BO-CLAHE enhancing neonatal chest X-ray image quality for improved lesion classification

Abstract

In the case of neonates, especially low birth weight preterm and high-risk infants, portable X-rays are frequently used. However, the image quality of portable X-rays is significantly lower compared to standard adult or pediatric X-rays, leading to considerable challenges in identifying abnormalities. Although attempts have been made to introduce deep learning to address these image quality issues, the poor quality of the images themselves hinders the training of deep learning models, further emphasizing the need for image enhancement. Additionally, since neonates have a high cell division rate and are highly sensitive to radiation, increasing radiation exposure to improve image quality is not a viable solution. Therefore, it is crucial to enhance image quality through preprocessing before training deep learning models. While various image enhancement methods have been proposed, Contrast Limited Adaptive Histogram Equalization (CLAHE) has been recognized as an effective technique for contrast-based image improvement. However, despite extensive research, the process of setting CLAHE's hyperparameters still relies on a brute force, manual approach, making it inefficient. To address this issue, we propose a method called Bayesian Optimization CLAHE(BO-CLAHE), which leverages Bayesian optimization to automatically select the optimal hyperparameters for X-ray images used in diagnosing lung diseases in preterm and high-risk neonates. The images enhanced by BO-CLAHE demonstrated superior performance across several classification models, with particularly notable improvements in diagnosing Transient Tachypnea of the Newborn (TTN). This approach not only reduces radiation exposure but also contributes to the development of AI-based diagnostic tools, playing a crucial role in the early diagnosis and treatment of preterm and high-risk neonates.

Keywords: Bayesian optimization; CLAHE; High-risk neonates; Neonatal chest X-ray; Preterm.

Figure 1

A comprehensive workflow for neonatal lung disease classification using BO-CLAHE-enhanced X-ray images. The process includes preprocessing with bone suppression and chest segmentation, image enhancement using Bayesian optimization, and final classification into five disease categories (Normal, RDS, Air Leak Syndrome, Atelectasis, and TTN) with evaluation based on accuracy, F1 score, AUC-ROC, and expert opinion.

Figure 2

Statistical evaluation of accuracy, F1 score, recall, precision, and AUC for various enhancement methods. Table (a) summarizes ANOVA results, while Tables (b–f) provide descriptive statistics, including mean, standard deviation, and quartiles for each metric.

Figure 3

Comparison of X-ray images for different conditions with and without enhancement.

Figure 4

AUC-ROC curves for five disease classes using GoogleNet and InceptionV3 models on original and enhanced images.

Figure 5

Comparison of original and enhanced X-ray images after bone removal for different conditions.

Figure 6

Chest segmentation result.

Algorithm 1

BO-CLAHE algorithm with SSIM and BRISQUE

Figure 7

Bayesian optimization.

Figure 8

Boxplot distributions of model-wise performance metrics, including (a) accuracy, (b) F1 score, (c) recall, (d) precision, and (e) AUC. Each plot compares the performance of different enhancement methods: origin, BPDFHE, PhyCV, Real-ESRGAN, and BO-CLAHE.