Manuscript for paper on lowest dose recognition of protein

This deliverable reports on the progress that has been achieved in the direction of the dose-effective imaging of organic molecules. We report several essential experimental advancements, including a successful experimental scheme of ghost imaging that could be realised in Modena without the need of travelling. Overall this deliverable includes material for at least 3 papers and a patent. The restriction on travels due to Covid-19 imposed hard constraints on the performance of the intended experiment. The complete experiment on proteins would, in fact, have required the simultaneous presence of three groups at the microscope in Jülich: FZJ personnel for main operation on the microscope, people from MU for the cryomicroscopy part, and people from CNR for the lateral control of software and for the coordination of the experiment. Moreover, the preparation of the experiment required the intervention of FEI personnel. Unfortunately, in spite of the partial release of the restrictions during the extra 6 months allotted to Q-SORT, for the reasons given above these conditions were impossible to meet. This was unforeseeable at the moment the project extension was asked. Given these hard constraints, we chose the combination of experimental activities that allowed us to achieve the greatest possible amount of project goals. In fact, Q-SORT scientists achieved more than what was foreseen; novel results include: 1) the first patentable phase plates of different design ( designed in WP3 D3.4 for protein recognition) able to control up to 24 electrodes by multiplexing the initial 8 inputs. 2) the first computational ghost imaging experiment in microscopy and its optimization for low dose regime 3) a state of the art neural network control for an efficient prediction of the generated caustic pattern. Two approaches to the protein recognition were chosen at the end of WP3: a) the extension of the OAM sorter with an additional programmable phase plate. Such approach is referred to as GOAMS (generalised OAM sorter) in D3.2 as b) the computational ghost imaging with increasingly complex masks. Under the given constraints, it was found that the computational ghost imaging approach (D3.2, task 3.1 & 3.2) was the most appropriate and effective to pursue - enabling the greatest advancement. This development is also consistent with contingency plan #6 “Test random or Hadamard sparse sampling”. The results have been also written as a manuscript that is going to be submitted soon. The hardware for the 3-element GOAMS (i.e. the second approach that was found to be most promising in WP3) was also developed, with the design of an additional element that allowed the recognition of the radial degree of freedom. Such extended “protein specific” phase modulator (D3.4) was also finished and tested as one of the design of the programmable phase plates. [...]