Quantifying U-Net uncertainty in multi-parametric MRI-based glioma segmentation by spherical image projection

Dose-Incorporated Deep Ensemble Learning for Improving Brain Metastasis Stereotactic Radiosurgery Outcome Prediction

PURPOSE

To develop a novel deep ensemble learning model for accurate prediction of brain metastasis (BM) local control outcomes after stereotactic radiosurgery (SRS).

METHODS AND MATERIALS

A total of 114 brain metastases (BMs) from 82 patients were evaluated, including 26 BMs that developed biopsy-confirmed local failure post-SRS. The SRS spatial dose distribution (Dmap) of each BM was registered to the planning contrast-enhanced T1 (T1-CE) magnetic resonance imaging (MRI). Axial slices of the Dmap, T1-CE, and planning target volume (PTV) segmentation (PTVseg) intersecting the BM center were extracted within a fixed field of view determined by the 60% isodose volume in Dmap. A spherical projection was implemented to transform planar image content onto a spherical surface using multiple projection centers, and the resultant T1-CE/Dmap/PTVseg projections were stacked as a 3-channel variable. Four Visual Geometry Group (VGG-19) deep encoders were used in an ensemble design, with each submodel using a different spherical projection formula as input for BM outcome prediction. In each submodel, clinical features after positional encoding were fused with VGG-19 deep features to generate logit results. The ensemble's outcome was synthesized from the 4 submodel results via logistic regression. In total, 10 model versions with random validation sample assignments were trained to study model robustness. Performance was compared with (1) a single VGG-19 encoder, (2) an ensemble with a T1-CE MRI as the sole image input after projections, and (3) an ensemble with the same image input design without clinical feature inclusion.

RESULTS

The ensemble model achieved an excellent area under the receiver operating characteristic curve (AUCROC: 0.89 ± 0.02) with high sensitivity (0.82 ± 0.05), specificity (0.84 ± 0.11), and accuracy (0.84 ± 0.08) results. This outperformed the MRI-only VGG-19 encoder (sensitivity: 0.35 ± 0.01, AUCROC: 0.64 ± 0.08), the MRI-only deep ensemble (sensitivity: 0.60 ± 0.09, AUCROC: 0.68 ± 0.06), and the 3-channel ensemble without clinical feature fusion (sensitivity: 0.78 ± 0.08, AUCROC: 0.84 ± 0.03).

CONCLUSIONS

Facilitated by the spherical image projection method, a deep ensemble model incorporating Dmap and clinical variables demonstrated excellent performance in predicting BM post-SRS local failure. Our novel approach could improve other radiation therapy outcome models and warrants further evaluation.

Thermal damage map prediction during irreversible electroporation with U-Net

Recent developments in cancer treatment with irreversible electroporation (IRE) have led to a renewed interest in developing a treatment planning system based on Deep-Learning methods. This paper will give an account of U-Net, as a Deep-Learning architecture usage for predicting thermal damage area during IRE. In this study, an irregular shape of the liver tumor with MIMICS and 3-Matic software was created from Magnetic Resonance Imaging (MRI) images. To create electric field distribution and thermal damage maps in IRE, COMSOL Multiphysics 5.3 finite element analysis was performed. It was decided to use the pair needle, single bipolar, and multi-tine electrodes with different geometrical parameters as electrodes. The U-Net was designed as a Deep-Learning network to train and predict the thermal damage area from electric field distribution in the IRE. The average DICE coefficient and accuracy of trained U-Net for predicting thermal damage area on test data sets were 0.96 and 0.98, respectively, for the dataset consisting of all electrode type electric field intensity images. This is the first time that U-Net has been used to predict thermal damage area. The results of this research support the idea that the U-Net can be used for predicting thermal damage areas during IRE as a treatment planning system.