Objective To propose a deep learning-based method for removing Gaussian and pepper noises of the surgical instrument visual images so as to recover the detailed features of the images.Methods A lightweight multi-task progressive network was constructed involving in a multi-feature fusion encoder-decoder network,an attention-guided network and a detail-recovery progressive network,which used the multi-feature fusion encoder-decoder network to predict and eliminate the noise information in the visual images,the attention-guided network to remove the residual noise and the detail-recovery progressive network to restore the underlying detail features of the denoised images.Some of the regular convolutions in the detail recovery progressive network were replaced with depth separable convolutions to realize lightweight design of the network constructed.Denoising experiments were conducted on the publicly available CBSD68 and Kodak24 datasets and the self-constructed surgical instrument noise dataset so as to compare the denoising effects of the network constructed and the traditional methods and the classification accuracies of ResNet-18 model and ResNet-34 model for the denosied images by the network and to analyze computing power and memory usage before and after the lightweight design.Results The network constructed gained better denoising effect than the classical methods for publicly available datasets,and ResNet-18 model and ResNet-34 model had higher accuracies when used to classify the images denoised by the network for the self-constructed surgical instrument noise dataset.Lightweight design had the parameter number and floating point operations(FLOPs)decreased by approximately 27.27％and 29.81％,respectively.Conclusion The proposed lightweight multi-task progressive network behaves well in denoising surgical instrument visual images with reduced computating power consump-tion and memory usage.[Chinese Medical Equipment Journal,2024,45(2):1-7]

Objective To propose a measurement method based on improved YOLOv8 for the anterior segment parameters of ultrasonic biological microscopy(UBM)images to solve the problems of the ophthalmic ultrasound images in low quantity,high annotation cost and weak model generalization ability.Methods Firstly,data enhancement using StyleGAN3 was carried out to improve YOLOv8 model in decreased sensitivity to UBM images and increased generalization ability;secondly,the pseudo-labels of virtual UBM images were generated based on the YOLOv8 model with the original UBM images as the dataset and the method of pseudo-label self-training in semi-supervised learning;finally,the YOLOv8 model was trained with the expan-ded dataset and improved with the global context network(GCNet)module,and the prediction results of the YOLOv8 model were sorted using the key-point ranking algorithm,and the measurements of the anterior segment physiological parameters were computed after screening qualified images based on prior knowledge.Results Compared with the hand-labeled results by the ophthalmologists,the localization error of the StyleGAN3 data-enhanced and self-trained YOLOv8 model was(61.94±40.66)μm,and the mean relative errors for the measurements of anterior chamber angle distance,pupil diameter,ciliary sulcus distance,central corneal thickness,anterior chamber depth and lens thickness were 0.62％,1.35％,0.68％,4.87％,0.93％and 0.75％,respectively.Conclusion The method proposed enhances the accuracy of the measurement method for anterior segment parameters of UBM images,and can meet real-time requirements.[Chinese Medical Equipment Journal,2024,45(2):8-16]

Objective To develop a damage control operation(DCO)simulation training platform for traumatic brain injury(TBI)in wartime based on mixed reality to open up a new path for surgical skills training of military surgeons.Methods The platform mainly consisted of wartime TBI DCO simulation training software,a surgical manikin and a HoloLens 2 MR device.The simulating training software was developed with C# language and the technologies of MR,basic gestures,spatial scanning positioning and etc on the basis of constructed surgical decision-making training system,virtual surgical environment and functional modules.The surgical manikin was customized with reference to the standard body type of an adult male with a height of 180 cm,and an electronic chip was developed and placed inside the head of the manikin to execute data matching with the simulation training software.The simulation training software was installed and run in the HoloLens 2 MR device to realize TBI DCO simulation training on the virtual reality interactive model.Results The platform developed implemented the functions of virtual reality interactive model reset positioning,operation simulation training,examination and on-site demonstration,which gained advantages in stimulating learning interest and facilitating risk-free,time-and space-indepen-dent,immersive and interactive learning and was generally recognized by the trainees.Conclusion The simulation training platform can be a supplementary to other training means to improve the ability of military surgeons in damage control operation.[Chinese Medical Equipment Journal,2024,45(2):17-21]

Objective To design lower limb motor detection system based on the computer vision and human key point detection neural network model to realize non-contact measurement of the lower limb motor.Methods Firstly,a motion data acquisition platform was built to acquire the lower limb motor images,and to extract the 2D image coordinates of the key points of the lower limb in each frame based on the OpenPose neural network model.Secondly,the spatial coordinates of the key points were determined based on the computer vision algorithm with the position information of the visual sensors in the motion data acquisition platform,so as to analyze the lower limb motor in three-dimensional space.Finally,the data compu-tation of the system in case of different motion speeds was discussed with cycling as an example and the motion data simulta-neously collected by the motion capture system as references.Results The Pearson's correlation coefficients of the data computed by the system and the data obtained by the motion capture system were 0.950,0.917,and 0.828 at cycling speeds of 0.8,1.3,and 2.1 m/s,respectively,which were all within the strong correlation range.Conclusion The system developed indicates the lower limb motor,and thus can be used for the lower limb motion detection.[Chinese Medical Equipment Journal,2024,45(2):22-27]

Objective To explore the data on the optimum ventilation cross section for the ultra-clean bench with finite element simulation software to meet the requirements of experimental personnel in normal conditions.Methods Firstly,a finite element model for the vertical flow ultra-clean bench was constructed with SolidWorks software,then fluid dynamics simulation for the model was carried out with Flow Simulation software to analyze the gas flow rate,ventilation volume and ventilation interference of the cross section in case of different openings of the adjustable baffle(the distance of the lower edge of the baffle from the clean desktop),finally the suitable baffle openning was determined through comprehensive analysis.Results Simulation results showed that the gas flow rate and ventilation interfence decreased with the increase of th baffle openning,and the ventilation volume peaked when the baffle openning was 225 mm and had the average value being 376.002 m3/h.Conclusion The ventilation interference,gas flow rate and ventilation volume prove to have optimum results in case the baffle openning is 225 mm and the ventilation cross-sectional area is 0.248 m2.[Chinese Medical Equipment Journal,2024,45(2):28-34]

Objective To optimize the batch dispensing parameters of dispensing robots to ensure the standardization and homogenization of finished infusion.Methods Five parameters for intelligent dispensing robots were optimized with orthogonal test including push and pull times,standing time,needle rotation speed,solvent volume and pulling angle,with the insoluble particles as the evaluation index.Four finished infusion solutions including pantoprazole sodium solution,esomeprazole sodium solution,omeprazole sodium solution and eplazole sodium solution prepared at room temperature for 0,2,4,6 and 8 h were compared in terms of appearance,pH,osmolality and number of insoluble particles;the amounts of drug residue in empty syringe bottles and the efficiency of batch dispensing of the four finished infusion solutions using dispensing robot and manual operation were observed.Results After optimization the batch dispensing parameters of the robot were set as 2 times(push and pull times),5 s(standing time),low speed(needle rotation speed),6 mL(solvent volume)and 90°(pulling angle).After 8 h of dispensing at room temperature,the eplazole sodium solution became light yellow and clarified,and the number of insoluble particles went beyond the prescribed value in Chinese Pharmacopoeia 2020;the remained three solutions were colorless and clarified at 0,2,4,6 and 8 h,and the number of insoluble particles complied with the requirements of Chinese Pharmacopoeia 2020;there were no significant changes in pH and osmolality at 0,2,4,6 and 8 h.The robot had less drug residue than manual operation,and the efficiency of two robots operated by one person was higher than that of one person manually or one robot by one person,and all the differences were statistically significant(P＜0.05).Conclusion The batch dispensing parameters optimized with orthogonal test contributes to the standardization and homogeneity of the finished solutions with decreased drug residue while increased batch dispensing efficiency.[Chinese Medical Equipment Journal,2024,45(2):35-40]

Objective To compare the properties of hip joint implants with plasma sprayed and sintered porous coating.Methods The biological fixation femoral stems by plasma spraying and sintered porous titanium coating were observed in terms of the microstructure,tensile properties and abrasion resistance of the coating materials by a fully automatic metallur-gical microscope,a material testing machine and an abrasion meter,and the two coating processes were evaluated by observing the stems'stereological structure and the bonding performance between the coatings and the substrates.Results Plasma spraying titanium coating made the stem having refined and uniform microstructure,a tensile strength of(1 065±21)MPa and an elongation of(15±2)％,which had a loose and porous structure,with an average thickness of 521 μm,an average percentage of pore volume of 37％,and an average pore-intercept of 135 μm,a shear strength with the substrate of(39±4)MPa,a tensile strength of(47±5)MPa and a mass loss after 100 wear cycles of 48 mg;sintered porous titanium coating made the stem having coarse grains in its microstructure,a tensile strength of(879±23)MPa and an elongation of(7±2)％,which had gently sloping surface and uniform distribution of pores,with an average thickness of 1 167 μm,an average percentage of pore volume of 41％,and an average pore-intercept of 240 μm,a shear strength with the substrate of(44±5)MPa,a tensile strength of(62±7)MPa and a mass loss after 100 wear cycles of 15 mg.Conclusion The hip implants with the two coating processes both have their own advantages and disadvantages and meet the requirements of relevant industry standards,which can be expected to obtain excellent in vivo performance.[Chinese Medical Equipment Journal,2024,45(2):41-45]

Objective To design a rehabilitation outpatient system for military convalescent institutions to solve the problems due to data confusion.Methods A rehabilitation outpatient system for military convalescent institutions was developed with Client/Server(C/S)architecture,front-end and back-end separation mode,C# language,PowerBuilder 9.0 development tool and.NET Framework 4.0 architecture,which was composed of five functional modules for outpatient fee management,serviceman medical management,medicine & medical material management,outpatient doctor management and system data maintenance.Results The system developed was highly compatible with the underlying data of the military convalescent information management system,and enhanced effectively the data connectivity between the key points for fee management,medical management and pharmacy management and etc.Conclusion The system developed contributes to data interaction,and enables military convalescent institutions to run the military convalescent information management system and rehabiliation outpatient system simultaneously.[Chinese Medical Equipment Journal,2024,45(2):46-51]

Objective To investigate the dosimetric effects of different calculation grid size(CGS)in helical tomotherapy(HT)planning system on stereotactic body radiation therapy(SBRT)for non-small cell lung cancer(NSCLC).Methods Nine NSCLC patients receiving radiation therapy for the first time at some hospital from March 2019 to December 2022 were selected as the subjects.SBRT planning was carried out through the HT system with three different CGS plans(Fine,Normal,and Coarse)and the same pitch,modulation factor(MF)and optimization conditions,and the target area indexes of the three CGS plans were compared including conformity index(CI),homogeneity index(HI),dosimetric parameters of the organ at risk(OAR),point dose verification pass rate,treatment time,number of monitor units and Sinograms.SPSS 22.0 was used for statistical analysis.Results For target area HI,there weres significant differences between CGS Fine plan and Coarse plan and between CGS Normal plan and Coarse plan(P＜0.05),while no statistical differences were found between CGS Fine plan and Normal plan(P＞0.05).For target area CI,there were significant differences between CGS Fine plan and Coarse plan(P＜0.05),while no statistical differences were found between CGS Fine plan and Normal plan and between CGS Normal plan and Coarse plan(P＞0.05).For OAR dosimetric parameters,CGS Fine plan and Coarse plan had significant differences in heart Dmax and Dmean,esophageal Dmax and Dmean,V5,V20,V30 and Dmean of the whole lung and affected lung,V5 and Dmax of the affected lung and heart V10 and V30(P＜0.05),CGS Normal plan and Coarse plan had obvious differences in esophageal Dmax(P＜0.05),and the remained dosimetric parameters were not statistically significant(P＞0.05).Fine,Normal and Coarse plans had the point dose verifica-tion pass rates being 0.96％,1.50％and 1.77％,respectively.In terms of treatment time and number of monitor units,there were significant differences between Fine plan and Coarse plan(P＜0.05)while no statistical differences were found between Fine and Normal plans and between Normal and Coarse plans(P＞0.05).Sinograms analyses showed Fine plan had evenly distributed segment color gradient,Coarse plan had areas of very dark and very light color gradients and Normal plan was somewhere in between.Conclusion Low CGS has to be used as much as possible to obtain accurate dose distribution during SBRT planning for NSCLC patients,which contributes to the execution of the radiation therapy plan and the prevention of ad-verse effects.[Chinese Medical Equipment Journal,2024,45(2):52-57]

Objective To verify and evaluate the performance of Mindray CX-9000 automatic coagulation analyzer.Methods Myriad CX-9000 automatic coagulation analyzer had its performance validated and evaluated by five routine coagulation tests with reference to the industry including standard prothrombin time(PT),activated partial thromboplastin time(APTT),fibrino-gen(FIB),thrombin time(TT)and D-dimer,so as to clarity whether the analyzer meets the clinical requirements in terms of precision,accuracy,linear range,carry-over and reference interval.Results The five routine coagulation tests had the intra-batch precision CVs lower than 3％,the inter-day precision CVs lower than 5％and the relative deviations of the accuracies within±10％.The slope k of the linear regression equation was 1.006 0 for FIB and 1.013 2 for D-dimer,and the correlation coefficient r was 0.999 3 for FIB and 0.997 5 for D-dimer,which showed high linearity.The carry-over rate of sample concen-tration(CR1)was 4.6％,and the carry-over rate between test items(CR2)was lower than 2％;the R value of the reference interval was 100％;the requirements of industrial standards were met.Conclusion Mindray CX-9000 automatic coagulation analyzer gains advantages in precision,accuracy,linear range and carry-over,and is worthy promoting clinically.[Chinese Medical Equipment Journal,2024,45(2):58-61]