Detecting pelvic fracture on 3D-CT using deep convolutional neural networks with multi-orientated slab images

Kazutoshi Ukai,Keigo Hayashi,Naomi Yagi,Syoji Kobashi,Hirotsugu Muratsu,Rashedur Rahman,Akihiro Maruo

doi:10.1038/s41598-021-91144-z

Kazutoshi Ukai, Keigo Hayashi + Show 5 more

Open Access

https://doi.org/10.1038/s41598-021-91144-z

Copy DOI

Abstract

Pelvic fracture is one of the leading causes of death in the elderly, carrying a high risk of death within 1 year of fracture. This study proposes an automated method to detect pelvic fractures on 3-dimensional computed tomography (3D-CT). Deep convolutional neural networks (DCNNs) have been used for lesion detection on 2D and 3D medical images. However, training a DCNN directly using 3D images is complicated, computationally costly, and requires large amounts of training data. We propose a method that evaluates multiple, 2D, real-time object detection systems (YOLOv3 models) in parallel, in which each YOLOv3 model is trained using differently orientated 2D slab images reconstructed from 3D-CT. We assume that an appropriate reconstruction orientation would exist to optimally characterize image features of bone fractures on 3D-CT. Multiple YOLOv3 models in parallel detect 2D fracture candidates in different orientations simultaneously. The 3D fracture region is then obtained by integrating the 2D fracture candidates. The proposed method was validated in 93 subjects with bone fractures. Area under the curve (AUC) was 0.824, with 0.805 recall and 0.907 precision. The AUC with a single orientation was 0.652. This method was then applied to 112 subjects without bone fractures to evaluate over-detection. The proposed method successfully detected no bone fractures in all except 4 non-fracture subjects (96.4%).

Highlights

Pelvic fracture can be considered as a significant health concern, representing one of the most common causes of hospitalization and mobility loss[1]
Some studies have detected various kinds of bone fractures on 2D X-ray radiographs based on deep convolutional neural networks (DCNNs)
The experimental results of the proposed method depend on parameters, Cth and Ith

Summary

Introduction

Pelvic fracture can be considered as a significant health concern, representing one of the most common causes of hospitalization and mobility loss[1]. Natural gaps exist between pelvic bones and could be incorrectly detected as fractures and increase the number of false-positive results Another method was proposed to detect fractures on CT images of traumatic pelvic injuries based on the registered active shape model and 2D stationary wavelet transform[19]. Thian et al.[21] proposed a method to detect wrist fractures using frontal or lateral X-ray radiographs based on faster regionbased convolutional neural network (Faster R-CNN) architecture. Cheng et al.[24] developed a human-algorithm integration system to improve the diagnosis of hip fracture Another method to classify proximal femur fracture from X-ray images was proposed based on a multistage architecture of successive CNNs in cascade along with gradient class activation maps (Grad-CAM) to visualize the most relevant areas of the images[25]. These methods were based on 2D images, and could not be applied directly to 3D images

Methods

Results

Conclusion