A new smart-cropping pipeline for prostate segmentation using deep learning networks

TL;DR

This paper introduces a deep learning pipeline that intelligently crops prostate MRI images to enhance segmentation accuracy by addressing class imbalance, outperforming standard cropping methods across multiple neural network architectures.

Contribution

The study proposes a novel smart-cropping method that improves prostate segmentation accuracy in MRI images by balancing foreground and background pixels, validated across five deep learning models.

Findings

Smart-cropping improves Dice scores by up to 8.9%.

All evaluated networks show enhanced segmentation accuracy with smart-cropping.

The approach effectively addresses class imbalance in prostate MRI segmentation.

Abstract

Prostate segmentation from magnetic resonance imaging (MRI) is a challenging task. In recent years, several network architectures have been proposed to automate this process and alleviate the burden of manual annotation. Although the performance of these models has achieved promising results, there is still room for improvement before these models can be used safely and effectively in clinical practice. One of the major challenges in prostate MR image segmentation is the presence of class imbalance in the image labels where the background pixels dominate over the prostate. In the present work we propose a DL-based pipeline for cropping the region around the prostate from MRI images to produce a more balanced distribution of the foreground pixels (prostate) and the background pixels and improve segmentation accuracy. The effect of DL-cropping for improving the segmentation performance compared to standard center-cropping is assessed using five popular DL networks for prostate segmentation, namely U-net, U-net+, Res Unet++, Bridge U-net and Dense U-net. The proposed smart-cropping outperformed the standard center cropping in terms of segmentation accuracy for all the evaluated prostate segmentation networks. In terms of Dice score, the highest improvement was achieved for the U-net+ and ResU-net++ architectures corresponding to 8.9% and 8%, respectively.