Study On The Technology Related To The Diagnosis System Of Interstitial Pneumonia After Lung Cancer Treatment Based On Multi-source Data

In recent years,as the incidence of lung cancer has gradually increased and the level of treatment for lung cancer has improved,adverse reactions following lung cancer treatment have become an increasing focus of research.Interstitial pneumonia after lung cancer treatment is a common adverse reaction that usually occurs after radiation therapy,after immunotherapy and after treatment with infection.If left unrecognized and untreated,the disease can deteriorate rapidly and even become lifethreatening.Treatment options for interstitial pneumonia vary from cause to cause.Traditional diagnostic methods rely on the judgement of CT(Computed Tomography)images by imaging physicians to achieve this,which is subjective and has a misdiagnosis rate that does not meet clinical needs.Therefore,it is crucial to establish a multi-source data-based diagnosis system for interstitial pneumonia after lung cancer treatment to improve the diagnostic accuracy and efficiency of interstitial pneumonia after lung cancer treatment.Based on the characteristics of CT images of lung,this paper revolves around increasing the global modelling capability of the model,the capability of local generalization bias and incorporating multimodal modelling of clinical metadata,and designs two diagnostic models for identifying different causes of interstitial pneumonia,with the main work as follows:(1)proposed a novel network of LS-MLP(Multilayer Perceptron for local and added sliding windows)based on MLP(Multilayer Perceptron)to increase the model’s ability to locally model lung CT images and to accurately capture the LS-MLP extracts global features through the MLP network and employs a sliding window to increase the exchange of information between different sequences.To demonstrate the effectiveness and potential of the network,the LS-MLP network was used to identify post-treatment interstitial pneumonia in 166 patients with a diagnostic accuracy of 0.889,outperforming existing image classification models.The experimental results show that the LS-MLP network has the ability to discriminate between interstitial pneumonia caused by radiation therapy,immunotherapy and infection in lung CT images,and can effectively distinguish between different causes of interstitial pneumonia.(2)Based on the clinical diagnosis process of physicians,this paper proposes a multimodal joint diagnosis method based on CT image data combined with clinical metadata to capture clinical information of patients with interstitial pneumonia to enrich features.A self-attentive mechanism is added to the multi-source fusion model to learn the weights of the feature vectors.The results of this study show that the multi-source fusion model has a higher diagnostic accuracy and reliability than LS-MLP,a diagnostic system that only identifies lung CT images of interstitial pneumonia caused by radiation therapy,immunotherapy,and infection.The results of this study show that the diagnosis of post-treatment interstitial pneumonia in lung cancer is highly accurate and reliable and can effectively address the limitations of traditional diagnostic methods.The results of this study are important for improving the early diagnosis of post-treatment interstitial pneumonia in lung cancer,as well as providing ideas and methods for medical research and treatment of lung cancer.