PaperlayerView exact revision on arXiv

Source-linked AI summary

Comparing Different Deep Learning Architectures for Classification of Chest Radiographs

Keno K. Bressem, Lisa Adams, Christoph Erxleben, Bernd Hamm, Stefan Niehues, Janis Vahldiek

arXiv:2002.08991v1cs.LGcs.CVeess.IV

TL;DR

Chest-radiograph classifiers commonly reuse deep architectures despite radiographs being grayscale and containing fewer classes than typical image datasets. This study compares neural-network architectures on CheXpert and finds that smaller models can match or surpass deeper networks while requiring fewer computational resources.

  • Problem

    It was unclear whether very deep neural networks are necessary for classifying comparatively simple grayscale chest radiographs.

    Source: Introduction · p. 2

  • Method

    The study systematically compared fifteen neural-network architectures on the CheXpert chest-radiograph dataset using AUROC, AUPRC, and training-time analyses.

    Source: Introduction · p. 2, Data preparation · p. 2, Results · p. 3

  • Results

    Smaller networks such as ResNet-34, AlexNet, and VGG-16 achieved comparable or superior classification performance to deeper models, with VGG-16 reaching an AUPRC of 0.709.

    Source: Overall best Performance · p. 4, Discussion · p. 4, Conclusion · p. 5

  • Takeaways & Limitations

    Smaller networks can reduce hardware requirements and training time without loss of performance in chest-radiograph classification.

    Source: Conclusion · p. 5

    The work compared architectures rather than completely optimizing one specific network.

    Source: Discussion · p. 5

Abstract

from arXiv · show

Chest radiographs are among the most frequently acquired images in radiology and are often the subject of computer vision research. However, most of the models used to classify chest radiographs are derived from openly available deep neural networks, trained on large image-datasets. These datasets routinely differ from chest radiographs in that they are mostly color images and contain several possible image classes, while radiographs are greyscale images and often only contain fewer image classes. Therefore, very deep neural networks, which can represent more complex relationships in image-features, might not be required for the comparatively simpler task of classifying grayscale chest radiographs. We compared fifteen different architectures of artificial neural networks regarding training-time and performance on the openly available CheXpert dataset to identify the most suitable models for deep learning tasks on chest radiographs. We could show, that smaller networks such as ResNet-34, AlexNet or VGG-16 have the potential to classify chest radiographs as precisely as deeper neural networks such as DenseNet-201 or ResNet-151, while being less computationally demanding.

Introduction

Chest radiograph classification remains an active research area despite widespread use of established neural-network architectures. This study tests whether shallower networks can match or exceed deeper models with fewer resources.

  • Because radiographs contain fewer relevant pathologies than many other imaging tasks, shallow networks may be sufficient for classification.Source: Introduction · p. 2
  • The study systematically examines fifteen openly available neural-network architectures for basic chest-radiograph classification.Source: Introduction · p. 2
  • The central hypothesis is that shallow networks might outperform deeper networks while requiring fewer resources.Source: Introduction · p. 2

Methods

The study compares fourteen convolutional neural networks using selected CheXpert labels and standardized training and evaluation procedures. Images were restricted to frontal radiographs and uncertainty-labeled cases were excluded.

  • Training used five CheXpert findings: cardiomegaly, edema, consolidation, atelectasis, and pleural effusion.Source: Data preparation · p. 2
  • Only frontal radiographs without uncertainty labels were used, leaving 135,494 images from 53,388 patients for training.Source: Data preparation · p. 2
  • Models were trained with 320 x 320 pixel images, batch sizes of 16 or 32, and staged classification-head then whole-network training.Source: Data augmentation · p. 2, Model training · p. 3
  • The analysis trained fourteen CNNs spanning ResNet, DenseNet, VGG, SqueezeNet, and AlexNet architectures.Source: Model training · p. 3
  • Performance was assessed with ROC and precision-recall curves using AUROC and AUPRC.Source: Evaluation · p. 3

Results

Performance depended on the metric: deeper networks generally led pooled AUROC, while shallower models achieved the highest pooled AUPRC. Training time increased substantially with network depth.

AUROC

AUPRC

Training time

  • AlexNet required 20 minutes for batch size 32, whereas DenseNet-201 required 5:11 hours with batch size 16.Source: Training time · p. 4
  • Increasing batch size from 16 to 32 accelerated training by 29.9% ± 9.34% on average.Source: Training time · p. 4

Discussion

The study finds that deeper networks do not necessarily perform better on chest radiographs. Smaller models can achieve comparable or superior classification while reducing computational demands.

  • More complex networks do not necessarily perform better than shallow networks for chest-radiograph classification.Source: Discussion · p. 4
  • AlexNet, ResNet-34, and VGG-16 may surpass deeper models such as ResNet-150 and DenseNet-201.Source: Discussion · p. 4
  • Lower resource demands may enable higher image resolutions and facilitate human-in-the-loop annotation workflows.Source: Discussion · p. 4
  • The work focused on architecture comparison rather than complete optimization of one specific network.Source: Discussion · p. 5
  • Smaller networks require less hardware and training time without loss of performance.Source: Conclusion · p. 5

Tables

The tables compare models using AUROC, AUPRC, and training duration. Deeper models generally performed better on AUROC, shallower models on AUPRC, while training time increased with network depth.

Receiver Operating Characteristic Curves

Figures 1–3 present ROC curves for all models, distinguishing individual trainings from pooled performance across five trainings.

Precision Recall Curves

Figures 1–3 also display precision-recall curves for all models, separating individual training runs from pooled five-training performance.

Related papers

Loading 2002.08991v1…