A Comparative Analysis of Optimization Algorithms for CNN-based Tuberculosis Detection from Chest Radiographs

Tuberculosis (TB) remains a global health challenge, particularly in resource-constrained settings. Early and accurate detection is crucial for effective treatment and preventing disease transmission. This paper proposes a convolutional neural network (CNN) approach for TB detection from chest radiographs and presents a comparative analysis of different optimization algorithms to enhance model performance. We evaluate four widely used optimizers: Stochastic Gradient Descent with Momentum (SGDM), Adaptive Gradient Algorithm (AdaGrad), Root Mean Square Propagation (RMSProp), and Adaptive Moment Estimation (Adam). Our experiments demonstrate that AdaGrad achieves the highest accuracy of 98.33% after five epochs, followed by SGDM at 97.75%. We further analyze convergence rates, training stability, and generalization capabilities of each optimizer. The findings provide valuable insights for implementing efficient deep learning models in medical imaging diagnostics with optimal training procedures.