Automatic Lung Tumor Volume Estimation Using PSO-TSVM Method

This research is driven by concerns over the increasing cost of patient care and demand for faster clinical interventions. It is based on the prime hypothesis that well-trained machine learning models can be used to improve the accuracy of pathology localization. A well-tested, robust and accurate model for tumor volume measurement and segmentation. The proposed method uses CT scans for diagnosis and combines particle swarm optimization with twin support vector machine to detect the lung infection. A two-stage framework called Suspicious Volumetric Tumor Segmentation (SVTS) following a segmentation classification pipeline is proposed for classification of normal and abnormal tumorous lung CT slices from a standard dataset. The experiments conducted on the dataset given the better result when compared to other machine learning methods. The performance of the proposed method measured by accuracy, precision, sensitivity, error rate and F1-Score. The proposed method achieves 96.18 % accuracy.