Prostate Segmentation from 3D MRI Using a Two-Stage Model and Variable-Input Based Uncertainty Measure

TL;DR

This paper introduces a two-stage 3D MRI prostate segmentation model with a novel variable-input uncertainty measure, demonstrating improved confidence estimation and post-processing to enhance segmentation accuracy and smoothness.

Contribution

It presents a new variable-input based uncertainty measure for 3D MRI segmentation, along with an uncertainty-guided post-processing method, advancing confidence estimation in deep learning models.

Findings

High correlation between the proposed uncertainty measure and ground-truth performance metrics

Uncertainty-guided post-processing improved segmentation label smoothness

The model demonstrated robustness across varied testing cases

Abstract

This paper proposes a two-stage segmentation model, variable-input based uncertainty measures and an uncertainty-guided post-processing method for prostate segmentation on 3D magnetic resonance images (MRI). The two-stage model was based on 3D dilated U-Nets with the first stage to localize the prostate and the second stage to obtain an accurate segmentation from cropped images. For data augmentation, we proposed the variable-input method which crops the region of interest with additional random variations. Similar to other deep learning models, the proposed model also faced the challenge of suboptimal performance in certain testing cases due to varied training and testing image characteristics. Therefore, it is valuable to evaluate the confidence and performance of the network using uncertainty measures, which are often calculated from the probability maps or their standard deviations with multiple model outputs for the same testing case. However, few studies have quantitatively compared different methods of uncertainty calculation. Furthermore, unlike the commonly used Bayesian dropout during testing, we developed uncertainty measures based on the variable input images at the second stage and evaluated its performance by calculating the correlation with ground-truth-based performance metrics, such as Dice score. For performance estimation, we predicted Dice scores and Hausdorff distance with the most correlated uncertainty measure. For post-processing, we performed Gaussian filter on the underperformed slices to improve segmentation quality. Using PROMISE-12 data, we demonstrated the robustness of the two-stage model and showed high correlation of the proposed variable-input based uncertainty measures with GT-based performance. The uncertainty-guided post-processing method significantly improved label smoothness.