Stacked BiLSTM with ResNet50 for Medical Image Classification

ABSTRACT

The healthcare sector plays a significant role in the industry, where a client looks for the highest amount of care and services, no matter the cost. However, this sector has not satisfied society's presumption, even if this industry consumes a considerable percentage of the national budget. In the past, medical experts have been looking for smart medical solutions. This work focuses on accurate and early detection of illness from various medical images. Early detection not only aids in the development of better medications but can also save a life in the long run. Deep learning provides an excellent solution for early medical imaging in healthcare. This paper proposed a Stacked-based BiLSTM with Resnet50 Model using an AdaSwarm optimizer to classify and analyze the medical illnesses from the different medical image datasets. For this study, four medical datasets were used as benchmarks Covid19, Pneumonia, Ma, and Lung Cancer. Accuracy, AUC, ROC, and F1 Score performance metrics are used to evaluate the prosed model from other models. The proposed model gives a mean ACCURACY, AUC, ROC, and F1 Score on these four datasets are 98%, 99%, 97%, and 98%, respectively. © 2022 IEEE.