利用卷积神经网络对杜兴氏肌营养不良症进行分类识别
|
摘要
杜兴氏肌营养不良(DMD)是一种致死性骨骼肌遗传病。传统的诊断方法中包含创伤性检查,会令患者产生极大的痛苦。本文基于受试者的磁共振图像(MRI),探索DMD的无创检测方法。共获取试验图像485幅,分成患者组和健康对照组,每组均包括T1和T2这2类加权图像。试验设计一个10隐层卷积神经网络(CNN),利用该网络直接读取T1和T2并分类识别。结果显示:1)随着网络参数的优化和迭代次数的增加,图像识别准确度分别达到99. 2%和98. 9%,结果收敛且稳定; 2) 2类加权图像T1和T2均能很好区分患者组与健康组; 3)与KNN、LR、DT及SVM等算法相比,CNN算法的分类准确度更高。CNN尤其提升了对于T2图像的识别准确度,大大发掘了T2图像的利用价值。因此,利用CNN对DMD图像进行分类识别,因其准确度高、无损图像信息等特点,有望为临床提供客观有效的辅助诊断手段。这是人工智能在DMD无创检测领域中有效的尝试探索。
Duchenne muscular dystrophy( DMD) is a fatal skeletal muscle hereditary disease. The conventional treatment is invasive,which incurs great sufferings. Therefore,this paper explores a non-invasive detection method on the basis of magnetic resonance images( MRI) of the patients. 485 experimental MRIs are obtained with the guidance of senior physicians of neuromuscular department. These images are divided into two control groups: the patient group and the healthy group; each group includes two weighted images,T1 and T2. A 10-hidden layer depth convolutional neural network( CNN) is designed and used to directly read T1 and T2,and classify them. The results show: firstly,by increasing the numbers of network parameters and iterative optimizations,the accuracies of image recognition have reached 99. 2% and 98. 9% respectively; secondly,both T1 and T2 can be used to well distinguish between patient and healthy groups; thirdly,in comparison with KNN,LR,DT and SVM algorithms,the accuracy of classification with the CNN algorithm is best. In particular,the CNN algorithm improves the recognition accuracy of T2 images,and greatly explores the utilization value of T2 images. Therefore,using CNN for DMD image classification and recognition,because of its high accuracy,lossless image information and other characteristics,it is expected to provide an objective and effective auxiliary diagnosis means for clinical; this is a new exploration of application of artificial intelligence in the field of DMD non-invasive detection.
Duchenne muscular dystrophy( DMD) is a fatal skeletal muscle hereditary disease. The conventional treatment is invasive,which incurs great sufferings. Therefore,this paper explores a non-invasive detection method on the basis of magnetic resonance images( MRI) of the patients. 485 experimental MRIs are obtained with the guidance of senior physicians of neuromuscular department. These images are divided into two control groups: the patient group and the healthy group; each group includes two weighted images,T1 and T2. A 10-hidden layer depth convolutional neural network( CNN) is designed and used to directly read T1 and T2,and classify them. The results show: firstly,by increasing the numbers of network parameters and iterative optimizations,the accuracies of image recognition have reached 99. 2% and 98. 9% respectively; secondly,both T1 and T2 can be used to well distinguish between patient and healthy groups; thirdly,in comparison with KNN,LR,DT and SVM algorithms,the accuracy of classification with the CNN algorithm is best. In particular,the CNN algorithm improves the recognition accuracy of T2 images,and greatly explores the utilization value of T2 images. Therefore,using CNN for DMD image classification and recognition,because of its high accuracy,lossless image information and other characteristics,it is expected to provide an objective and effective auxiliary diagnosis means for clinical; this is a new exploration of application of artificial intelligence in the field of DMD non-invasive detection.
引文
[1]黄艳军,郭虎,何燕,等. Duchenne肌营养不良15例临床及基因分析并文献复习[J].中华临床医师杂志(电子版),2014,8(7):1275-1278.
[2]李悦,李小明,卜碧涛.磁共振在肌肉疾病诊断中的应用[J].神经损伤与功能重建,2011,6(4):302-304.
[3]郭媛,庹军,吴波.线粒体脑肌病:六例报告并文献复习[J].中国现代神经疾病杂志,2014,14(7):621-624.
[4] JONES T A,WAYTE S C,REDDY N L,et al. Identification of an optimal threshold for detecting human brown adipose tissue using receiver operating characteristic analysis of IDEAL MRI fat fraction maps[J]. Magnetic Resonance Imaging,2018,51:61-68.
[5]许晓杰,卢辉群,周玮,等.应用磁共振成像观察首发强迫症患者海马体积及相关静脉的改变[J].江苏大学学报(医学版),2017,27(6):500-503.
[6]宁殿秀,郎志谨,唐乐梅,等.颞叶癫痫的磁共振CUBE成像技术临床应用价值[J].临床放射学杂志,2017,36(11):1562-1565.
[7] EL HOUBY E M F. Framework of computer aided diagnosis systems for cancer classification based on medical images[J]. Journal of Medical Systems,2018,42(8),doi:10. 1007/s10916-018-1010-x.
[8] ELSALAMONY H A. Detection of anaemia disease in human red blood cells using cell signature,neural networks and SVM[J]. Multimedia Tools and Applications,2018,77(12):15047-15074.
[9]贾紫婷.基于脉冲耦合神经网络的医学图像融合算法[D].太原:中北大学,2018.
[10] FITZMAURICE G M,LAIRD N M. Binary response models and logistic regression[M]//International Encyclopedia of the Social&Behavioral Sciences. 2015:587-595.
[11] HU H,PANG L,SHI Z Z. Image matting in the perception granular deep learning[J]. Knowledge-based Systems,2016,102:51-63.
[12] WANG W N,ZHAO M Q,WANG L,et al. A multi-scene deep learning model for image aesthetic evaluation[J].Signal Processing:Image Communication,2016,47:511-518.
[13] ISIN A,DIREKOGLU C, AH M. Review of MRI-based brain tumor image segmentation using deep learning methods[J]. Procedia Computer Science,2016,102:317-324.
[14]刘建伟,刘媛,罗雄麟.深度学习研究进展[J].计算机应用研究,2014,31(7):1921-1930.
[15]孙志远,鲁成祥,史忠植,等.深度学习研究与进展[J].计算机科学,2016,43(2):1-8.
[16] ZHANG M H,MA J S,SHEN Y,et al. Optimal classification for the diagnosis of Duchenne muscular dystrophy images using support vector machines[J]. International Journal of Computer Assisted Radiology&Surgery,2016,11(9):1755-1763.
[17]古冬连,金观桥,苏丹柯.磁共振弛豫时间T1、T2的测量方法及其在肿瘤中的应用价值[J].广西医学,2017,39(5):686-688.
[18]赵芳.磁共振图像对软组织肉瘤组织学级别的诊断价值[D].济南:山东大学,2014.
[19]周超,徐军,罗波.基于深度卷积网络和结合策略的乳腺组织病理图像细胞核异型性自动评分[J].中国生物医学工程学报,2017,36(3):276-283.
[20]虞仁和. SPSS 18及其医学应用[M].长沙:中南大学出版社,2012.
[21]林辉煌.基于预决策的卷积神经网络加速研究[D].北京:北京工业大学,2016.
[22]唐启永,徐凯.基于小波变换和区域分割的红外可见光彩色图像融合算法[J].光学技术,2015,41(1):68-71.
[23]胡珊,张俊杰,徐超,等.基于小波变换的骨肉瘤CR图像分析[J].中国数字医学,2014,9(6):71-74.
[24] XUE M F,LI X P,LIU Y M,et al. Denoising of X-ray pulsar observed profile using biorthogonal lifting wavelet transform[J]. Journal of Systems Engineering and Electronics,2016,27(3):514-523.
[2]李悦,李小明,卜碧涛.磁共振在肌肉疾病诊断中的应用[J].神经损伤与功能重建,2011,6(4):302-304.
[3]郭媛,庹军,吴波.线粒体脑肌病:六例报告并文献复习[J].中国现代神经疾病杂志,2014,14(7):621-624.
[4] JONES T A,WAYTE S C,REDDY N L,et al. Identification of an optimal threshold for detecting human brown adipose tissue using receiver operating characteristic analysis of IDEAL MRI fat fraction maps[J]. Magnetic Resonance Imaging,2018,51:61-68.
[5]许晓杰,卢辉群,周玮,等.应用磁共振成像观察首发强迫症患者海马体积及相关静脉的改变[J].江苏大学学报(医学版),2017,27(6):500-503.
[6]宁殿秀,郎志谨,唐乐梅,等.颞叶癫痫的磁共振CUBE成像技术临床应用价值[J].临床放射学杂志,2017,36(11):1562-1565.
[7] EL HOUBY E M F. Framework of computer aided diagnosis systems for cancer classification based on medical images[J]. Journal of Medical Systems,2018,42(8),doi:10. 1007/s10916-018-1010-x.
[8] ELSALAMONY H A. Detection of anaemia disease in human red blood cells using cell signature,neural networks and SVM[J]. Multimedia Tools and Applications,2018,77(12):15047-15074.
[9]贾紫婷.基于脉冲耦合神经网络的医学图像融合算法[D].太原:中北大学,2018.
[10] FITZMAURICE G M,LAIRD N M. Binary response models and logistic regression[M]//International Encyclopedia of the Social&Behavioral Sciences. 2015:587-595.
[11] HU H,PANG L,SHI Z Z. Image matting in the perception granular deep learning[J]. Knowledge-based Systems,2016,102:51-63.
[12] WANG W N,ZHAO M Q,WANG L,et al. A multi-scene deep learning model for image aesthetic evaluation[J].Signal Processing:Image Communication,2016,47:511-518.
[13] ISIN A,DIREKOGLU C, AH M. Review of MRI-based brain tumor image segmentation using deep learning methods[J]. Procedia Computer Science,2016,102:317-324.
[14]刘建伟,刘媛,罗雄麟.深度学习研究进展[J].计算机应用研究,2014,31(7):1921-1930.
[15]孙志远,鲁成祥,史忠植,等.深度学习研究与进展[J].计算机科学,2016,43(2):1-8.
[16] ZHANG M H,MA J S,SHEN Y,et al. Optimal classification for the diagnosis of Duchenne muscular dystrophy images using support vector machines[J]. International Journal of Computer Assisted Radiology&Surgery,2016,11(9):1755-1763.
[17]古冬连,金观桥,苏丹柯.磁共振弛豫时间T1、T2的测量方法及其在肿瘤中的应用价值[J].广西医学,2017,39(5):686-688.
[18]赵芳.磁共振图像对软组织肉瘤组织学级别的诊断价值[D].济南:山东大学,2014.
[19]周超,徐军,罗波.基于深度卷积网络和结合策略的乳腺组织病理图像细胞核异型性自动评分[J].中国生物医学工程学报,2017,36(3):276-283.
[20]虞仁和. SPSS 18及其医学应用[M].长沙:中南大学出版社,2012.
[21]林辉煌.基于预决策的卷积神经网络加速研究[D].北京:北京工业大学,2016.
[22]唐启永,徐凯.基于小波变换和区域分割的红外可见光彩色图像融合算法[J].光学技术,2015,41(1):68-71.
[23]胡珊,张俊杰,徐超,等.基于小波变换的骨肉瘤CR图像分析[J].中国数字医学,2014,9(6):71-74.
[24] XUE M F,LI X P,LIU Y M,et al. Denoising of X-ray pulsar observed profile using biorthogonal lifting wavelet transform[J]. Journal of Systems Engineering and Electronics,2016,27(3):514-523.