A Quality Control System for Automated Prostate Segmentation on T2-Weighted MRI.

Mohammed R S Sunoqrot,Tone F Bathen,Gabriel A Nketiah,Mattijs Elschot,Olmo Zavala-Romero,Elise Sandsmark,Radka Stoyanova,Kirsten M Selnæs

doi:10.3390/diagnostics10090714

Abstract

Computer-aided detection and diagnosis (CAD) systems have the potential to improve robustness and efficiency compared to traditional radiological reading of magnetic resonance imaging (MRI). Fully automated segmentation of the prostate is a crucial step of CAD for prostate cancer, but visual inspection is still required to detect poorly segmented cases. The aim of this work was therefore to establish a fully automated quality control (QC) system for prostate segmentation based on T2-weighted MRI. Four different deep learning-based segmentation methods were used to segment the prostate for 585 patients. First order, shape and textural radiomics features were extracted from the segmented prostate masks. A reference quality score (QS) was calculated for each automated segmentation in comparison to a manual segmentation. A least absolute shrinkage and selection operator (LASSO) was trained and optimized on a randomly assigned training dataset (N = 1756, 439 cases from each segmentation method) to build a generalizable linear regression model based on the radiomics features that best estimated the reference QS. Subsequently, the model was used to estimate the QSs for an independent testing dataset (N = 584, 146 cases from each segmentation method). The mean ± standard deviation absolute error between the estimated and reference QSs was 5.47 ± 6.33 on a scale from 0 to 100. In addition, we found a strong correlation between the estimated and reference QSs (rho = 0.70). In conclusion, we developed an automated QC system that may be helpful for evaluating the quality of automated prostate segmentations.

Highlights

Prostate cancer is one of the most commonly diagnosed cancers among men worldwide [1]
We explored the importance of the radiomics features the system is based on and compared a generalizable model with models trained on specific combinations of dataset and segmentation algorithm
We found a strong correlation between the Reference quality scores (rQSs) and estimated quality score (eQS) and mean absolute error (MAE) values less than the standard deviation between the experts and the second observer segmentations (5.37 vs. 7.76), implying that errors were in an acceptable range

Summary

Introduction

Prostate cancer is one of the most commonly diagnosed cancers among men worldwide [1]. Automated computer-aided detection and diagnosis (CAD) systems, which exploit the quantitative information in MR images, are providing promising solutions to overcome these limitations of qualitative image interpretation and support clinical decision making [6,8]. The segmentation of the organ of interest, in this case the prostate gland, constitutes one of the first important steps in a CAD system workflow [7,9]. This step helps remove irrelevant image information and facilitates subsequent extraction of quantitative image features (radiomics) from sub-regions/volumes such as tumors for further analysis or diagnosis. Manual segmentation of the prostate, which is traditionally performed on T2-weighted (T2W) MR images by radiologists, is a time-consuming task

Objectives

Methods

Results

Discussion

Conclusion