An Optimized EfficientNetB0 Framework with CLAHE-Based Preprocessing for Accurate Multi-Class Chest X-Ray Classification (DenseNet121/MobileNetV2/EfficientNetB0)

Abstract

We deposit here the versioned code corresponding to the results reported in our manuscript . The archive contains the complete implementation for training and evaluating three CNN architectures (DenseNet121, MobileNetV2, EfficientNetB0) for thoracic finding classification using the NIH ChestX-ray dataset, including preprocessing routines and evaluation utilities used to compute the reported performance metrics. To comply with dataset terms, no image data are included; the dataset should be downloaded independently from the official NIH Clinical Center repository.