Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Classifying Histopathologic Glioblastoma Sub-regions with EfficientNet

Sanyukta Adap, Ujjwal Baid, Spyridon Bakas

TL;DR

This work tackles the heterogeneity of glioblastoma by classifying six histopathologic GBM sub-regions in H&E-stained slides using a four-step deep-learning pipeline built on EfficientNet variants. The BraTS-Path dataset, with large-scale 512×512 RGB image collections across six regions, supports transfer learning from ImageNet and a 5-fold cross-validated, ensembling approach to improve robustness. The EfficientNet-B1 variant demonstrated strong generalization, achieving a hold-out validation F1 of 0.546 and final testing F1 of 0.517, though a gap remains between training and unseen data. The study contributes a publicly available codebase and highlights the need for diverse, regularized training to translate these methods into clinically reliable GBM diagnostics.

Abstract

Glioblastoma (GBM) is the most common aggressive, fast-growing brain tumor, with a grim prognosis. Despite clinical diagnostic advancements, there have not been any substantial improvements to patient prognosis. Histopathological assessment of excised tumors is the first line of clinical diagnostic routine. We hypothesize that automated, robust, and accurate identification of distinct histological sub-regions within GBM could contribute to morphologically understanding this disease at scale. In this study, we designed a four-step deep learning approach to classify six (6) histopathological regions and quantitatively evaluated it on the BraTS-Path 2024 challenge dataset, which includes digitized Hematoxylin \& Eosin (H\&E) stained GBM tissue sections annotated for six distinct regions. We used the challenge's publicly available training dataset to develop and evaluate the effectiveness of several variants of EfficientNet architectures (i.e., B0, B1, B2, B3, B4). EfficientNet-B1 and EfficientNet-B4 achieved the best performance, achieving an F1 score of 0.98 in a 5-fold cross-validation configuration using the BraTS-Path training set. The quantitative performance evaluation of our proposed approach with EfficientNet-B1 on the BraTS-Path hold-out validation data and the final hidden testing data yielded F1 scores of 0.546 and 0.517, respectively, for the associated 6-class classification task. The difference in the performance on training, validation, and testing data highlights the challenge of developing models that generalize well to new data, which is crucial for clinical applications. The source code of the proposed approach can be found at the GitHub repository of Indiana University Division of Computational Pathology: https://github.com/IUCompPath/brats-path-2024-enet.

Classifying Histopathologic Glioblastoma Sub-regions with EfficientNet

TL;DR

This work tackles the heterogeneity of glioblastoma by classifying six histopathologic GBM sub-regions in H&E-stained slides using a four-step deep-learning pipeline built on EfficientNet variants. The BraTS-Path dataset, with large-scale 512×512 RGB image collections across six regions, supports transfer learning from ImageNet and a 5-fold cross-validated, ensembling approach to improve robustness. The EfficientNet-B1 variant demonstrated strong generalization, achieving a hold-out validation F1 of 0.546 and final testing F1 of 0.517, though a gap remains between training and unseen data. The study contributes a publicly available codebase and highlights the need for diverse, regularized training to translate these methods into clinically reliable GBM diagnostics.

Abstract

Glioblastoma (GBM) is the most common aggressive, fast-growing brain tumor, with a grim prognosis. Despite clinical diagnostic advancements, there have not been any substantial improvements to patient prognosis. Histopathological assessment of excised tumors is the first line of clinical diagnostic routine. We hypothesize that automated, robust, and accurate identification of distinct histological sub-regions within GBM could contribute to morphologically understanding this disease at scale. In this study, we designed a four-step deep learning approach to classify six (6) histopathological regions and quantitatively evaluated it on the BraTS-Path 2024 challenge dataset, which includes digitized Hematoxylin \& Eosin (H\&E) stained GBM tissue sections annotated for six distinct regions. We used the challenge's publicly available training dataset to develop and evaluate the effectiveness of several variants of EfficientNet architectures (i.e., B0, B1, B2, B3, B4). EfficientNet-B1 and EfficientNet-B4 achieved the best performance, achieving an F1 score of 0.98 in a 5-fold cross-validation configuration using the BraTS-Path training set. The quantitative performance evaluation of our proposed approach with EfficientNet-B1 on the BraTS-Path hold-out validation data and the final hidden testing data yielded F1 scores of 0.546 and 0.517, respectively, for the associated 6-class classification task. The difference in the performance on training, validation, and testing data highlights the challenge of developing models that generalize well to new data, which is crucial for clinical applications. The source code of the proposed approach can be found at the GitHub repository of Indiana University Division of Computational Pathology: https://github.com/IUCompPath/brats-path-2024-enet.

Paper Structure

This paper contains 18 sections, 2 equations, 3 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Materials and Methods
  3. Data
  4. Pre-processing
  5. Deconvolution:
  6. Data Augmentation:
  7. Normalization:
  8. Model Training
  9. EfficientNet:
  10. Cross-Validation:
  11. Loss:
  12. Metrics:
  13. Inference
  14. Experimental Design
  15. Results
  16. ...and 3 more sections

Figures (3)

  • Figure 1: Sample images in the training data. 1-Cellular Tumor (CT), 2-Pseudopalisading Necrosis (PN), 3-Infiltration into Cortex (IC), 4-Geographic Necrosis (NC), 5-Microvascular Proliferation (MP), and 6-Penetration into White Matter (WM).
  • Figure 2: Training F1 scores and Loss Curves for ENet-B1 and ENet-B4 per Epoch with Gamma = 0.9.
  • Figure 3: Cross Validation Training Metric Curves for ENet-B1 and ENet-B4 per Epoch.