Font Size: a A A

Morphologic Identification Of Clinically Encountered Moulds Based On An Artificial Intelligence Technology And The Analysis Of Differences In Pathogenicity Patterns And Host Transcriptomics Of Pulmonary Aspergillus Species

Posted on:2024-09-09Degree:DoctorType:Dissertation
Country:ChinaCandidate:R JingFull Text:PDF
GTID:1524306938457244Subject:Medical Technology
Abstract/Summary:PDF Full Text Request
Backgrounds:Invasive pulmonary aspergillosis(IPA)is the most serious deep infectious disease of pulmonary aspergillosis,which can threaten human’s life.With the widespread use of antibiotics,hormones,and immunosuppressive agents,as well as the increase in invasive treatments,the incidence of IPA has gradually increased,often afecting immunocompromised or compromised patients.Aspergillus fumigatus has been reported as the main pathogen of IPA.However,it has been reported that IPA caused by non-Aspergillus fumigatus have been gradually increasing,including Aspergillus flavus,Aspergillus niger,Aspergillus terreus and Aspergillus nidulans,etc.However,the ability to identify moulds in China is still relatively weak.This study aims to develop a rapid detecting method for moulds identification.In addition,the differences in pathogenicity among indicated pulmonary Aspergillus species,as well as the similarities and differences of the host response mechanism,were explored to lay a theoretical foundation for a better understanding of the pathogenicity of IPA infected by the indicated pulmonary Aspergillus species,which is helpful to assist clinical diagnosis and treatment of IPA.Methods:Based on the principle of ResNet-50 deep convolutional neural network,XMVision Fungus AI system was established for identification of clinically encountered moulds by creating training sets,test sets and evaluation sets.Automatic color equalization(ACE)and adaptive contrast enhancement(ACE)are used to normalize the images acquired from different settings,to improve identification ability of XMVision Fungus AI.The clinical strains were screened from Department of Clinical Laboratory,Peking Union Medical College Hospital during 2015-2020,and from northern cooperative hospitals of China Hospital Invasive Fungal Surveillance Net(CHIF-NET)during 2017-2018.In vitro stress responses against pulmonary Aspergillus species were performed,including metal ion pressure,osmotic pressure,oxidative pressure,cell wall/cell membrane damage pressure,high temperature pressure,and antifungal susceptibility testing(Sensititre YeastsOne).The virulence of pulmonary Aspergillus species was initially screened based on C.mellonella infection model.Meanwhile,the representative strains of each pulmonary Aspergillus species were screened by combination of strains information and relevant patient’s cases.By establishing immunosuppressive mice model of invasive pulmonary aspergillosis,the differences and similarities of pathogenicity among pulmonary Aspergillus species were verified.By an analysis of host transcriptomic,the differences and similarities of immune responses and the functional characteristics of pathogenic pathways of each pulmonary aspergillus species were explored.Results:XMVision Fungus AI system successfully achieved a rapid and accurate identification of nine taxa of clinical moulds,including Aspergillus fumigatus complex,Aspergillus flavus complex,Aspergillus niger complex,Aspergillus terreus complex,Aspergillus nidulans,Aspergillus sydowii/Aspergillus versicolor,Syncephalastrum racemosum,Fusarium spp and Penicillium spp.,and the overall accuracy rate reached 93.00%which was higher than that of human readers.By combination of survival curves of C.mellonella and mice,as well as HE&PAS staining and tunel testing on lung tissues,the pathogenicity of each pulmonary Aspergillus species was verified by using the reference strain Af293,with Aspergillus flavus>Aspergillus fumigatus>Aspergillus niger,and Aspergillus tubingensis and Aspergillus terreus were similar in pathogenicity;with the pathogenicity of Aspergillus felis>Aspergillus fumigatus>Aspergillus lentulus among the Aspergillus fumigatus complex.Through specific staining on macrophages and mycelium,it was observed under confocal laser scanning microscope that hypervirulent pulmonary Aspergillus species invaded the host lung mainly in the form of mycelium,and could penetrate the macrophages to ’escape’ the innate immune response.However,hypovirulent pulmonary Aspergillus species mainly invaded the host lung in the form of spores and could be phagocytic by macrophages,possibly ’symbiotic’ with the host during the later stage of infection.Host transcriptome analysis showed that differentially expressed genes(DEGs)increased significantly in the hypervirulent’ group at 24 h after infection,while DEGs increased gradually in the ’hypovirulent’ group at 72 h.The results of DEGs enrichment of ’hypervirulent’ group showed that GO functions were mainly enriched in immune response,developmental function of host epidermal and epithelial cells,and host apoptosis.KEGG enrichment analysis showed that the TNF signaling pathway and IL-17 signaling pathway were significantly different between hypervirulent and hypovirulent species.Real-time fluorescence quantitative PCR(RT-PCR)showed that the expression of BIRC2 gene and IL15 gene was significantly up-regulated in the TNF signaling pathway in the hypovirulent species(P<0.05),while the expression of other related genes compared with hypervirulent species was significantly down-regulated(P<0.05).In addition,the relative expression of LCAM1,CXCL1,LCN2,MAPK3K8 and FOSL1 in Aspergillus terreus were significantly down-regulated compared with that of Af293 reference strain,Aspergillus fumigatus and Aspergillus felis(P<0.05).Conclusions:XMVision Fungus AI with its advantages of convenient operation and lower cost,is suitable for the rapid identification on clinical moulds,and is more suitable for primary hospitals or general hospitals with large amount of specimens,which can also solve the shortage of professional personnel and assist the clinical laboratory to identify common moulds in clinic.Both hypervirulent and hypo virulent species showed different pathogenic characteristics during the infection.The hypervirulent species infected the host could induce excessive inflammatory response,suggesting that relevant clinical treatments could be given to prevent and treat excessive inflammatory response in immunocompromised patients on the basis of antifungal drugs used in the previous treatment of IPA,to protect against immune reconstitution syndrome(IRS).Although the hypovirulent species could only induce a weak inflammatory response to the host,it could not be ignored that the hypovirulent species might ’symbiosis’ with the host in the later stage of infection,suggesting that clinical treatment of IPA infection caused by the hypovirulent species should be aimed at fungal clearance by using antifungals to prevent the occurrence of chronic pulmonary aspergillosis(CPA).In addition,the cIAP1 protein regulated by BIRC2 gene may serve as a targeted agent for IPA patients who are infected by the hypervirulent species.
Keywords/Search Tags:ResNet-50, XMVision Fungus AI, invasive pulmonary aspergillosis, pathogenicity, immune response
PDF Full Text Request
Related items