ML and DL-based Segmentation and Classification of Different Types of Venous Thrombosis

Venous thrombosis (VT) can happen for no obvious cause, but it is generally formed in deep veins in our body due to injury or too sluggish blood flow. It is a devastating condition with severe morbidity and fatality rates. This model involves a novel, impressive pre-processing technique that could prove to be an enhancement tool for different types of thrombus by reducing non-uniform revelation while preserving its edges. Usually, algorithms used for classification estimation fail to include constraints like sharp edges, spatial smoothness, orientation, intensity and lesion variation features while preserving their naturalness, which is essential in defining the state of the affected anatomical region. Deep learning (DL) based VGGNet-19 is used to classify VT, which is often underdiagnosed and preventable. Performance analysis is conducted and outcomes are compared to existing techniques to prove effectiveness.