Non-invasive assessment of tissue perfusion and arterial blood flow has gained increased interest in the recent past for monitoring various physiological dysfunctions. Assessment of arterial flow can additionally indicate the presence of plaque depositions. Laser speckle imaging and associated processing techniques have the potential to present a noncontact, useful and inexpensive method of achieving this goal. However, the derived flow parameters offer present poor sensitivity due to various factors such as tissue static and dynamic optical properties. Also, the simultaneous quantitative estimation of microcirculation parameters such as flow velocity and scatterer concentration are challenging. The introduction of machine learning methods into laser speckle image analysis can help meet these challenges to a great extent. The temporal fractal-based approach to measure tissue perfusion parameters using time-varying dynamic speckle signals captured by a low-cost laser speckle imaging system is presented. Further, an approach for the simultaneous extraction of perfusion parameters, using multi-target regression techniques applied to the extracted features from acquired laser speckle images is demonstrated. The results show that the multi-target regression trees act as an effective tool for the simultaneous extraction of flow velocity and scatterer concentration, at reasonable speed and computational efficiency that help develop a real-time tool for flow analysis. In the second part of the work, spatial and temporal analysis of the speckle images captured from the designed carotid artery phantom with different percentages of plaque is carried out. Identification of different grades of atherosclerosis is achieved by integrating machine learning approach to the acquired laser speckle images. The results show the potential of laser speckle imaging for assessing dynamics in macrovascular flow and subsequent classification of mild, moderate and high-risk plaque deposition stages