A multicenter study of 2.5D convolutional neural networks based on multi-sequence MRI in distinguishing meningioma

Share this content in WeChat

• Clinical Article •

GUO Kaican LIU Ting LIU Gaoyuan ZHANG Yong LIU Xiangchu LU Zhongyan ZHOU Yuanlin LI Bing

Cite this article as: GUO K C, LIU T, LIU G Y, et al. A multicenter study of 2.5D convolutional neural networks based on multi-sequence MRI in distinguishing meningioma[J]. Chin J Magn Reson Imaging, 2025, 16(2): 20-28. DOI:10.12015/issn.1674-8034.2025.02.004.

Abstract

[Abstract] Objective To explore the value of 2.5D convolutional neural networks (CNN) based on T2WI, DWI, and enhanced T1WI sequences in distinguishing meningiomas from other similar-appearing tumors.Materials and Methods A total of 674 cases with histopathologically confirmed meningiomas and non-meningiomas with similar imaging features were retrospectively collected from three hospitals (A, B, and C). Among them, 414 cases from hospital A (meningiomas, n = 178; non-meningiomas, n = 236) were used as the training set, 95 cases from hospital B (meningiomas, n = 41; non-meningiomas, n = 54) were used as the test set, and 165 cases from hospital C (meningiomas, n = 78; non-meningiomas: n = 87) were used as the validation set. All cases were classified into five categories: solitary fibrous tumor/ hemangiopericytoma (Class_0), meningioma (Class_1), lymphoma (Class_2), metastatic tumor (Class_3), and cartilage-derived and other similar-appearing tumors (Class_4). A Gradient Boosted Decision Trees (GBDT) model was constructed based on MRI features, and three types of 2.5D CNN, namely ResNet50, DenseNet169, and ResNext50_32x4d, were developed using the input MRI images. After a comprehensive comparison of the performance of these models, the optimal model was selected. Six radiologists with varying levels of experience (two at each level of junior, intermediate, and senior) independently diagnosed cases in the validation set to assess the consistency of the optimal model's diagnostic outcomes with those of radiologists with different levels of experience.Results Among the four multi-class diagnostic models, ResNext50_32x4d was determined to be the optimal model, with accuracies of 86.7%, 82.1%, and 80.6% in the training, test, and validation sets, respectively. Six radiologists with varying levels of diagnostic experience (designed as Radiologist A through Radiologist F) achieved accuracies of 61.2%, 66.3%, 72.1%, 77.9%, 80.1% and 83.2% in thevalidation set, respectively. The optimal model showed better consistency with the diagnostic outcomes of the two senior radiologists, with intraclass correlation coefficients (ICC) of 0.735 and 0.862, respectively.Conclusions The developed 2.5D CNN model based on multi-sequences MRI has good classification and prediction performance in the differential diagnosis of meningiomas, providing valuable reference for distinguishing meningiomas from other brain tumors.

[Keywords] meningioma；magnetic resonance imaging；deep learning；convolutional neural network；differential diagnosis

Contributor Information

GUO Kaican¹ LIU Ting¹ LIU Gaoyuan¹ ZHANG Yong^1* LIU Xiangchu² LU Zhongyan¹ ZHOU Yuanlin¹ LI Bing³

¹ Department of Radiology, Deyang People's Hospital, Deyang 618000, China

² Department of Radiology, Mianzhu People's Hospital, Mianzhu 618200, China

³ Department of Radiology, Affiliated Hospital of North Sichuan Medical College, Nanchong 637000, China

Corresponding author: ZHANG Y, E-mail: 759740128@qq.com

Conflicts of interest None.

Publication Information

Received 2024-09-05

Accepted 2025-01-10

DOI: 10.12015/issn.1674-8034.2025.02.004

Text

References

[1]

LOUIS D N, PERRY A, WESSELING P, et al. The 2021 WHO Classification of Tumors of the Central Nervous System: a summary[J]. Neuro Oncol, 2021, 23(8): 1231-1251. DOI: 10.1093/neuonc/noab106.

[2]

HAN T, ZHOU J L. Advances in imaging study on grading and typing of meningiomas[J]. Chin J Magn Reson Imaging, 2021, 12(7): 94-97. DOI: 10.12015/issn.1674-8034.2021.07.022.

[3]

LI Z H, YOU H, FENG F. MRI features of dural lesions[J]. Chin J Radiol, 2024, 58(5): 558-563. DOI: 10.3760/cma.j.cn112149-20240209-00069.

[4]

LIU X, DENG J, SUN Q, et al. Differentiation of intracranial solitary fibrous tumor/hemangiopericytoma from atypical meningioma using apparent diffusion coefficient histogram analysis[J]. Neurosurg Rev, 2022, 45(3): 2449-2456. DOI: 10.1007/s10143-022-01771-x.

[5]

FORST D A, JONES P S. Skull base tumors[J]. Continuum (Minneap Minn), 2023, 29(6): 1752-1778. DOI: 10.1212/CON.0000000000001361.

[6]

GOKOGLU A, ORUNOGLU M, EKSI M S, et al. Dural and calvarial metastasis of thyroid follicular carcinoma mimicking Sindou type 6 parafalcine meningioma[J]. J Cancer Res Ther, 2023, 19(7): 2098-2100. DOI: 10.4103/jcrt.jcrt_2017_21.

[7]

ZHAO H M, ZHANG H. Research progress of intelligent image prediction of MGMT methylation status in high-grade glioma[J]. Chin J Magn Reson Imaging, 2022, 13(2): 130-132, 136. DOI: 10.12015/issn.1674-8034.2022.02.032.

[8]

JIANG X, ZHAO H, SALDANHA O L, et al. An MRI deep learning model predicts outcome in rectal cancer[J/OL]. Radiology, 2023, 307(5): e222223 [2024-09-05]. https://pubmed.ncbi.nlm.nih.gov/37278629/. DOI: 10.1148/radiol.222223.

[9]

JIANG Y, ZHOU K, SUN Z, et al. Non-invasive tumor microenvironment evaluation and treatment response prediction in gastric cancer using deep learning radiomics[J/OL]. Cell Rep Med. 2023, 4(8): 101146 [2024-09-05]. https://pubmed.ncbi.nlm.nih.gov/37557177/. DOI: 10.1016/j.xcrm.2023.101146.

[10]

MUHAMMAD K, KHAN S, SER J D, et al. Deep learning for multigrade brain tumor classification in smart healthcare systems: A prospective survey[J]. IEEE Trans Neural Netw Learn Syst, 2021, 32(2): 507-522. DOI: 10.1109/TNNLS.2020.2995800.

[11]

GAUR L, BHANDARI M, RAZDAN T, et al. Explanation-driven deep learning model for prediction of brain tumour status using MRI image data[J/OL]. Front Genet, 2022, 13: 822666 [2024-09-05]. https://pubmed.ncbi.nlm.nih.gov/35360838/. DOI: 10.3389/fgene.2022.822666.

[12]

MAHUM R, SHARAF M, HASSAN H, et al. A robust brain tumor detector using BiLSTM and mayfly optimization and multi-level thresholding[J/OL]. Biomedicines, 2023, 11(6): 1715 [2024-09-05]. https://pubmed.ncbi.nlm.nih.gov/37371810/. DOI: 10.3390/biomedicines11061715.

[13]

LU Z Y, ZHANG Y, LIU X C, et al. The value of multi-sequence MRI-based radiomics in differential diagnosis of meningioma[J]. Chin J Magn Reson Imaging, 2024, 15(5): 47-54. DOI: 10.12015/issn.1674-8034.2024.05.009.

[14]

LIU T, ZHOU Y M, LU Z Y, et al. Application of MRI radiomics model based on deep learning features in differential diagnosis for skull-based meningioma[J]. DOI: 10.3969/j.issn.1672-0512.2024.05.009.

[15]

PARK M J, KIM H S, JAHNG G H, et al. Semiquantitative assessment of intratumoral susceptibility signals using non-contrast-enhanced high-field high-resolution susceptibility-weighted imaging in patients with gliomas: comparison with MR perfusion imaging[J]. AJNR Am J Neuroradiol, 2009, 30(7): 1402-1408. DOI: 10.3174/ajnr.A1593.

[16]

CHEN X H, ZHANG R D, ZHOU Y S, et al. Multi-sequence MRI-based convolutional neural network predicts the methylation status of MGMT promoter in glioma[J]. Chin J Magn Reson Imaging, 2023, 14(8): 34-39, 78. DOI: 10.12015/issn.1674-8034.2023.08.005.

[17]

LI X, LU Y, XIONG J, et al. Presurgical differentiation between malignant hemangiopericytoma and angiomatous meningioma by a radiomics approach based on texture analysis[J]. J Neuroradiol, 2019, 46(5): 281-287. DOI: 10.1016/j.neurad.2019.05.013.

[18]

DONG D, FANG M J, TANG L, et al. Deep learning radiomic nomogram can predict the number of lymph node metastasis in locally advanced gastric cancer: an international multicenter study[J]. Ann Oncol. 2020, 31(7): 912-920. DOI: 10.1016/j.annonc.2020.04.003.

[19]

LI N, MO Y, HUANG C, et al. A clinical semantic and radiomics nomogram for predicting brain invasion in WHO grade II meningioma based on tumor and tumor-to-brain interface features[J/OL]. Front Oncol, 2021, 11: 752158 [2024-09-05]. https://pubmed.ncbi.nlm.nih.gov/34745982/. DOI: 10.3389/fonc.2021.752158.

[20]

YANG L, XU P, ZHANG Y, et al. A deep learning radiomics model may help to improve the prediction performance of preoperative grading in meningioma[J]. Neuroradiology, 2022, 64(7): 1373-1382. DOI: 10.1007/s00234-022-02894-0.

[21]

JOO L, PARK J E, PARK S Y, et al. Extensive peritumoral edema and brain-to-tumor interface MRI features enable prediction of brain invasion in meningioma: development and validation[J]. Neuro Oncol, 2021, 23(2): 324-333. DOI: 10.1093/neuonc/noaa190.

[22]

BI Y Z, BAI J, BAI P R, et al. Machine learning models based on radiomics in differentiating solitary fibrous tumor from angiomatous meningioma[J]. Chin J Magn Reson Imaging, 2023, 14(09): 50-55. DOI: 10.12015/issn.1674-8034.2023.09.009.

[23]

HAN T, LIU X W, JIANG J, et al. Differential diagnosis of angiomatous meningioma and atypical meningioma based on contrast enhanced T1-weighted images histogram analysis[J]. Chin J Magn Reson Imaging, 2024, 15(3): 37-42. DOI: 10.12015/issn.1674-8034.2024.03.007.

[24]

HE W, XIAO X, LI X, et al. Whole-tumor histogram analysis of apparent diffusion coefficient in differentiating intracranial solitary fibrous tumor/hemangiopericytoma from angiomatous meningioma[J]. Eur J Radiol, 2019, 112: 186-191. DOI: 10.1016/j.ejrad.2019.01.023.

[25]

ZHANG Y, SHANG L, CHEN C, et al. Machine-learning classifiers in discrimination of lesions located in the anterior skull base[J/OL]. Front Oncol, 2020, 10: 752 [2024-09-05]. https://pubmed.ncbi.nlm.nih.gov/32547944/. DOI: 10.3389/fonc.2020.00752.

[26]

AL-ZOGHBY A M, AL-AWADLY E M K, MOAWAD A, et al. Dual deep CNN for tumor brain classification[J/OL]. Diagnostics (Basel), 2023, 13(12): 2050 [2024-09-05]. https://pubmed.ncbi.nlm.nih.gov/37370945/. DOI: 10.3390/diagnostics13122050.

[27]

RASHEED Z, MA Y K, ULLAH I, et al. Brain tumor classification from MRI using image enhancement and convolutional neural network techniques[J/OL]. Brain Sci, 2023, 13(9): 1320 [2024-09-05]. https://pubmed.ncbi.nlm.nih.gov/37759920/. DOI: 10.3390/brainsci13091320.

[28]

ZHANG H, ZHANG S, SUN J W, et al. Research progress of deep learning brain tumor MRI image classification[J]. Chin J Magn Reson Imaging, 2023, 14(1): 166-171, 193. DOI: 10.12015/issn.1674-8034.2023.01.031.

[29]

ABDUSALOMOV A B, MUKHIDDINOV M, WHANGBO T K. Brain tumor detection based on deep learning approaches and magnetic resonance imaging[J/OL]. Cancers (Basel), 2023, 15(16): 4172 [2024-09-05]. https://pubmed.ncbi.nlm.nih.gov/37627200/. DOI: 10.3390/cancers15164172.

[30]

HUANG M, XIONG Z Y, ZHU J L. MRI brain tumor classification based on multi-scale residual network[J]. Chin J Magn Reson Imaging, 2023, 14(1): 124-129. DOI: 10.12015/issn.1674-8034.2023.01.022.

[31]

HU X Y, LIU Y, NI C X, et al. An aided diagnosis model of acoustic neuroma and meningioma in cerebellopontine angle based on deep learning[J]. Software engineering, 2023, 26(11): 20-24, 58. DOI: 10.19644/j.cnki.issn2096-1472.2023.011.005.

[32]

LI Z X, XIE C Q, LI S L, et al. The value of predicting the PR expression status of meningiomas based on MRI features[J]. Chin J Magn Reson Imaging, 2022, 13(7): 1-5. DOI: 10.12015/issn.1674-8034.2022.07.001.

[33]

LIU H, QIAN H, LI X, ZUO F, et al. Clinial features, individualized treatment and long-term surgical outcomes of skull base meningiomas with extracranial extensions[J/OL]. Front Oncol, 2020, 10: 1054 [2024-09-05]. https://pubmed.ncbi.nlm.nih.gov/32714869/. DOI: 10.3389/fonc.2020.01054.

[34]

TIAN X, WANG J, WEN Y, et al. Multi-attribute scientific documents retrieval and ranking model based on GBDT and LR[J]. Math Biosci Eng, 2022, 19(4): 3748-3766. DOI: 10.3934/mbe.2022172.

[35]

ZHOU T, YE X, LU H, et al. Dense Convolutional Network and Its Application in Medical Image Analysis[J/OL]. Biomed Res Int. 2022, 2022: 2384830 [2024-09-05]. https://pubmed.ncbi.nlm.nih.gov/35509707/. DOI: 10.1155/2022/2384830.

[36]

NOFALLAH S, MEHTA S, MERCAN E, et al. Machine learning techniques for mitoses classification[J/OL]. Comput Med Imaging Graph, 2021, 87: 101832 [2024-09-05]. https://pubmed.ncbi.nlm.nih.gov/33302246/. DOI: 10.1016/j.compmedimag.2020.101832.

[37]

LEE H, EUN Y, HWANg J Y, et al. Explainable deep learning algorithm for distinguishing incomplete Kawasaki disease by coronary artery lesions on echocardiographic imaging[J/OL]. Comput Methods Programs Biomed, 2022, 223: 106970 [2024-09-05]. https://pubmed.ncbi.nlm.nih.gov/35772231/. DOI: 10.1016/j.cmpb.2022.106970.

[38]

CHEN R J, LU M Y, WANG J, et al. Pathomic fusion: an integrated framework for fusing histopathology and genomic features for cancer diagnosis and prognosis[J]. IEEE Trans Med Imaging, 2022, 41(4): 757-770. DOI: 10.1109/TMI.2020.3021387.

[39]

HU Z Y, WEI W, HU W Z, et al. Prediction of IDH mutations in glioma based on MRI multiparametric image fusion and DenseNet network[J]. Chin J Magn Reson Imaging, 2023, 14(7): 10-17. DOI: 10.12015/issn.1674-8034.2023.07.003.

[40]

SUN K, ZHANG J, LIU Z, et al. A deep learning radiomics analysis for identifying sinus invasion in patients with meningioma before operation using tumor and peritumoral regions[J/OL]. Eur J Radiol, 2022, 149: 110187 [2024-09-05]. https://pubmed.ncbi.nlm.nih.gov/35183900/. DOI: 10.1016/j.ejrad.2022.110187.

[41]

YOO Y S, KIM D, YANG S, et al. Comparison of 2D, 2.5D, and 3D segmentation networks for maxillary sinuses and lesions in CBCT images[J/OL]. BMC Oral Health, 2023, 23(1): 866 [2024-09-05]. https://pubmed.ncbi.nlm.nih.gov/37964229/. DOI: 10.1186/s12903-023-03607-6.

PREV MRI deep learning study to predict progression-free survival in brain glioma

NEXT Comparative study of myocardial mechanical function changes before and after coronary artery bypass grafting using cardiac magnetic resonance feature tracking technique

LINK

Tel & Fax: +8610-67113815 E-mail: editor@cjmri.cn