Deep Learning Model for Intracranial Hemangiopericytoma and Meningioma Classification

Performance of deep learning models for automatic histopathological grading of meningiomas: a systematic review and meta-analysis

Background

Accurate preoperative grading of meningiomas is crucial for selecting the most suitable treatment strategies and predicting patient outcomes. Traditional MRI-based assessments are often insufficient to distinguish between low- and high-grade meningiomas reliably. Deep learning (DL) models have emerged as promising tools for automated histopathological grading using imaging data. This systematic review and meta-analysis aimed to comprehensively evaluate the diagnostic performance of deep learning (DL) models for meningioma grading.

Methods

This study was conducted in accordance with the PRISMA-DTA guidelines and was prospectively registered on the Open Science Framework. A systematic search of PubMed, Scopus, and Web of Science was performed up to March 2025. Studies using DL models to classify meningiomas based on imaging data were included. A random-effects meta-analysis was used to pool sensitivity, specificity, accuracy, and area under the receiver operating characteristic curve (AUC). A bivariate random-effects model was used to fit the summary receiver operating characteristic (SROC) curve. Study quality was assessed using the Newcastle-Ottawa Scale, and publication bias was evaluated using Egger's test.

Results

Twenty-seven studies involving 13,130 patients were included. The pooled sensitivity was 92.31% (95% CI: 92.1-92.52%), specificity 95.3% (95% CI: 95.11-95.48%), and accuracy 97.97% (95% CI: 97.35-97.98%), with an AUC of 0.97 (95% CI: 0.96-0.98). The bivariate SROC curve demonstrated excellent diagnostic performance, characterized by a relatively narrow 95% confidence interval despite moderate to high heterogeneity (I2 = 79.7%, p < 0.001).

Conclusion

DL models demonstrate high diagnostic accuracy for automatic meningioma grading and could serve as valuable clinical decision-support tools.

Systematic review registration

DOI: 10.17605/OSF.IO/RXEBM.

Deep learning radiomic nomogram outperforms the clinical model in distinguishing intracranial solitary fibrous tumors from angiomatous meningiomas and can predict patient prognosis

OBJECTIVES

To evaluate the value of a magnetic resonance imaging (MRI)-based deep learning radiomic nomogram (DLRN) for distinguishing intracranial solitary fibrous tumors (ISFTs) from angiomatous meningioma (AMs) and predicting overall survival (OS) for ISFT patients.

METHODS

In total, 1090 patients from Beijing Tiantan Hospital, Capital Medical University, and 131 from Lanzhou University Second Hospital were categorized as primary cohort (PC) and external validation cohort (EVC), respectively. An MRI-based DLRN was developed in PC to distinguish ISFTs from AMs. We validated the DLRN and compared it with a clinical model (CM) in EVC. In total, 149 ISFT patients were followed up. We carried out Cox regression analysis on DLRN score, clinical characteristics, and histological stratification. Besides, we evaluated the association between independent risk factors and OS in the follow-up patients using Kaplan-Meier curves.

RESULTS

The DLRN outperformed CM in distinguishing ISFTs from AMs (area under the curve [95% confidence interval (CI)]: 0.86 [0.84-0.88] for DLRN and 0.70 [0.67-0.72] for CM, p < 0.001) in EVC. Patients with high DLRN score [per 1 increase; hazard ratio (HR) 1.079, 95% CI: 1.009-1.147, p = 0.019] and subtotal resection (STR) [per 1 increase; HR 2.573, 95% CI: 1.337-4.932, p = 0.004] were associated with a shorter OS. A statistically significant difference in OS existed between the high and low DLRN score groups with a cutoff value of 12.19 (p < 0.001). There is also a difference in OS between total excision (GTR) and STR groups (p < 0.001).

CONCLUSION

The proposed DLRN outperforms the CM in distinguishing ISFTs from AMs and can predict OS for ISFT patients.

CLINICAL RELEVANCE STATEMENT

The proposed MRI-based deep learning radiomic nomogram outperforms the clinical model in distinguishing ISFTs from AMs and can predict OS of ISFT patients, which could guide the surgical strategy and predict prognosis for patients.

KEY POINTS

Distinguishing ISFTs from AMs based on conventional radiological signs is challenging. The DLRN outperformed the CM in our study. The DLRN can predict OS for ISFT patients.

Preoperative diagnosis of meningioma sinus invasion based on MRI radiomics and deep learning: a multicenter study

OBJECTIVE

Exploring the construction of a fusion model that combines radiomics and deep learning (DL) features is of great significance for the precise preoperative diagnosis of meningioma sinus invasion.

MATERIALS AND METHODS

This study retrospectively collected data from 601 patients with meningioma confirmed by surgical pathology. For each patient, 3948 radiomics features, 12,288 VGG features, 6144 ResNet features, and 3072 DenseNet features were extracted from MRI images. Thus, univariate logistic regression, correlation analysis, and the Boruta algorithm were applied for further feature dimension reduction, selecting radiomics and DL features highly associated with meningioma sinus invasion. Finally, diagnosis models were constructed using the random forest (RF) algorithm. Additionally, the diagnostic performance of different models was evaluated using receiver operating characteristic (ROC) curves, and AUC values of different models were compared using the DeLong test.

RESULTS

Ultimately, 21 features highly associated with meningioma sinus invasion were selected, including 6 radiomics features, 2 VGG features, 7 ResNet features, and 6 DenseNet features. Based on these features, five models were constructed: the radiomics model, VGG model, ResNet model, DenseNet model, and DL-radiomics (DLR) fusion model. This fusion model demonstrated superior diagnostic performance, with AUC values of 0.818, 0.814, and 0.769 in the training set, internal validation set, and independent external validation set, respectively. Furthermore, the results of the DeLong test indicated that there were significant differences between the fusion model and both the radiomics model and the VGG model (p < 0.05).

CONCLUSIONS

The fusion model combining radiomics and DL features exhibits superior diagnostic performance in preoperative diagnosis of meningioma sinus invasion. It is expected to become a powerful tool for clinical surgical plan selection and patient prognosis assessment.

Apparent Diffusion Coefficient (ADC) and Magnetic Resonance Imaging (MRI) Nomogram for Differentiating a Solitary Fibrous Tumor (World Health Organization Grade II) From an Angiomatous Meningioma

INTRODUCTION

 Accurate preoperative differentiation between intracranial solitary fibrous tumor (SFT, World Health Organization grade II) and angiomatous meningioma (AM) is crucial for surgical planning and prognosis prediction. While conventional magnetic resonance imaging (MRI) is widely used, distinguishing these tumors based on imaging alone remains challenging. This study aimed to evaluate clinical and MRI features to improve diagnostic accuracy between SFT and AM, focusing on the apparent diffusion coefficient (ADC) and conventional MRI parameters.

METHODS

 A retrospective analysis was conducted on 51 patients (23 with SFT and 28 with AM) confirmed by pathology. Clinical and MRI characteristics were assessed using t-tests and chi-square tests. Logistic regression analysis was performed to identify independent predictors, and receiver operating characteristic (ROC) curve analysis evaluated diagnostic performance. A nomogram integrating ADC values with conventional MRI features was developed and validated using calibration curves.

RESULTS

 Significant differences in tumor shape, cystic necrosis, T1-weighted imaging and T2-weighted imaging signal intensities, and ADC values were observed between SFT and AM (p < 0.05). Logistic regression analysis confirmed these factors as independent predictors, with ADC demonstrating the highest diagnostic performance at an optimal cutoff value of 1.08 × 10-³ mm²/second. The ROC analysis showed that combining ADC with conventional MRI features improved diagnostic accuracy. The calibration curve demonstrated strong agreement between nomogram predictions and actual outcomes.

CONCLUSION

 Integrating ADC values with clinical and MRI features provides a reliable method for differentiating intracranial SFT from AM. This approach enhances diagnostic precision, aiding in optimized clinical decision-making and surgical planning.

Unveiling Thymoma Typing Through Hyperspectral Imaging and Deep Learning

Thymoma, a rare tumor from thymic epithelial cells, presents diagnostic challenges because of the subjective nature of traditional methods, leading to high false-negative rates and long diagnosis times. This study introduces a thymoma classification technique that integrates hyperspectral imaging with deep learning. We initially capture pathological slice images of thymoma using a hyperspectral camera and delineate regions of interest to extract spectral data. This data undergoes reflectance calibration and noise reduction. Subsequently, we transform the spectral data into two-dimensional images via the Gramian Angular Field (GAF) method. A variant residual network is then utilized to extract features and classify these images. Our results demonstrate that this model significantly enhances classification accuracy and efficiency, achieving an average accuracy of 95%. The method proves highly effective in automated thymoma diagnosis, optimizing data utilization, and feature representation learning.

Cystic intracranial solitary fibrous tumor: a case report

Solitary fibrous tumor (SFT) is a rare spindle cell tumor originating from mesenchymal tissue, and even rarer when it occurs intracranially. This case report described a 42-year-old man who presented with headache and limb weakness for more than 10 days. Magnetic resonance imaging (MRI) showed a well-defined multicompartmental cystic space-occupying lesion in the left occipital region, with surrounding edema and a compressed left lateral ventricle, the mass growing across the cerebellar vermis, which was initially diagnosed as hemangioblastoma. Neurosurgery was utilized to successfully remove the mass, and intracranial solitary fibrous tumor (ISFT) was identified by postoperative pathological analysis. Here, this article describes the imaging manifestations and pathologic features of a case of cystic intracranial solitary fibrous tumor, aiming to improve the understanding and diagnosis of this disease in order to provide an accurate therapy plan.

The role of tumor parenchyma and brain cortex signal intensity ratio in differentiating solitary fibrous tumors and meningiomas

BACKGROUND

Solitary fibrous tumors (SFT) and meningiomas (MA) have similar clinical and radiographic presentations but require different treatment approaches and have different prognoses. This emphasizes the importance of a correct preoperative diagnosis of SFT versus MA.

OBJECTIVE

In this study, investigated the differences in imaging characteristics between SFT and MA to improve the accuracy of preoperative imaging diagnosis of SFT.

METHODS

The clinical and imaging data of 26 patients with SFT and 104 patients with MA who were pathologically diagnosed between August 2017 and December 2022, were retrospectively analyzed. The clinical and imaging differences between SFT and MA, as well as between the various pathological grades of SFT, were analyzed.

RESULTS

Age, gender, cystic change, flow void phenomenon, yin-yang sign, lobulation, narrow base, tumor/cortex signal ratio (TCSR) > 1.0 in T1-weighted imaging (T1WI), TCSR ≥ 1.1 in T2-weighted imaging (T2WI), peritumoral edema, and absence of dural tail sign varied between SFT and MA. As per the receiver operating characteristic (ROC) curve analysis, TCSR > 1 in T1WI has the maximum diagnostic accuracy for SFT. Cranial or venous sinus invasion had a positive effect on SFT (Grade III, World Health Organization (WHO) grading).

CONCLUSION

Among the many radiological and clinical distinctions between SFT and MA, TCSR ≥ 1 exhibits the highest predictive efficacy for SFT; while cranial or venous sinus invasion may be a predictor of WHO grade III SFT.

Solitary Fibrous Tumor of the Central Nervous System: A Report of Two Cases with Emphasis on Diagnostic Pitfalls

Solitary fibrous tumor (SFT) is a rare primary central nervous system neoplasm that usually presents as a dural-based mass. Awareness of the entity is limited by the rarity of the tumor which renders it prone to misdiagnosis. We present two cases of SFT located in the right parafalx and intraventricular region. The cases were classified as WHO grade 1 and grade 2, respectively. The present study discusses the radiological, histomorphological, and immunohistochemical features of SFT, with emphasis on potential diagnostic pitfalls that may lead to erroneous diagnosis.

Solitary fibrous tumor of the central nervous system invading and penetrating the skull: A case report

Solitary fibrous tumor (SFT) of the central nervous system is a rare spindle cell tumor of mesenchymal origin. The present study reports the case of a 44-year-old male patient with SFT. Magnetic resonance imaging demonstrated that the majority of the intracranial tumors exhibited uneven low signals on T1-weighted imaging (T1WI) and low mixed signals on T2WI, and there was an enhancement on enhanced scanning. Furthermore, the distal part of the left occipital lobe exhibited hypersignals on T1WI and T2WI, and this was significantly enhanced following enhanced scanning. The lower part of the scalp exhibited low signals on T1WI and high signals on T2WI, and there was no notable enhancement following enhanced scanning. Magnetic resonance spectroscopy demonstrated an elevated choline/creatine peak in the solid part of the tumor. Under the microscope, the tumor exhibited characteristic 'staghorn-shaped' blood vessels. As SFT is difficult to differentially diagnose via imaging, immunohistochemical analysis of CD34, vimentin and signal transducer and activator of transcription 6 was performed for the definitive diagnosis of SFT. Of note, surgical resection was the preferred treatment for SFT; however, due to the rarity of the tumor, subsequent adjuvant therapy and prognosis require further investigation.