Detection and visualization of abnormality in chest radiographs using modality-specific convolutional neural network ensembles.

Sivaramakrishnan Rajaraman,Incheol Kim,Sameer K Antani

doi:10.7717/peerj.8693

Sivaramakrishnan Rajaraman, Incheol Kim + Show 1 more

Open Access

https://doi.org/10.7717/peerj.8693

Copy DOI

Abstract

Convolutional neural networks (CNNs) trained on natural images are extremely successful in image classification and localization due to superior automated feature extraction capability. In extending their use to biomedical recognition tasks, it is important to note that visual features of medical images tend to be uniquely different than natural images. There are advantages offered through training these networks on large scale medical common modality image collections pertaining to the recognition task. Further, improved generalization in transferring knowledge across similar tasks is possible when the models are trained to learn modality-specific features and then suitably repurposed for the target task. In this study, we propose modality-specific ensemble learning toward improving abnormality detection in chest X-rays (CXRs). CNN models are trained on a large-scale CXR collection to learn modality-specific features and then repurposed for detecting and localizing abnormalities. Model predictions are combined using different ensemble strategies toward reducing prediction variance and sensitivity to the training data while improving overall performance and generalization. Class-selective relevance mapping (CRM) is used to visualize the learned behavior of the individual models and their ensembles. It localizes discriminative regions of interest (ROIs) showing abnormal regions and offers an improved explanation of model predictions. It was observed that the model ensembles demonstrate superior localization performance in terms of Intersection of Union (IoU) and mean Average Precision (mAP) metrics than any individual constituent model.

Highlights

Computer-aided diagnosis (CADx) tools have gained immense prominence in medicine by augmenting clinical expertise and reducing observer variability (Bar et al, 2015)
We propose modality-specific ensemble learning toward improving abnormality detection in chest X-rays (CXRs)
Since the coarse models have already learned the ability to classify the normal and abnormal CXRs with a large-scale dataset with a range of data distributions, the learned weights served as a promising initialization for a related abnormality classification task in a dataset with a different distribution, as compared to ImageNet weights learned from stock photographic images that are visually distinct from medical images

Summary

Introduction

Computer-aided diagnosis (CADx) tools have gained immense prominence in medicine by augmenting clinical expertise and reducing observer variability (Bar et al, 2015). Datadriven deep learning (DL) algorithms using convolutional neural networks (CNNs) have been successfully applied to chest X-ray (CXR) screening (Singh et al, 2018; Rajpurkar et al, 2018; Qin et al, 2018; Irvin et al, 2019; Pasa et al, 2019). The CXRs are analyzed for typical abnormalities to localize suspicious regions Detection and visualization of abnormality in chest radiographs using modality-specific convolutional neural network ensembles. Generalization would require retraining models on additional expert-annotated training data and labels.

Objectives

Methods

Results

Discussion

Conclusion