计算机科学 ›› 2025, Vol. 52 ›› Issue (6A): 240700010-8.doi: 10.11896/jsjkx.240700010
关昕1, 杨雪永1, 杨啸林2, 孟祥福1
GUAN Xin1, YANG Xueyong1, YANG Xiaolin2, MENG Xiangfu1
摘要： 肿瘤突变负荷与非小细胞肺癌的免疫治疗疗效呈正相关,在临床实践中一般通过全外显子组测序来测量肿瘤突变负荷。然而,全外显子组测序操作复杂耗时、价格昂贵,导致大多数医院无法使用。基于此,提出了一种成本低、周期短、准确率高的基于病理组织切片预测肺腺癌肿瘤突变负荷的深度学习模型DBFormer。首先,颜色反卷积结构将输入模型的数字病理图像的RGB和HED图像信息相结合,丰富输入的病理图像信息,使模型更加适合医学任务分类;其次,图像通过四层金字塔结构,每层都包括一个最大池化层和一个DBFormer块,最大池化层减小图像尺寸、提升特征矩阵维度,DBFormer块包含归一化层和双重路由注意力机制对图像进行特征提取和处理;最后,从公开数据集TCGA-LUAD中随机选取337张和200张肺癌组织病理图像,分别构建二分类和三分类数据集进行实验。在二分类数据集上DBFormer模型的AUC,F1-Score,Precision,Recall,分别达到了99.7%,97.3%,97.6%,97.2%;在三分类数据集上DBFormer的Accuracy,Precision,Recall,F1-Score分别达到了97.3%,97.0%,97.0%,97.1%。实验结果表明,DBFormer模型相较于经典深度学习模型,在基于数字病理图像预测肺腺癌肿瘤突变负荷任务上具有更加优异的性能。
中图分类号:
| [1]PANG S,ZHANG Y,DING M,et al.A Deep Model for Lung Cancer Type Identification by Densely Connected Convolutional Networks and Adaptive Boosting[J].IEEE Access,2020,8:4799-4805. [2]SIEGEL R L,KRATZER T B,GIAQUINTO A N,et al.Cancer statistics,2025[J].Ca,2025,75(1):10. [3]RAMALINGAM S,BELANI C P.State-of-the-art chemotherapy for advanced non-small cell lung cancer[J].Seminars in Oncology,2004,31:68-74. [4]ZHOU C C,WANG J,WANG B C,et al.Chinese Experts Consensus on Immune Checkpoint Inhibitors for Non-small Cell Lung Cancer(2020 Version)[J].Chinese Journal of Lung Cancer,2021,24(4):217-235. [5]SETORDZI P,CHANG X,LIU Z,et al.The recent advances of PD-1 and PD-L1 checkpoint signaling inhibition for breast cancer immunotherapy[J].European Journal of Pharmacology,2021,895:173867. [6]CHALMERS Z R,CONNELLY C F,FABRIZIOD,et al.Analysis of 100 000 human cancer genomes reveals the landscape of tumor mutational burden[J].Genome Medicine,2017,9(1):34. [7]YARCHOAN M,HOPKINS A,JAFFEE E M.Tumor Muta-tional Burden and Response Rate to PD-1 Inhibition[J].New England Journal of Medicine,2017,377(25):2500-2501. [8]COUDRAY N,OCAMPO P S,SAKELLAROPOULOS T,et al.Classification and mutation prediction from non-small cell lung cancer histopathology images using deep learning[J].Nature Medicine,2018,24(10):1559-1567. [9]YU W,LUO M,ZHOU P,et al.Metaformer is act-ually what you need for vision[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2022:10819-10829. [10]KRIZHEVSKY A,SUTSKEVER I,HINTONG E.ImageNetclassification with deep convolutional neural networks[J].Communications of the ACM,2017,60(6):84-90. [11]LIN M,CHEN Q,YAN S.Network in network[J].arXiv:1312.4400,2013. [12]HUANG G,LIU Z,VAN DER MAATENL,et al.Deep Convolutional Networks for Large-Scale Image Recognition[C]//ICLR.2017. [13]SZEGEDY C,LIU W,JIA Y Q,et al.Going deeper with convolutions[C]//IEEE Conference on Computer Vision and Pattern Recognition(CVPR).Boston,MA,USA:IEEE,2015:1-9. [14]HE K,ZHANG X,REN S,et al.Deep residual le-arning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2016:770-778. [15]XIE S,GIRSHICK R,DOLLAR P,et al.Aggregated ResidualTransformations for Deep Neural Networks[C]//IEEE Conference on Computer Vision and Pattern Recognition(CVPR).Honolulu,HI:IEEE,2017:5987-5995. [16]HUANG G,LIU Z,VAN DER MAATEN L,et al.DenselyConnected Convolutional Networks[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).Honolulu,HI:IEEE,2017:2261-2269. [17]JANG H J,SONG I H,LEES H.Deep Learning for Automatic Subclassification of Gastric Carcinoma Using Whole-Slide Histopathology Images[J].Cancers,2021,13(15):3811. [18]ZHOU P,CAO Y,LI M,et al.HCCANet:histopathological image grading of colorectal cancer using CNN based on multichannel fusion attention mechanism[J].Scientific Reports,2022,12(1):15103. [19]WANG D,KHOSLA A,GARGEYA R,et al.Deep learning for identifying metastatic breast cancer[J].arXiv:1606.05718,2016. [20]LAM L H T,CHU N T,TRANT O,et al.A Radiomics-Based Machine Learning Model for Prediction of Tumor Mutational Burden in Lower-Grade Gliomas[J].Cancers,2022,14(14):3492. [21]VEERARAGHAVAN H,FRIEDMAN C F,DELAIR D F,et al.Machine learning-based prediction of microsatellite instability and high tumor mutation burden from contrastenhanced computed tomography in endometrial cancers[J].Scientific Reports,2020,10(1):17769. [22]SALDANHA O L,LOEFFLER C M L,NIEHUES J M,et al.Self-supervised attention-based deep learning for pancancer mutation prediction from histopathology[J].NPJ Precision Oncology,2023,7(1):35. [23]SHAMAI G,LIVNE A,POLÓNIA A,et al.Deep learning-based image analysis predicts PD-L1 status from H&E-stained histopathology images in breast cancer[J].Nature Communications,2022,13(1):6753. [24]SADHWANI A,CHANG H W,BEHROOZ A,et al.Comparative analysis of machine learning approaches to classify tumor mutation burden in lung adenocarcinoma using histopathology images[J].Scientific Reports,2021,11(1):16605. [25]CHEN Y,LIU J,JIANG P,et al.Predicting Tumor MutationBurden of TNBC Based on Nuclei Scores of Histopathological Images[C]//IEEE International Conference on Bioinformatics and Biomedicine(BIBM).Las Vegas,NV,USA:IEEE,2022:936-941. [26]JAIN M S,MASSOUD T F.Predicting tumour mutational burden from histopathological images using multiscale deep learning[J].Nature Machine Intelligence,2020,2(6):356-362. [27]CHEN S,XIANG J,WANGX,et al.Pan-cancer computational histopathology reveals tumor mutational burden status through weakly-supervised deep learning[J].arXiv:2204.03257,2022. [28]HUANG K,LIN B,LIU J,et al.Predicting colorectal cancertumor mutational burden from histopathological images and clinical information using multi-modal deep learning[J].Bioinformatics,2022,38(22):5108-5115. |
|
||
首页