The intent of this paper is to conjure a reflexive real-time spotting of thrombosis in the femoral nerve. This paper exhibits blood clot recognition model of the ultrasound (US) femoral nerve images using the techniques of active contour, shape based geometric boundary descriptors and texture based local descriptors. Specially, Local Mesh Pattern (LMeP) and Gray-Level Co-occurrence Matrix (GLCM) are amalgamated as a novel texture based local descriptor called Local Mesh Cooccurrence Pattern (LMCoP). It is decidedly used to extract the local directional information from the US image. By enforcing Random Forest (RF) classifier model; the severity of the thrombus is automatically classified. The classifiers performance were studied and compared.