Exploring the Trade-Off between generalist and specialized Models: A center-based comparative analysis for glioblastoma segmentation.

INTRODUCTION: Inherent variations between inter-center data can undermine the robustness of segmentation models when applied at a specific center (dataset shift). We investigated whether specialized center-specific models are more effective compared to generalist models based on multi-center data, and how center-specific data could enhance the performance of generalist models within a particular center using a fine-tuning transfer learning approach. For this purpose, we studied the dataset shift at center level and conducted a comparative analysis to assess the impact of data source on glioblastoma segmentation models. METHODS & MATERIALS: The three key components of dataset shift were studied: prior probability shift-variations in tumor size or tissue distribution among centers; covariate shift-inter-center MRI alterations; and concept shift-different criteria for tumor segmentation. BraTS 2021 dataset was used, which includes 1251 cases from 23 centers. Thereafter, 155 deep-learning models were developed and compared, including 1) generalist models trained with multi-center data, 2) specialized models using only center-specific data, and 3) fine-tuned generalist models using center-specific data. RESULTS: The three key components of dataset shift were characterized. The amount of covariate shift was substantial, indicating large variations in MR imaging between different centers. Glioblastoma segmentation models tend to perform best when using data from the application center. Generalist models, trained with over 700 samples, achieved a median Dice score of 88.98%. Specialized models surpassed this with 200 cases, while fine-tuned models outperformed with 50 cases. CONCLUSIONS: The influence of dataset shift on model performance is evident. Fine-tuned and specialized models, utilizing data from the evaluated center, outperform generalist models, which rely on data from other centers. These approaches could encourage medical centers to develop customized models for their local use, enhancing the accuracy and reliability of glioblastoma segmentation in a context where dataset shift is inevitable.

The prognostic relevance of a gene expression signature in MRI-defined highly vascularized glioblastoma.

Background: The vascular heterogeneity of glioblastomas (GB) remains an important area of research, since tumor progression and patient prognosis are closely tied to this feature. With this study, we aim to identify gene expression profiles associated with MRI-defined tumor vascularity and to investigate its relationship with patient prognosis. Methods: The study employed MRI parameters calculated with DSC Perfusion Quantification of ONCOhabitats glioma analysis software and RNA-seq data from the TCGA-GBM project dataset. In our study, we had a total of 147 RNA-seq samples, which 15 of them also had MRI parameter information. We analyzed the gene expression profiles associated with MRI-defined tumor vascularity using differential gene expression analysis and performed Log-rank tests to assess the correlation between the identified genes and patient prognosis. Results: The findings of our research reveal a set of 21 overexpressed genes associated with the high vascularity pattern. Notably, several of these overexpressed genes have been previously implicated in worse prognosis based on existing literature. Our log-rank test further validates that the collective upregulation of these genes is indeed correlated with an unfavorable prognosis. This set of genes includes a variety of molecules, such as cytokines, receptors, ligands, and other molecules with diverse functions. Conclusions: Our findings suggest that the set of 21 overexpressed genes in the High Vascularity group could potentially serve as prognostic markers for GB patients. These results highlight the importance of further investigating the relationship between the molecules such as cytokines or receptors underlying the vascularity in GB and its observation through MRI and developing targeted therapies for this aggressive disease.