Proteins are involved in all tasks of life, and their characterization is essential to understand the underlying mechanisms of biological processes. We present a method called "differential visual proteomics" geared to study proteome-wide structural changes of proteins and protein-complexes between a disturbed and an undisturbed cell or between two cell populations. To implement this method, the cells are lysed and the lysate is prepared in a lossless manner for single-particle electron microscopy (EM). The samples are subsequently imaged in the EM. Individual particles are computationally extracted from the images and pooled together, while keeping track of which particle originated from which specimen. The extracted particles are then aligned and classified. A final quantitative analysis of the particle classes found identifies the particle structures that differ between positive and negative control samples. The algorithm and a graphical user interface developed to perform the analysis and to visualize the results were tested with simulated and experimental data. The results are presented, and the potential and limitations of the current implementation are discussed. We envisage the method as a tool for the untargeted profiling of the structural changes in the proteome of single-cells as a response to a disturbing force.