A deep learning model for differentiating paediatric intracranial germ cell tumour subtypes and predicting survival with MRI: a multicentre prospective study.

BACKGROUND: The pretherapeutic differentiation of subtypes of primary intracranial germ cell tumours (iGCTs), including germinomas (GEs) and nongerminomatous germ cell tumours (NGGCTs), is essential for clinical practice because of distinct treatment strategies and prognostic profiles of these diseases. This study aimed to develop a deep learning model, iGNet, to assist in the differentiation and prognostication of iGCT subtypes by employing pretherapeutic MR T2-weighted imaging. METHODS: The iGNet model, which is based on the nnUNet architecture, was developed using a retrospective dataset of 280 pathologically confirmed iGCT patients. The training dataset included 83 GEs and 117 NGGCTs, while the retrospective internal test dataset included 31 GEs and 49 NGGCTs. The model's diagnostic performance was then assessed with the area under the receiver operating characteristic curve (AUC) in a prospective internal dataset (n = 22) and two external datasets (n = 22 and 20). Next, we compared the diagnostic performance of six neuroradiologists with or without the assistance of iGNet. Finally, the predictive ability of the output of iGNet for progression-free and overall survival was assessed and compared to that of the pathological diagnosis. RESULTS: iGNet achieved high diagnostic performance, with AUCs between 0.869 and 0.950 across the four test datasets. With the assistance of iGNet, the six neuroradiologists' diagnostic AUCs (averages of the four test datasets) increased by 9.22% to 17.90%. There was no significant difference between the output of iGNet and the results of pathological diagnosis in predicting progression-free and overall survival (P = .889). CONCLUSIONS: By leveraging pretherapeutic MR imaging data, iGNet accurately differentiates iGCT subtypes, facilitating prognostic evaluation and increasing the potential for tailored treatment.

A Fully Automated Deep-Learning Model for Predicting the Molecular Subtypes of Posterior Fossa Ependymomas Using T2-Weighted Images.

PURPOSE: We aimed to develop and validate a deep learning (DL) model to automatically segment posterior fossa ependymoma (PF-EPN) and predict its molecular subtypes [Group A (PFA) and Group B (PFB)] from preoperative MR images. EXPERIMENTAL DESIGN: We retrospectively identified 227 PF-EPNs (development and internal test sets) with available preoperative T2-weighted (T2w) MR images and molecular status to develop and test a 3D nnU-Net (referred to as T2-nnU-Net) for tumor segmentation and molecular subtype prediction. The network was externally tested using an external independent set [n = 40; subset-1 (n = 31) and subset-2 (n =9)] and prospectively enrolled cases [prospective validation set (n = 27)]. The Dice similarity coefficient was used to evaluate the segmentation performance. Receiver operating characteristic analysis for molecular subtype prediction was performed. RESULTS: For tumor segmentation, the T2-nnU-Net achieved a Dice score of 0.94 ± 0.02 in the internal test set. For molecular subtype prediction, the T2-nnU-Net achieved an AUC of 0.93 and accuracy of 0.89 in the internal test set, an AUC of 0.99 and accuracy of 0.93 in the external test set. In the prospective validation set, the model achieved an AUC of 0.93 and an accuracy of 0.89. The predictive performance of T2-nnU-Net was superior or comparable to that of demographic and multiple radiologic features (AUCs ranging from 0.87 to 0.95). CONCLUSIONS: A fully automated DL model was developed and validated to accurately segment PF-EPNs and predict molecular subtypes using only T2w MR images, which could help in clinical decision-making.

Automated Determination of the H3 K27-Altered Status in Spinal Cord Diffuse Midline Glioma by Radiomics Based on T2-Weighted MR Images.

BACKGROUND AND PURPOSE: Conventional MR imaging is not sufficient to discern the H3 K27-altered status of spinal cord diffuse midline glioma. This study aimed to develop a radiomics-based model based on preoperative T2WI to determine the H3 K27-altered status of spinal cord diffuse midline glioma. MATERIALS AND METHODS: Ninety-seven patients with confirmed spinal cord diffuse midline gliomas were retrospectively recruited and randomly assigned to the training (n = 67) and test (n = 30) sets. One hundred seven radiomics features were initially extracted from automatically-segmented tumors on T2WI, then 11 features selected by the Pearson correlation coefficient and the Kruskal-Wallis test were used to train and test a logistic regression model for predicting the H3 K27-altered status. Sensitivity analysis was performed using additional random splits of the training and test sets, as well as applying other classifiers for comparison. The performance of the model was evaluated through its accuracy, sensitivity, specificity, and area under the curve. Finally, a prospective set including 28 patients with spinal cord diffuse midline gliomas was used to validate the logistic regression model independently. RESULTS: The logistic regression model accurately predicted the H3 K27-altered status with accuracies of 0.833 and 0.786, sensitivities of 0.813 and 0.750, specificities of 0.857 and 0.833, and areas under the curve of 0.839 and 0.818 in the test and prospective sets, respectively. Sensitivity analysis confirmed the robustness of the model, with predictive accuracies of 0.767-0.833. CONCLUSIONS: Radiomics signatures based on preoperative T2WI could accurately predict the H3 K27-altered status of spinal cord diffuse midline glioma, providing potential benefits for clinical management.

"Soap bubble" sign as an imaging marker for posterior fossa ependymoma Group B.

PURPOSE: To investigate the predictive value of the "soap bubble" sign on molecular subtypes (Group A [PFA] and Group B [PFB]) of posterior fossa ependymomas (PF-EPNs). METHODS: MRI scans of 227 PF-EPNs (internal retrospective discovery set) were evaluated by two independent neuroradiologists to assess the "soap bubble" sign, which was defined as clusters of cysts of various sizes that look like "soap bubbles" on T2-weighted images. Two independent cohorts (external validation set [n = 31] and prospective validation set [n = 27]) were collected to validate the "soap bubble" sign. RESULTS: Across three datasets, the "soap bubble" sign was observed in 21 PFB cases (7.4% [21/285] of PF-EPNs and 12.9% [21/163] of PFB); none in PFA. Analysis of the internal retrospective discovery set demonstrated substantial interrater agreement (1st Rating: κ = 0.71 [0.53-0.90], 2nd Rating: κ = 0.83 [0.68-0.98]) and intrarater agreement (Rater 1: κ = 0.73 [0.55-0.91], Rater 2: κ = 0.74 [0.55-0.92]) for the "soap bubble" sign; all 13 cases positive for the "soap bubble" sign were PFB (p = 0.002; positive predictive value [PPV] = 100%, negative predictive value [NPV] = 44%, sensitivity = 10%, specificity = 100%). The findings from the external validation set and the prospective validation set were similar, all cases positive for the "soap bubble" sign were PFB (p < 0.001; PPV = 100%). CONCLUSION: The "soap bubble" sign represents a highly specific imaging marker for the PFB molecular subtype of PF-EPNs.

Amide proton transfer-weighted imaging and derived radiomics in the classification of adult-type diffuse gliomas.

OBJECTIVES: To evaluate the utility of amide proton transfer-weighted (APTw) MRI imaging and its derived radiomics in classifying adult-type diffuse glioma. MATERIALS AND METHODS: In this prospective study, APTw imaging was performed on 129 patients with adult-type diffuse gliomas. The mean APTw-related metrics (chemical exchange saturation transfer ratio (CESTR), CESTR normalized with the reference value (CESTRnr), and relaxation-compensated inverse magnetization transfer ratio (MTRRex)) and radiomic features within 3D tumor masks were extracted. APTw-radiomics models were developed using a support vector machine (SVM) classifier. Sensitivity analysis with tumor area of interest, different histogram cutoff values, and other classifiers were conducted. RESULTS: CESTR, CESTRnr, and MTRRex in glioblastomas were all significantly higher (p < 0.0003) than those of oligodendrogliomas and astrocytomas, with no significant difference between oligodendrogliomas and astrocytomas. The APTw-related metrics for IDH-wildtype and high-grade gliomas were significantly higher (p < 0.001) than those for the IDH-mutant and low-grade gliomas, with area under the curve (AUCs) of 0.88 for CESTR. The CESTR-radiomics models demonstrated accuracies of 84% (AUC 0.87), 83% (AUC 0.83), 90% (AUC 0.95), and 84% (AUC 0.86) in predicting the IDH mutation status, differentiating glioblastomas from astrocytomas, distinguishing glioblastomas from oligodendrogliomas, and determining high/low grade prediction, respectively, but showed poor performance in distinguishing oligodendrogliomas from astrocytomas (accuracy 63%, AUC 0.63). The sensitivity analysis affirmed the robustness of the APTw signal and APTw-derived radiomics prediction models. CONCLUSION: APTw imaging, along with its derived radiomics, presents a promising quantitative approach for prediction IDH mutation and grading adult-type diffuse glioma. CLINICAL RELEVANCE STATEMENT: Amide proton transfer-weighted imaging, a quantitative imaging biomarker, coupled with its derived radiomics, offers a promising non-invasive approach for predicting IDH mutation status and grading adult-type diffuse gliomas, thereby informing individualized clinical diagnostics and treatment strategies. KEY POINTS: • This study evaluates the differences of different amide proton transfer-weighted metrics across three molecular subtypes and their efficacy in classifying adult-type diffuse glioma. • Chemical exchange saturation transfer ratio normalized with the reference value and relaxation-compensated inverse magnetization transfer ratio effectively predicts IDH mutation/grading, notably the first one. • Amide proton transfer-weighted imaging and its derived radiomics holds potential to be used as a diagnostic tool in routine clinical characterizing adult-type diffuse glioma.

Deep Learning for Noninvasive Assessment of H3 K27M Mutation Status in Diffuse Midline Gliomas Using MR Imaging.

BACKGROUND: Determination of H3 K27M mutation in diffuse midline glioma (DMG) is key for prognostic assessment and stratifying patient subgroups for clinical trials. MRI can noninvasively depict morphological and metabolic characteristics of H3 K27M mutant DMG. PURPOSE: This study aimed to develop a deep learning (DL) approach to noninvasively predict H3 K27M mutation in DMG using T2-weighted images. STUDY TYPE: Retrospective and prospective. POPULATION: For diffuse midline brain gliomas, 341 patients from Center-1 (27 ± 19 years, 184 males), 42 patients from Center-2 (33 ± 19 years, 27 males) and 35 patients (37 ± 18 years, 24 males). For diffuse spinal cord gliomas, 133 patients from Center-1 (30 ± 15 years, 80 males). FIELD STRENGTH/SEQUENCE: 5T and 3T, T2-weighted turbo spin echo imaging. ASSESSMENT: Conventional radiological features were independently reviewed by two neuroradiologists. H3 K27M status was determined by histopathological examination. The Dice coefficient was used to evaluate segmentation performance. Classification performance was evaluated using accuracy, sensitivity, specificity, and area under the curve. STATISTICAL TESTS: Pearson's Chi-squared test, Fisher's exact test, two-sample Student's t-test and Mann-Whitney U test. A two-sided P value <0.05 was considered statistically significant. RESULTS: In the testing cohort, Dice coefficients of tumor segmentation using DL were 0.87 for diffuse midline brain and 0.81 for spinal cord gliomas. In the internal prospective testing dataset, the predictive accuracies, sensitivities, and specificities of H3 K27M mutation status were 92.1%, 98.2%, 82.9% in diffuse midline brain gliomas and 85.4%, 88.9%, 82.6% in spinal cord gliomas. Furthermore, this study showed that the performance generalizes to external institutions, with predictive accuracies of 85.7%-90.5%, sensitivities of 90.9%-96.0%, and specificities of 82.4%-83.3%. DATA CONCLUSION: In this study, an automatic DL framework was developed and validated for accurately predicting H3 K27M mutation using T2-weighted images, which could contribute to the noninvasive determination of H3 K27M status for clinical decision-making. EVIDENCE LEVEL: 2 Technical Efficacy: Stage 2.