目的 探讨大矩阵联合全模型迭代重建（Iterative Model Reconstruction，IMR）在人工智能（Artificial Intelligence， AI）检出肺结节中的应用价值，并比较滤波反投影（Filter Back Projection，FBP）、1024×1024 iDose4、1024×1024 IMR 重建算法对图像质量的影响。方法 将临床怀疑肺结节要求行胸部薄层CT的60名患者纳入本次研究，所有患者扫描结束 后对原始数据进行常规FBP、1024×1024 iDose4、1024×1024 IMR算法重建，比较三组重建图像噪声标准差（Standard Deviation，SD）、信噪比（Signal to Noise Ratio，SNR）、对比噪声比（Contrast to Noise Ratio，CNR），比较三组图像在 AI系统下对肺结节的检出率、真阳性数以及假阳性率的影响。结果 FBP、1024×1024 iDose4、1024×1024 IMR降低噪声能 力依次提高，三组图像SNR、CNR按照此顺序依次提高，随着图像SNR、CNR的提高，图像的空间分辨率也随之提升，AI 识别肺结节能力得到提升，AI识别肺结节敏感度IMR组相较于常规FBP组提高16.9%，同时假阳性率降低17.8%。结论 大矩 阵联合IMR技术能提高图像空间分辨率，在AI识别肺结节过程中可以提高对磨玻璃结节的检出能力，值得临床推广。

Objective To explore the application value of large matrix combined with full iterative model reconstruction (IMR) in artificial intelligence (AI) detection of lung nodules, and compare the effects of filter back projection (FBP), 1024×1024 iDose4, and 1024×1024 IMR reconstruction algorithms on image quality. Methods A total of 60 patients with suspected pulmonary nodules who required chest thin-layer CT were included in this study. After scanning, the original data of all patients were analyzed by routine FBP, 1024×1024 iDose4, 11024×1024 IMR algorithm reconstruction. Image noise standard deviation (SD), signal to noise ratio (SNR) and contrast to noise ratio (CNR) of three groups of reconstructed images were compared. The effects of the three groups of images on the detection rate, true positive number and false positive rate of pulmonary nodules under the artificial intelligence diagnosis system (CAD) were compared. Results The noise reduction ability of FBP, 1024×1024 iDose4 and 1024×1024 IMR was improved in turn. The SNR and CNR of the three groups of images were improved in this order. With the improvement of SNR and CNR, the spatial resolution of the images was also improved, and the ability of AI to identify pulmonary nodules was improved. Compared with the conventional FBP group, the sensitivity of AI to identify lung nodules in the IMR group was increased by 16.9%, and the false positive rate was reduced by 17.8%. Conclusion Large matrix combined with IMR technology can improve the spatial resolution of image and improve the detection ability of ground glass nodules in the process of CAD recognition of pulmonary nodules, which is worthy of clinical promotion.