Abstract
Spatial intratumoral heterogeneity poses a significant challenge for accurate response assessment in glioblastoma. Multimodal imaging coupled with advanced image analysis has the potential to unravel this response heterogeneity. Based on automated tumor segmentation and longitudinal registration with follow-up imaging, we categorized contrast-enhancing voxels of 61 patients with suspected recurrence of glioblastoma into either true tumor progression (TP) or pseudoprogression (PsP). To allow the unbiased analysis of semantically related image regions, adjacent voxels with similar values of cerebral blood volume (CBV), FET-PET, and contrast-enhanced T1w were automatically grouped into supervoxels. We then extracted first-order statistics as well as texture features from each supervoxel. With these features, a Random Forest classifier was trained and validated employing a 10-fold cross-validation scheme. For model evaluation, the area under the receiver operating curve, as well as classification performance metrics were calculated. Our image analysis pipeline enabled reliable spatial assessment of tumor response. The predictive model reached an accuracy of 80.0% and a macro-weighted AUC of 0.875, which takes class imbalance into account, in the hold-out samples from cross-validation on supervoxel level. Analysis of feature importances confirmed the significant role of FET-PET-derived features. Accordingly, TP- and PsP-labeled supervoxels differed significantly in their 10th and 90th percentile, as well as the median of tumor-to-background normalized FET-PET. However, CBV- and T1c-related features also relevantly contributed to the model's performance. Disentangling the intratumoral heterogeneity in glioblastoma holds immense promise for advancing precise local response evaluation and thereby also informing more personalized and localized treatment strategies in the future.
Overview
- The study aims to develop an image analysis pipeline for accurately assessing response to treatment in glioblastoma patients using multimodal imaging and advanced image analysis techniques. The study uses automated tumor segmentation and longitudinal registration with follow-up imaging to categorize contrast-enhancing voxels into true tumor progression (TP) or pseudoprogression (PsP). The study then extracts first-order statistics and texture features from supervoxels and trains a Random Forest classifier to predict TP and PsP. The model achieves an accuracy of 80.0% and a macro-weighted AUC of 0.875 in the hold-out samples from cross-validation on supervoxel level. The study's findings suggest that FET-PET-derived features play a significant role in predicting tumor response, while CBV and T1c-related features also contribute to the model's performance. The study's findings have the potential to inform more personalized and localized treatment strategies in the future.
Comparative Analysis & Findings
- The study compares the outcomes observed under different experimental conditions or interventions, specifically the use of multimodal imaging and advanced image analysis techniques to assess response to treatment in glioblastoma patients. The study identifies significant differences in the results between true tumor progression (TP) and pseudoprogression (PsP) labeled supervoxels, with TP- and PsP-labeled supervoxels differing significantly in their 10th and 90th percentile, as well as the median of tumor-to-background normalized FET-PET. The study also finds that CBV and T1c-related features also relevantly contribute to the model's performance. The key findings of the study suggest that disentangling the intratumoral heterogeneity in glioblastoma holds immense promise for advancing precise local response evaluation and informing more personalized and localized treatment strategies in the future.
Implications and Future Directions
- The study's findings have significant implications for the field of research and clinical practice, as they provide a reliable and accurate method for assessing response to treatment in glioblastoma patients using multimodal imaging and advanced image analysis techniques. The study identifies limitations, such as the need for larger sample sizes and the need to validate the model on independent datasets. Future research directions could include exploring the use of other imaging modalities, such as magnetic resonance spectroscopy (MRS), and incorporating clinical and genetic data to further improve the predictive model. The study's findings also highlight the importance of personalized and localized treatment strategies in glioblastoma, and the need for further research to develop more effective and targeted treatments.