Improving the accuracy in localisation microscopy for samples that consist of straight lines using prior knowledge

We have investigated the possibility of using prior knowledge about a sample in order to improve the image quality of localisation microscopy. We assumed that all fluorescent molecules were aligned on one, or two parallel line segments and used this assumption to develop an estimator for the emitter locations and the line parameters. Using simulation we have shown that the Maximum-Likelihood estimator we defined is able to reduce the Root Mean Square localisation error up to maximum factor of 1? ?2 compared to ordinary computational methods that do not use any prior knowledge.Imaging Science & TechnologyApplied Science