Lightweight Networks for COVID-19 Detection from Chest X-Ray Images inside a Low-Tier Android Device

The efforts to inoculate majority of the population have been slower than expected and this is especially true for lower income countries. This problem has caused a lot of worries and further accentuates the importance of timely and effective mass testing considering the emergence of newer variants. The RT-PCR is still the gold standard diagnostic test for COVID-19 detection, but its limitations has led researchers and scientists to explore supplementary screening methods. One effective tool to consider is Chest X-Ray (CXR) imaging and combining it with deep learning has piqued attention from the artificial intelligence (AI) community. To further contribute to this research area, this work focuses on creating, evaluating, and comparing lightweight and mobile-phone-suitable COVID-detecting models. These transfer learning models together with their corresponding dynamic-range quantized versions are first tested according to their classification performance. Afterwards, the models are pushed in a low-tier phone to measure their resource consumption and inference timings. Results show that the utilization of EfficientNetB0 and MobileNetV3 (Small & Large) architectures for transfer learning without any quantization can produce at least 91 % overall average accuracy for 3-class classification scheme. For systems requiring more efficient models, using the quantized versions of the transfer learning models particularly with EfficientNetB0 and MobileNetV3Large as foundation can render at most 0.79 % accuracy loss but still show more than 95% f1-scores for the COVID-19 class. © 2022 IEEE.

Detecting COVID-19 from Chest X-Ray Images using a Lightweight Deep Transfer Learning Model with Improved Contrast Enhancement Technique

Despite the vaccinations, the emergence of new and more contagious variants of the COVID-19 disease has continued to pose threats and challenges to our lives. Until herd immunity is achieved, it is important to continuously perform screening tests to control and minimize the transmissions. Due to the reported shortcomings of the RT-PCR, the utilization of deep learning for detecting COVID-19 from Chest X-Ray (CXR) images has gathered a lot of interest from researchers. As a contribution to the field, this study proposes a deep learning pipeline that utilizes transfer learning and image enhancement techniques to classify whether a given CXR image exhibits characteristics of COVID-19 infection, pneumonia infection, or normal/healthy lungs. For a lighter approach, the small pre-trained model named EfficientNetB0 is used as the base model for the transfer learning method. To improve the network's performance, a sequence of contrast enhancement techniques namely the Multi-Scale Retinex (MSR) and Contrast Limited Adaptive Histogram Equalization (CLAHE) is introduced in the pipeline and employed as a pre-processing step. Gathered from a 10-fold cross-validation method in a dataset with 3729 images per class, results show that the proposed approach achieves an average overall accuracy of 92.089% with 98.431% average precision, 95.119% average recall, and 96.741% average f1-score for the COVID-19 class. © 2021 IEEE.