基于课程学习训练的聚合注意力网络Multi-SEANet用于MRI图像的格里森级别组无创预测

doi:10.1007/s12204-022-2502-x

J Shanghai Jiaotong Univ Sci ›› 2024, Vol. 29 ›› Issue (1): 109-119.doi: 10.1007/s12204-022-2502-x

基于课程学习训练的聚合注意力网络Multi-SEANet用于MRI图像的格里森级别组无创预测

沈傲1, 2，胡冀苏2, 3，金鹏飞4，周志勇2，钱旭升2, 3，郑毅2，包婕4，王希明4，戴亚康1, 2

（1.长春理工大学计算机科学与技术学院，长春 130022；2. 中国科学院苏州生物医学工程技术研究所，江苏苏州 215163；3. 中国科学技术大学生物医学工程学院 (苏州)，江苏苏州 215163；4. 苏州大学附属第一医院放射科，江苏苏州 215000）

接受日期:2022-03-31 出版日期:2024-01-28 发布日期:2024-01-24

Ensemble Attention Guided Multi-SEANet Trained with Curriculum Learning for Noninvasive Prediction of Gleason Grade Groups from MRI

(1. School of Computer Science and Technology, Changchun University of Science and Technology, Changchun 130022, China; 2. Suzhou Institute of Biomedical Engineering and Technology, Chinese Academy of Sciences, Suzhou 215163, Jiangsu, China; 3. School of Biomedical Engineering (SooChow), University of Science and Technology of China, Suzhou 215163, Jiangsu, China; 4. Department of Radiology, The First Affiliated Hospital of Soochow University, Suzhou 215006, Jiangsu, China)

Accepted:2022-03-31 Online:2024-01-28 Published:2024-01-24

摘要/Abstract

摘要： 格里森分级组(Gleason grade group, GG)是评估前列腺癌恶性程度的重要依据，但需要侵入性活检才能获得病理结果。为了无创地预测GG，提出了一种基于课程学习训练的集成注意力模块的多尺度卷积神经网络的自动预测方法。首先，提出了基于感兴趣区域图像的病灶注意力图，并与残存注意力模块相融合，使网络更加关注病灶区域。其次，结合特征金字塔网络，使网络更好地学习病灶区域的多尺度信息。最后，在网络训练中，提出了基于视觉评价与病理分级一致性差异的课程学习网络训练框架，进一步提高了网络的预测性能。实验结果表明，该方法在预测GG性能方面优于传统网络模型。二次加权Kappa结果为0.4711，用于评估临床显著性癌症的阳性预测值为0.9369。

关键词: 前列腺癌，格里森级别组，双参数核磁共振图像，深度学习，课程学习

Abstract: The Gleason grade group (GG) is an important basis for assessing the malignancy of prostate cancer, but it requires invasive biopsy to obtain pathology. To noninvasively evaluate GG, an automatic prediction method is proposed based on multi-scale convolutional neural network of the ensemble attention module trained with curriculum learning. First, a lesion-attention map based on the image of the region of interest is proposed in combination with the bottleneck attention module to make the network more focus on the lesion area. Second, the feature pyramid network is combined to make the network better learn the multi-scale information of the lesion area. Finally, in the network training, a curriculum based on the consistency gap between the visual evaluation and the pathological grade is proposed, which further improves the prediction performance of the network. Experimental results show that the proposed method is better than the traditional network model in predicting GG performance. The quadratic weighted Kappa is 0.471 1 and the positive predictive value for predicting clinically significant cancer is 0.936 9.

Key words: prostate cancer, Gleason grade groups (GGs), bi-parametric magnetic resonance imaging, deep learning, curriculum learning

中图分类号:

沈傲1, 2，胡冀苏2, 3，金鹏飞4，周志勇2，钱旭升2, 3，郑毅2，包婕4，王希明4，戴亚康1, 2. 基于课程学习训练的聚合注意力网络Multi-SEANet用于MRI图像的格里森级别组无创预测[J]. J Shanghai Jiaotong Univ Sci, 2024, 29(1): 109-119.

SHEN Ao^1,2‡ (沈傲), HU Jisu^2,3‡ (胡冀苏), JIN Pengfei⁴ (金鹏飞), ZHOU Zhiyong² (周志勇), QIAN Xusheng^2,3 (钱旭升), ZHENG Yi² (郑毅), BAO Jie⁴ (包婕), WANG Ximing^4∗ (王希明), DAI Yakang^1,2∗ (戴亚康). Ensemble Attention Guided Multi-SEANet Trained with Curriculum Learning for Noninvasive Prediction of Gleason Grade Groups from MRI[J]. J Shanghai Jiaotong Univ Sci, 2024, 29(1): 109-119.

参考文献

[1] SCHELTEMA M J, CHANG J I, STRICKER P D, et al. Diagnostic accuracy of 68Ga-prostate-specific membrane antigen (PSMA) positron-emission tomography (PET) and multiparametric (mp)MRI to detect intermediate-grade intra-prostatic prostate cancer using whole-mount pathology: Impact of the addition of 68Ga-PSMA PET to mpMRI [J]. BJU International, 2019, 124: 42-49.
[2] DROST F J H, OSSES D F, NIEBOER D, et al. Prostate MRI, with or without MRI-targeted biopsy, and systematic biopsy for detecting prostate cancer [J]. The Cochrane Database of Systematic Reviews, 2019, 4(4): CD012663.
[3] FEHR D, VEERARAGHAVAN H, WIBMER A, et al. Automatic classification of prostate cancer Gleason scores from multiparametric magnetic resonance images [J]. Proceedings of the National Academy of Sciences of the United States of America, 2015, 112(46): E6265-E6273.
[4] JENSEN C, CARL J, BOESEN L, et al. Assessment of prostate cancer prognostic Gleason grade group using zonal: Specific features extracted from biparametric MRI using a KNN classifier [J]. Journal of Applied Clinical Medical Physics, 2019, 20(2): 146-153.
[5] CAO R M, MOHAMMADIAN BAJGIRAN A, AFSHARI MIRAK S, et al. Joint prostate cancer detection and gleason score prediction in mp-MRI via FocalNet [J]. IEEE Transactions on Medical Imaging, 2019, 38(11): 2496-2506.
[6] ARMATO S G, HUISMAN H, DRUKKER K, et al. PROSTATEx Challenges for computerized classification of prostate lesions from multiparametric mag netic resonance images [J]. Journal of Medical Imaging, 2018, 5(4): 044501.
[7] ABRAHAM B, NAIR M S. Computer-aided classification of prostate cancer grade groups from MRI images using texture features and stacked sparse autoencoder [J]. Computerized Medical Imaging and Graphics, 2018, 69: 60-68.
[8] ABRAHAM B, NAIR M S. Automated grading of prostate cancer using convolutional neural network and ordinal class classifier [J]. Informatics in Medicine Unlocked, 2019, 17: 100256.
[9] WANG X, ZHAO X Y, LI Q, et al. Can peritumoral radiomics increase the efficiency of the prediction for lymph node metastasis in clinical stage T1 lung adenocarcinoma on CT? [J]. European Radiology, 2019, 29(11): 6049-6058.
[10] HAN D, WEI Y, WANG X D, et al. Association of peripheral nerve invasion with clinicopathological factors and prognosis of colorectal cancer [J]. Chinese Journal of Gastrointestinal Surgery, 2017, 20(1): 62-66.
[11] HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition [C]//2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas: IEEE, 2016: 770-778.
[12] LIN T Y, DOLLAR P, GIRSHICK R, et al. Feature pyramid networks for object detection [C]//2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu: IEEE, 2017: 936-944.
[13] IKECHUKWU A V, MURALI S, DEEPU R, et al. ResNet-50 vs VGG-19 vs training from scratch: A comparative analysis of the segmentation and classi- fication of Pneumonia from chest X-ray images [J]. Global Transitions Proceedings, 2021, 2(2): 375-381.
[14] WOO S, PARK J, LEE J Y, et al. CBAM: Convolutional block attention module [M]// Computer vision - ECCV 2018. Cham: Springer, 2018: 3-19.
[15] HOU Y, ZHANG Y H, BAO J, et al. Artificial intelligence is a promising prospect for the detection of prostate cancer extracapsular extension with mpMRI: A two-center comparative study [J]. European Journal of Nuclear Medicine and Molecular Imaging, 2021, 48(12): 3805-3816.
[16] EPSTEIN J I, ZELEFSKY M J, SJOBERG D D, et al. A contemporary prostate cancer grading system: A validated alternative to the Gleason score [J]. European Urology, 2016, 69(3): 428-435.
[17] TANG Y X, WANG X S, HARRISON A P, et al. Attention-guided curriculum learning for weakly supervised classification and localization of thoracic diseases on chest radiographs [M]//Machine learning in medical imaging. Cham: Springer, 2018: 249-258.
[18] LI H Y, LIU X B, BOUMARAF S, et al. A new three-stage curriculum learning approach for deep network based liver tumor segmentation [C]//2020 International Joint Conference on Neural Networks. Glasgow: IEEE, 2020: 1-6. [19] ISENSEE F, JAEGER P F, KOHL S A A, et al. nnUNet: a self-configuring method for deep learning-based biomedical image segmentation [J]. Nature Methods, 2021, 18(2): 203-211.
[20] HU J, SHEN L, SUN G. Squeeze-and-excitation networks [C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City: IEEE, 2018: 7132-7141.
[21] HUANG G, LIU Z, VAN DER MAATEN L, et al. Densely connected convolutional networks [C]//2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu: IEEE, 2017: 2261-2269.
[22] LIAO J H, DUAN H H, DAI H M, et al. Automatic detection of intracranial aneurysm from digital subtraction angiography with cascade networks [C]//2nd International Conference on Artificial Intelligence and Pattern Recognition. Beijing: ACM, 2019: 18-23.
[23] GHOSH S, SANTOSH K C. Tumor segmentation in brain MRI: U-nets versus feature pyramid network [C]//2021 IEEE 34th International Symposium on Computer-Based Medical Systems. Aveiro: IEEE, 2021: 31-36.
[24] WANG Z, LIU C, MA L H. LandmarkNet: A 2D digital radiograph landmark estimator for registration [J]. BMC Medical Informatics and Decision Making, 2020, 20(1): 168.
[25] XIE X Z, NIU J W, LIU X F, et al. A survey on incorporating domain knowledge into deep learning for medical image analysis [J]. Medical Image Analysis, 2021, 69: 101985.
[26] SANFORD T, HARMON S A, TURKBEY E B, et al. Deep-learning-based artificial intelligence for PIRADS classification to assist multiparametric prostate MRI interpretation: A development study [J]. Journal of Magnetic Resonance Imaging, 2020, 52(5): 1499-1507.
[27] BARENTSZ J O, WEINREB J C, VERMA S, et al. Synopsis of the PI-RADS v2 guidelines for multiparametric prostate magnetic resonance imaging and recommendations for use [J]. European Urology, 2016, 69(1): 41-49.

基于课程学习训练的聚合注意力网络Multi-SEANet用于MRI图像的格里森级别组无创预测

Ensemble Attention Guided Multi-SEANet Trained with Curriculum Learning for Noninvasive Prediction of Gleason Grade Groups from MRI

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 1

编辑推荐

Metrics

本文评价