Preservation of High Frequency Content for Deep Learning-Based Medical Image Classification

TL;DR

This paper introduces a DWT-based method to preserve high-frequency information in chest radiographs, improving CNN classification accuracy without extra computational cost, aiding medical diagnosis.

Contribution

The paper presents a novel DWT-based technique that enhances high-frequency content retention in radiographs, improving deep learning classification performance with minimal modifications to existing CNNs.

Findings

Improved CNN accuracy on NIH Chest-8 and ImageNet datasets.

High-frequency preservation boosts structure recognition in radiographs.

Method requires minimal changes to existing CNN input pipelines.

Abstract

Chest radiographs are used for the diagnosis of multiple critical illnesses (e.g., Pneumonia, heart failure, lung cancer), for this reason, systems for the automatic or semi-automatic analysis of these data are of particular interest. An efficient analysis of large amounts of chest radiographs can aid physicians and radiologists, ultimately allowing for better medical care of lung-, heart- and chest-related conditions. We propose a novel Discrete Wavelet Transform (DWT)-based method for the efficient identification and encoding of visual information that is typically lost in the down-sampling of high-resolution radiographs, a common step in computer-aided diagnostic pipelines. Our proposed approach requires only slight modifications to the input of existing state-of-the-art Convolutional Neural Networks (CNNs), making it easily applicable to existing image classification frameworks. We show that the extra high-frequency components offered by our method increased the classification performance of several CNNs in benchmarks employing the NIH Chest-8 and ImageNet-2017 datasets. Based on our results we hypothesize that providing frequency-specific coefficients allows the CNNs to specialize in the identification of structures that are particular to a frequency band, ultimately increasing classification performance, without an increase in computational load. The implementation of our work is available at github.com/DeclanMcIntosh/LeGallCuda.