面向急性缺血性脑卒中的CT生成MRI算法

doi:10.19678/j.issn.1000-3428.0068435

摘要/Abstract

摘要：

急性缺血性脑卒中病灶在计算机断层扫描（CT）上表现不明显，但在核磁共振成像（MRI）上可以清晰显示。然而，当患者体内有金属植入物等特殊情况则无法进行MRI检测，使得患者的治疗受到延误。通过CT生成MRI可在急性缺血性脑卒中的诊断和治疗中起到至关重要的作用，但现有的医学影像跨模态生成方法从CT得到的MRI缺乏病灶信息且边界模糊。为了解决上述问题，提出一种基于影像组学和扩散生成对抗网络的急性缺血性脑卒中CT生成MRI算法，通过影像组学在CT上进行病灶特征增强，突出生成MRI的病灶信息，引入梯度损失为生成图像与真实图像增加边缘感知约束，提升生成MRI的质量。在ISLES2018挑战赛数据集上的实验结果表明，该算法生成的MRI在整体上的峰值信噪比为23.051 dB，结构相似度为0.798，皮尔逊相关系数为0.969，并且病灶区域的互信息为2.075，与现有的生成模型相比，该算法的各项指标均达到最优。此外，经3名经验丰富的放射科医生在生成的MRI上确定病灶并进行阳性/阴性分类，其中生成的MRI中无错误病灶，且分类准确率可达到86.61%，可作为一种辅助工具协助医生进行诊断。

关键词: 医学图像生成, 影像组学, 扩散生成对抗网络, 计算机断层扫描, 核磁共振成像

Abstract:

The lesions of acute ischemic stroke are not clearly discernible on Computed Tomography(CT), but they are distinctly visible in Magnetic Resonance Imaging(MRI). However, in cases where patients have a metal implant in their body, MRI detection becomes unfeasible, delaying the patient's treatment. MRI generated from CT plays a crucial role in the diagnosis and treatment of acute ischemic stroke. However, MRI obtained from CT using the current medical image cross-modal generation method lacks lesion information and exhibits blurred boundaries. To address these issues, this study proposes a CT-generated MRI algorithm for acute ischemic stroke based on radiomics and diffusion Generative Adversarial Network(GAN). Through imaging radiomics, lesion features are enhanced on CT to highlight lesion information generated by MRI. Additionally, gradient loss is introduced to increase the edge perception constraints between generated and real images, thereby enhancing the quality of the generated MRI. The results of experiments conducted on the ISLES2018 challenge dataset show that the overall Peak Signal-to-Noise Ratio(PSNR)of the generated MRI is 23.051 dB, the Structural Similarity(SSIM) is 0.798, the Pearson Correlation Coefficient(PCC) is 0.969, and the Mutual Information(MI)of the lesion region is 2.075. These results indicate that the proposed algorithm is optimal compared with the existing generation models. In addition, three experienced radiologists identified the lesions on the generated MRI and classified them as positive or negative. No incorrect lesions are identified in the generated MRI, and the classification accuracy reached 86.61%, demonstrating its potential as an auxiliary tool to assist doctors in diagnosis.

Key words: medical image generation, radiomics, diffusion Generated Adversarial Network(GAN), Computed Tomography(CT), Magnetic Resonance Imaging(MRI)

张美美, 秦品乐, 柴锐, 曾建潮, 翟双姣, 闫俊义, 冯二燕. 面向急性缺血性脑卒中的CT生成MRI算法[J]. 计算机工程, 2024, 50(2): 317-326.

Meimei ZHANG, Pinle QIN, Rui CHAI, Jianchao ZENG, Shuangjiao ZHAI, Junyi YAN, Eryan FENG. CT-Generated MRI Algorithm for Acute Ischemic Stroke[J]. Computer Engineering, 2024, 50(2): 317-326.

https://www.ecice06.com/CN/Y2024/V50/I2/317

图/表 12

图1 本文算法总体结构

Fig.1 Overall structure of algorithm in this paper

图2 组学特征相关性热力图

Fig.2 Correlation heat map of radiomics feature

图3 生成器网络结构

Fig.3 Network structure of generator

图4 生成器部分模块结构

Fig.4 Structure of generator partial module

图5 对比实验可视化结果

Fig.5 Visualization results of comparative experiments

图6 消融实验可视化结果

Fig.6 Visualization results of ablation experiment

图7 损失曲线

Fig.7 Loss curve

参考文献 29

1	ROTH G A, JOHNSON C O, ABATE K H, et al. The burden of cardiovascular diseases among US states, 1990-2016. JAMA Cardiology, 2018, 3 (5): 375- 389. doi: 10.1001/jamacardio.2018.0385
2	SACCO R L, KASNER S E, BRODERICK J P, et al. An updated definition of stroke for the 21st century: a statement for healthcare professionals from the American heart association/american stroke association. Stroke, 2013, 44 (7): 2064- 2089. doi: 10.1161/STR.0b013e318296aeca
3	LI W, LI Y F, QIN W J, et al. Magnetic resonance image synthesis from brain computed tomography images based on deep learning methods for magnetic resonance-guided radiotherapy. Quantitative Imaging in Medicine and Surgery, 2020, 10 (6): 1223- 1236. doi: 10.21037/qims-19-885
4	WANG G T, SONG T, DONG Q, et al. Automatic ischemic stroke lesion segmentation from computed tomography perfusion images by image synthesis and attention-based deep neural networks. Medical Image Analysis, 2020, 65, 101787. doi: 10.1016/j.media.2020.101787
5	HU N, ZHANG T W, WU Y F, et al. Detecting brain lesions in suspected acute ischemic stroke with CT-based synthetic MRI using generative adversarial networks. Annals of Translational Medicine, 2022, 10 (2): 35. doi: 10.21037/atm-21-4056
6	FENG E, QIN P, CHAI R, et al. MRI generated from CT for acute ischemic stroke combining radiomics and generative adversarial networks. IEEE Journal of Biomedical and Health Informatics, 2022, 26 (12): 6047- 6057. doi: 10.1109/JBHI.2022.3205961
7	LÜ Q, WANG G. Conversion between CT and MRI images using diffusion and score-matching models[EB/OL]. [2023-08-10]. https://arxiv.org/abs/2209.12104.
8	MENG X X, GU Y N, PAN Y S, et al. A novel unified conditional score-based generative framework for multi-modal medical image completion[EB/OL]. [2023-08-10]. https://arxiv.org/abs/2207.03430.
9	OZBEY M, DALMAZ O, DAR S U H, et al. Unsupervised medical image translation with adversarial diffusion models. IEEE Transactions on Medical Imaging, 2023, 42 (12): 3524- 3539. doi: 10.1109/TMI.2023.3290149
10	LAMBIN P, RIOS-VELAZQUEZ E, LEIJENAAR R, et al. Radiomics: extracting more information from medical images using advanced feature analysis. European Journal of Cancer, 2012, 48 (4): 441- 446. doi: 10.1016/j.ejca.2011.11.036
11	GE G, ZHANG J. Feature selection methods and predictive models in CT lung cancer radiomics. Journal of Applied Clinical Medical Physics, 2023, 24 (1): 13869. doi: 10.1002/acm2.13869
12	CHEN Q, XIA T Y, ZHANG M Y, et al. Radiomics in stroke neuroimaging: techniques, applications, and challenges. Aging and Disease, 2021, 12 (1): 143- 154. doi: 10.14336/AD.2020.0421
13	SIKIÖ M, KÖLHI P, RYYMIN P, et al. MRI texture analysis and diffusion tensor imaging in chronic right hemisphere ischemic stroke. Journal of Neuroimaging, 2015, 25 (4): 614- 619. doi: 10.1111/jon.12185
14	ORTIZ-RAMÓN R, VALDÉS HERNÁNDEZ M D C. Identification of the presence of ischaemic stroke lesions by means of texture analysis on brain magnetic resonance images. Computerized Medical Imaging and Graphics, 2019, 74, 12- 24. doi: 10.1016/j.compmedimag.2019.02.006
15	HOFMEISTER J, BERNAVA G, ROSI A, et al. Clot-based radiomics predict a mechanical thrombectomy strategy for successful recanalization in acute ischemic stroke. Stroke, 2020, 51 (8): 2488- 2494. doi: 10.1161/STROKEAHA.120.030334
16	GUAN Y, WANG P, WANG Q, et al. Separability of acute cerebral infarction lesions in CT based radiomics: toward artificial intelligence-assisted diagnosis. BioMed Research International, 2020, 2020, 8864756. doi: 10.1155/2020/8864756
17	NIE D, TRULLO R, LIAN J, et al. Medical image synthesis with context-aware generative adversarial networks[C]//Proceedings of the 20th International Conference on Medical Image Computing and Computer Assisted Intervention. Berlin, Germany: Springer, 2017: 417-425.
18	YANG H R, SUN J, CARASS A, et al. Unsupervised MR-to-CT synthesis using structure-constrained CycleGAN. IEEE Transactions on Medical Imaging, 2020, 39 (12): 4249- 4261. doi: 10.1109/TMI.2020.3015379
19	BEN-COHEN A, KLANG E, RASKIN S P, et al. Cross-modality synthesis from CT to PET using FCN and GAN networks for improved automated lesion detection. Engineering Applications of Artificial Intelligence, 2019, 78, 186- 194. doi: 10.1016/j.engappai.2018.11.013
20	ARMANIOUS K, JIANG C M, FISCHER M, et al. MedGAN: medical image translation using GANs. Computerized Medical Imaging and Graphics, 2020, 79, 101684. doi: 10.1016/j.compmedimag.2019.101684
21	王红玉, 朱天薏, 冯筠, 等. 基于生成对抗网络的乳腺MRI图像生成. 西北大学学报(自然科学版), 2023, 53 (3): 348- 358. URL
	WANG H Y, ZHU T Y, FENG J, et al. Breast MRI image generation method based on generative adversarial network. Journal of Northwest University (Natural Science Edition), 2023, 53 (3): 348- 358. URL
22	刘少鹏, 赵慧民, 洪佳明, 等. 面向医学图像生成的鲁棒条件生成对抗网络. 电子学报, 2023, 51 (2): 427- 437. URL
	LIU S P, ZHAO H M, HONG J M, et al. Medical image synthesis using robust conditional GAN. Acta Electronica Sinica, 2023, 51 (2): 427- 437. URL
23	PINAYA W H L, GRAHAM M S, GRAY R, et al. Fast unsupervised brain anomaly detection and segmentation with diffusion models[C]//Proceedings of International Conference on Medical Image Computing and Computer Assisted Intervention. Berlin, Germany: Springer, 2022: 705-714.
24	DUAN Y Y, QIN J, QIU W Q, et al. Performance of a generative adversarial network using ultrasound images to stage liver fibrosis and predict cirrhosis based on a deep-learning radiomics nomogram. Clinical Radiology, 2022, 77 (10): 723- 731. doi: 10.1016/j.crad.2022.06.003
25	PAN S Y, FLORES J, LIN C T, et al. Generative adversarial networks and radiomics supervision for lung lesion synthesis[C]//Proceedings of SPIE'21. Washington D. C., USA: IEEE Press, 2021: 167-172.
26	CHEN J H, BERMEJO I, DEKKER A, et al. Generative models improve radiomics performance in different tasks and different datasets: an experimental study. Physica Medica, 2022, 98, 11- 17. doi: 10.1016/j.ejmp.2022.04.008
27	RONNEBERGER O, FISCHER P, BROX T. U-net: convolutional networks for biomedical image segmentation[C]//Proceedings of the 18th International Conference on Lecture Notes in Computer Science. Berlin, Germany: Springer, 2015: 234-241.
28	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2016: 770-778.
29	SONG Y, SOHL-DICKSTEIN J N, KINGMA D P, et al. Score-based generative modeling through stochastic differential equations[EB/OL]. [2023-08-10]. https://arxiv.org/abs/2011.13456.

[1]	田炜, 雷志超, 王楚正. 基于多期相注意力融合网络的肝脏病灶CT影像分类研究[J]. 计算机工程, 2023, 49(2): 231-237.
[2]	杨本臣, 贾宇航, 金海波. 融合多分支特征的肝脏和肝脏肿瘤的体积分割[J]. 计算机工程, 2023, 49(10): 194-201.
[3]	周博超, 韩雨男, 桂志国, 李郁峰, 张权. 基于VGG网络与深层字典的低剂量CT图像去噪算法[J]. 计算机工程, 2022, 48(4): 191-196,205.
[4]	唐江平, 周晓飞, 贺鑫, 褚晓文, 李世锋, 常庆蕊, 张继勇. 基于深度学习的新型冠状病毒肺炎诊断研究综述[J]. 计算机工程, 2021, 47(5): 1-15.
[5]	刘天宇, 姜威威, 何江萍, 韩金仓. 基于HC-CFCN模型的肝脏CT图像分割[J]. 计算机工程, 2020, 46(2): 268-273.
[6]	陈扬洋, 钱鹏江, 赵开发, 苏冠豪. 基于mDixon序列下腹部MRI数据的sCT生成方法[J]. 计算机工程, 2019, 45(7): 273-281.
[7]	高媛,贾紫婷,秦品乐,王丽芳. 基于压缩感知与自适应PCNN的医学图像融合[J]. 计算机工程, 2018, 44(9): 224-229.
[8]	冯宝,刘晓刚. 基于先验信息的脑图像数据信息提取算法[J]. 计算机工程, 2015, 41(9): 194-198.
[9]	冯宝,秦传波. 基于凸优化的脑图像数据盲信号分离算法[J]. 计算机工程, 2015, 41(8): 233-237.
[10]	冯宝,刘晓刚. 基于字典稀疏性的脑图像数据盲分离方法[J]. 计算机工程, 2015, 41(12): 269-274.
[11]	袁静. 基于压缩感知的核磁共振成像重构算法[J]. 计算机工程, 2015, 41(10): 270-274.
[12]	王海军, 柳明. 一种基于局部信息的多相脑图分割模型[J]. 计算机工程, 2012, 38(7): 185-187.
[13]	陈亮, 陈允杰. 基于改进高斯混合模型的MR图像分割[J]. 计算机工程, 2012, 38(17): 242-244,248.

选择文件类型/文献管理软件名称

选择包含的内容