and reliability value,is more practical in comprehensively assessing certain clinical circumstances,and has better accuracy with clinically significant guidelines.

Impact of hospital type on risk-adjusted,traffic-related 30-day mortality:a population-based registry study

Background:Traffic incidents are still a major contributor to hospital admissions and trauma-related mortality.The aim of this nationwide study was to examine risk-adjusted traffic injury mortality to determine whether hospital type was an independent survival factor.Methods:Data on all patients admitted to Swedish hospitals with traffic-related injuries,based on International Classification of Diseases codes,between 2001 and 2011 were extracted from the Swedish inpatient and cause of death registries.Using the binary outcome measure of death or survival,data were analysed using logistic regression,adjusting for age,sex,comorbidity,severity of injury and hospital type.The severity of injury was established using the International Classification of Diseases Injury Severity Score(ICISS).Results:The final study population consisted of 152,693 hospital admissions.Young individuals(0–25 years of age)were overrepresented,accounting for 41%of traffic-related injuries.Men were overrepresented in all age categories.Fatalities at university hospitals had the lowest mean(SD)ICISS 0.68(0.19).Regional and county hospitals had mean ICISS 0.75(0.15)and 0.77(0.15),respectively,for fatal traffic incidents.The crude overall mortality in the study population was 1193,with a mean ICISS 0.72(0.17).Fatalities at university hospitals had the lowest mean ICISS 0.68(0.19).Regional and county hospitals had mean ICISS 0.75(0.15)and 0.77(0.15),respectively,for fatal traffic incidents.When regional and county hospitals were merged into one group and its risk-adjusted mortality compared with university hospitals,no significant difference was found.A comparison between hospital groups with the most severely injured patients(ICISS0.85)also did not show a significant difference(odds ratio,1.13;95%confidence interval,0.97–1.32).Conclusions:This study shows that,in Sweden,the type of hospital does not influence risk adjusted traffic related mortality,where the most severely injured patients are transported to the university hospitals and centralization of treatment is common.