Compared to other modes, bicyclists are disproportionally affected by crashes considering their low mode share. There is evidence that crashes between bicyclists and motorized vehicle take place at road segments and signalized intersections where bicycle treatments (e.g., bike lanes) are present, urging for in-dept analysis of the safety impact of the various bicycle treatment types. Additionally, it is important to identify sensor types that have the potential to advance field data collection and traffic monitoring in multi-modal road environments. In this dissertation, three approaches, namely crash analysis, traffic conflict analysis, and analysis of driver speeding and glancing behavior, were implemented to investigate the safety impact of bicycle treatments at the segment- and the intersection-levels on bicycle safety. Prediction models were developed to predict bicycle-motorized vehicle crashes at road segments and signalized intersections, and traffic conflicts between straight-going bicyclists and right-turning vehicles at signalized intersections. Driver speeding and glancing behavior was analysed for the segment and the intersection levels. A mode classification framework to classify trajectories recorded using a radar-based sensor was developed to test the feasibility of using radar-based sensors in field studies. The findings of this dissertation contribute to bicycle safety research in terms of quantifying the safety impact of various bicycle treatment types and how to assess and also, by showing how to assess bicycle safety. The findings of this research have the potential to stand as a valuable tool for transportation policymakers and officials in charge of establishing safe bicycle networks.