Share：
Share this content in WeChat
X
[Chinese][PDF]331
• Clinical Article •
Correlation study of local habitat entropy based on multimodal MRI for predicting IDH molecular status in adult-type diffuse glioma
HU Mingxue  WANG Peng  LIU Yanhao  XIE Shenghui  HE Jinlong  WU Qiong  GAO Yang 

Cite this article as: HU M X, WANG P, LIU Y H, et al. Correlation study of local habitat entropy based on multimodal MRI for predicting IDH molecular status in adult-type diffuse glioma[J]. Chin J Magn Reson Imaging, 2025, 16(5): 120-126, 142. DOI:10.12015/issn.1674-8034.2025.05.019.


[Abstract] Objective To characterize the heterogeneity of adult-type diffuse gliomas using local habitat entropy based on multimodal magnetic resonance imaging and to develop and validate a comprehensive model for predicting isocitrate dehydrogenase molecular status.Materials and Methods A retrospective collection and analysis were performed on data obtained from the Affiliated Hospital of Inner Mongolia Medical University, the University of California, San Francisco and The Cancer Genome Atlas, encompassing a total of 533 subjects. Six types of conventional magnetic resonance images (T2, T1, T2-FLAIR, T1-CE, DWI, and ADC) were used for further image preprocessing. The preprocessing pipeline included N4 bias field correction, super-resolution reconstruction based on a migration model, isotropic resampling, and image normalization. An improved nn-Unet was employed to automatically segment tumor regions, followed by manual confirmation and correction. Habitat local entropy values were obtained for the entire lesion area, using a 3 × 3 × 3 matrix considering the size of the region of interest. During this process, global image discretization was performed according to the entire cohort, meaning the discretization histogram was based on the actual maximum and minimum values of the cohort, and finally adjusted to 32 Bins with equal interval Bin width. K-means was used to generate habitats based on T1-CE and T2-FLAIR matching, with the number of cluster centers ranging from 2 to 5. Then, 16 first-order features of different habitat subregions were extracted from all modalities. The UCSF public database served as the training set, and internal validation was performed using 10-fold cross-validation. The remaining two databases were used as independent test sets. A multi-pipeline approach (240 basic pipelines) was used to construct machine learning models. Feature selection and hyperparameter tuning were performed through cross-validation. The diagnostic performance of the models was evaluated using receiver operating characteristic (ROC) curves, and the DeLong test was used to compare model differences. The deviation between the model predictions and actual results was visualized using calibration curves. Decision curve analysis was employed to determine the clinical net benefit.Results When the number of cluster centers was set to 2 or 3, the corresponding Calinski-Harabasz indices were 95 080 and 100 379, respectively, the Silhouette coefficients were 0.477 and 0.422, and the Davies-Bouldin indices were 0.741 and 0.810. Since the results for cluster centers of 4 and 5 were suboptimal, subsequent analyses were conducted only for clusters 2 and 3. All three multimodal models (whole lesion area, cluster 2, and cluster 3) demonstrated excellent diagnostic performance, with AUC values ranging from 0.942 to 0.974 in the training set and from 0.739 to 0.864 in the test sets. Specifically, when the cluster number was 2, the sensitivity was higher in both independent test sets (95.2% and 80.0%, respectively). Conversely, when the cluster number was 3, the specificity was higher in both independent test sets (72.2% and 89.2%, respectively). The calibration curves and decision analysis curves for all three models indicated high and similar predictive consistency and clinical applicability.Conclusions Local habitat entropy based on multimodal MRI provides valuable information on the heterogeneity of adult-type diffuse gliomas. The combined application of local features and habitat analysis offers new insights and methods for the non-invasive assessment of various pathological abnormalities.
[Keywords] glioma；genotyping；isocitrate dehydrogenase；magnetic resonance imaging；unsupervised segmentation；radiomics；entropy

HU Mingxue   WANG Peng   LIU Yanhao   XIE Shenghui   HE Jinlong   WU Qiong   GAO Yang*  

Department of Radiology, Affiliated Hospital of Inner Mongolia Medical University, Hohhot 010059, China

Corresponding author: GAO Y, E-mail: 1390903990@qq.com

Conflicts of interest   None.

Received  2025-02-22
Accepted  2025-04-10
DOI: 10.12015/issn.1674-8034.2025.05.019
Cite this article as: HU M X, WANG P, LIU Y H, et al. Correlation study of local habitat entropy based on multimodal MRI for predicting IDH molecular status in adult-type diffuse glioma[J]. Chin J Magn Reson Imaging, 2025, 16(5): 120-126, 142. DOI:10.12015/issn.1674-8034.2025.05.019.

[1]
WELLER M, WEN P Y, CHANG S M, et al. Glioma[J/OL]. Nat Rev Dis Primers, 2024, 10(1): 33 [2025-02-20]. https://doi.org/10.1038/s41572-024-00516-y. DOI: 10.1038/s41572-024-00516-y.
[2]
WEN P Y, WELLER M, LEE E Q, et al. Glioblastoma in adults: a Society for Neuro-Oncology (SNO) and European Society of Neuro-Oncology (EANO) consensus review on current management and future directions[J]. Neuro Oncol, 2020, 22(8): 1073-1113. DOI: 10.1093/neuonc/noaa106.
[3]
LING A L, SOLOMON I H, LANDIVAR A M, et al. Clinical trial links oncolytic immunoactivation to survival in glioblastoma[J]. Nature, 2023, 623(7985): 157-166. DOI: 10.1038/s41586-023-06623-2.
[4]
MILLER J J, GONZALEZ CASTRO L N, MCBRAYER S, et al. Isocitrate dehydrogenase (IDH) mutant gliomas: a Society for Neuro-Oncology (SNO) consensus review on diagnosis, management, and future directions[J]. Neuro Oncol, 2023, 25(1): 4-25. DOI: 10.1093/neuonc/noac207.
[5]
LU J, XU W, CHEN X, et al. Noninvasive prediction of IDH mutation status in gliomas using preoperative multiparametric MRI radiomics nomogram: A mutlicenter study[J]. Magn Reson Imaging, 2023, 104: 72-79. DOI: 10.1016/j.mri.2023.09.001.
[6]
WANG P, XIE S, WU Q, et al. Model incorporating multiple diffusion MRI features: development and validation of a radiomics-based model to predict adult-type diffuse gliomas grade[J]. Eur Radiol, 2023, 33(12): 8809-8820. DOI: 10.1007/s00330-023-09861-0.
[7]
TIAN H, WU H, WU G, et al. Noninvasive prediction of TERT promoter mutations in high‐grade glioma by radiomics analysis based on multiparameter MRI[J/OL]. Biomed Res Int, 2020, 2020(1): 3872314 [2025-02-20]. https://doi.org/10.1155/2020/3872314. DOI: 10.1155/2020/3872314.
[8]
WANG R, GAO Y. Research progress of multimodal MRI radiomics and deep learning in glioma[J]. Chin J Magn Reson Imaging, 2024, 15(7): 165-172. DOI: 10.12015/issn.1674-8034.2024.07.028.
[9]
SINGH G, MANJILA S, SAKLA N, et al. Radiomics and radiogenomics in gliomas: a contemporary update[J]. Br J Cancer, 2021, 125(5): 641-657. DOI: 10.1038/s41416-021-01387-w.
[10]
WANG H, ZHANG S, XING X, et al. Radiomic study on preoperative multi‐modal magnetic resonance images identifies IDH‐mutant TERT promoter-mutant gliomas[J]. Cancer Med, 2023, 12(3): 2524-2537. DOI: 10.1002/cam4.5097.
[11]
MATHUR R, WANG Q, SCHUPP P G, et al. Glioblastoma evolution and heterogeneity from a 3D whole-tumor perspective[J/OL]. Cell, 2024, 187(2): 446-463. e16 [2025-02-20]. https://doi.org/10.1016/j.cell.2023.12.013. DOI: 10.1016/j.cell.2023.12.013.
[12]
DERCLE L, LU L, SCHWARTZ L H, et al. Radiomics response signature for identification of metastatic colorectal cancer sensitive to therapies targeting EGFR pathway[J]. J Natl Cancer Inst, 2020, 112(9): 902-912. DOI: 10.1093/jnci/djaa017.
[13]
XIE C, YANG P, ZHANG X, et al. Sub-region based radiomics analysis for survival prediction in oesophageal tumours treated by definitive concurrent chemoradiotherapy[J]. EBioMedicine, 2019, 44: 289-297. DOI: 10.1016/j.ebiom.2019.05.023.
[14]
SUO S, ZHANG D, CHENG F, et al. Added value of mean and entropy of apparent diffusion coefficient values for evaluating histologic phenotypes of invasive ductal breast cancer with MR imaging[J]. Eur Radiol, 2019, 29: 1425-1434. DOI: 10.1007/s00330-018-5667-9.
[15]
AHN S Y, PARK C M, PARK S J, et al. Prognostic value of computed tomography texture features in non-small cell lung cancers treated with definitive concomitant chemoradiotherapy[J]. Invest Radiol, 2015, 50(10): 719-725. DOI: 10.1097/RLI.0000000000000174.
[16]
LIU Y H, GAO Y. Implications of habitat imaging-based multisequence MRI in adult-type diffuse glioma[J]. Chin J Magn Reson Imaging, 2023, 14(9): 119-124. DOI: 10.12015/issn.1674-8034.2023.09.022.
[17]
LIU Y, WANG P, WANG S, et al. Heterogeneity matching and IDH prediction in adult-type diffuse gliomas: a DKI-based habitat analysis[J/OL]. Front Oncol, 2023, 13: 1202170 [2025-02-20]. https://doi.org/10.3389/fonc.2023.1202170. DOI: 10.3389/fonc.2023.1202170.
[18]
KIM M, PARK J E, KIM H S, et al. Spatiotemporal habitats from multiparametric physiologic MRI distinguish tumor progression from treatment-related change in post-treatment glioblastoma[J]. Eur Radiol, 2021, 31: 6374-6383. DOI: 10.1007/s00330-021-07718-y.
[19]
ZHANG C, WANG P, HE J, et al. Super-resolution reconstruction improves multishell diffusion: using radiomics to predict adult-type diffuse glioma IDH and grade[J/OL]. Front Oncol, 2024, 14: 1435204 [2025-02-20]. https://doi.org/10.3389/fonc.2024.1435204. DOI: 10.3389/fonc.2024.1435204.
[20]
LUU H M, PARK S H. Extending nn-UNet for brain tumor segmentation[C]//International MICCAI brainlesion workshop. Cham: Springer International Publishing, 2021: 173-186. DOI: 10.1007/978-3-031-09002-8_16.
[21]
ZHAO W, HU Z, KAZEROONI A F, et al. Physics-Informed Discretization for Reproducible and Robust Radiomic Feature Extraction Using Quantitative MRI[J]. Invest Radiol, 2024, 59(5): 359-371. DOI: 10.1097/RLI.0000000000001026.
[22]
TROJANI V, BASSI M C, VERZELLESI L, et al. Impact of Preprocessing parameters in Medical Imaging-based Radiomic studies: a systematic review[J/OL]. Cancers, 2024, 16(15): 2668 [2025-02-20]. https://doi.org/10.3390/cancers16152668. DOI: 10.3390/cancers16152668.
[23]
XU H, LV W, ZHANG H, et al. Evaluation and optimization of radiomics features stability to respiratory motion in 18F‐FDG 3D PET imaging[J]. Med Phys, 2021, 48(9): 5165-5178. DOI: 10.1002/mp.15022.
[24]
LEE J Y, LEE K, SEO B K, et al. Radiomic machine learning for predicting prognostic biomarkers and molecular subtypes of breast cancer using tumor heterogeneity and angiogenesis properties on MRI[J]. Eur Radiol, 2022, 32: 650-660. DOI: 10.1007/s00330-021-08146-8.
[25]
CHOE J, LEE S M, DO K H, et al. Prognostic value of radiomic analysis of iodine overlay maps from dual-energy computed tomography in patients with resectable lung cancer[J]. Eur Radiol, 2019, 29: 915-923. DOI: 10.1007/s00330-018-5639-0.
[26]
GAO Y, CHENG S, ZHU L, et al. A systematic review of prognosis predictive role of radiomics in pancreatic cancer: heterogeneity markers or statistical tricks?[J]. Eur Radiol, 2022, 32(12): 8443-8452. DOI: 10.1007/s00330-022-08922-0.
[27]
ZHANG H, ZHANG H, ZHANG Y, et al. Deep learning radiomics for the assessment of telomerase reverse transcriptase promoter mutation status in patients with glioblastoma using multiparametric MRI[J]. J Magn Reson Imaging, 2023, 58(5): 1441-1451. DOI: 10.1002/jmri.28671.
[28]
ZHANG H, OUYANG Y, ZHANG Y, et al. Sub-region based radiomics analysis for prediction of isocitrate dehydrogenase and telomerase reverse transcriptase promoter mutations in diffuse gliomas[J/OL]. Clin Radiol, 2024, 79(5): e682-e691 [2025-02-20]. https://doi.org/10.1016/j.crad.2024.01.030. DOI: 10.1016/j.crad.2024.01.030.
[29]
PEI D, GUAN F, HONG X, et al. Radiomic features from dynamic susceptibility contrast perfusion-weighted imaging improve the three-class prediction of molecular subtypes in patients with adult diffuse gliomas[J]. Eur Radiol, 2023, 33(5): 3455-3466. DOI: 10.1007/s00330-023-09459-6.
[30]
VERMA R, CORREA R, HILL V B, et al. Tumor habitat–derived radiomic features at pretreatment MRI that are prognostic for progression-free survival in glioblastoma are associated with key morphologic attributes at histopathologic examination: A feasibility study[J/OL]. Radiol Artif Intell, 2020, 2(6): e190168 [2025-02-20]. https://doi.org/10.1148/ryai.2020190168. DOI: 10.1148/ryai.2020190168.
[31]
NI J, ZHANG H, YANG Q, et al. Machine-Learning and Radiomics-Based Preoperative Prediction of Ki-67 Expression in Glioma Using MRI Data[J]. Acad Radiol, 2024, 31(8): 3397-3405. DOI: 10.1016/j.acra.2024.02.009.
[32]
KASAP D N G, MORA N G N, BLÖMER D A, et al. Comparison of MRI Sequences to Predict IDH Mutation Status in Gliomas Using Radiomics-Based Machine Learning[J/OL]. Biomedicines, 2024, 12(4): 725 [2025-02-20]. https://doi.org/10.3390/biomedicines12040725. DOI: 10.3390/biomedicines12040725.
[33]
BOBHOLZ S A, LOWMAN A K, BARRINGTON A, et al. Radiomic features of multiparametric MRI present stable associations with analogous histological features in patients with brain cancer[J/OL]. Tomography, 2020, 6(2): 160 [2025-02-20]. https://doi.org/10.18383/j.tom.2019.00029. DOI: 10.18383/j.tom.2019.00029.
[34]
WEI R, LU S, LAI S, et al. A subregion-based RadioFusionOmics model discriminates between grade 4 astrocytoma and glioblastoma on multisequence MRI[J/OL]. J Cancer Res Clin Oncol, 2024, 150(2): 73 [2025-02-20]. https://doi.org/10.1007/s00432-023-05603-3. DOI: 10.1007/s00432-023-05603-3.

PREV Three-dimensional pseudo-continuous arterial spin labeling reveals cerebral perfusion abnormalities in children with global developmental delay
NEXT Differential diagnosis of autoimmune encephalitis and herpes simplex virus encephalitis using radiomics models based on multimodal MRI
  


LINK


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