Systematically promoting imaging techniques and new means for advanced image analysis is one of HIP's broad ambitions, and this is where Ben Engel's work comes in to play. Ben and his team are at the forefront of the resolution limits in cryo-electron tomography (cryo-ET). The efficient, automated and meaningful analysis of information-rich three-dimensional cellular tomograms is at the core of their scientific success.
Faced with similar challenges from their own nanoscale microscopy data, the groups of Ben Engel from Helmholtz Munich, Oliver Daumke and Dagmar Kainmüller from MDC, aim to develop a machine-learning (ML) model capable of generalizing across multiple domains. Simpler put: the proposed ML solution should alleviate the need for every individual lab to create application-specific training data sets by means of tedious manual data annotation. The anticipated ML-model will be operational across research organizations and scientific fields, opening the full potential of ML methods for accurate automated tomogram segmentation and increasing accessibility to groups who do not (yet) pursue application-specific ML developments.
Beside some operational funds, the project will be spearheaded by a Postdoctoral scientist, bridging between the partner labs. Ben is convinced that ‘this endeavor will not only tangibly connect us to our colleagues at MDC, but will also trigger further internal synergies, as we are already collaborating closely with the team of Tingying Peng at Helmholtz AI.’ He concludes: ‘This new project will accelerate our efforts towards rapid and generalizable image analyses tools, bring the ambitions of HIP to life and strengthening the position of the Helmholtz Association in this exciting and highly interdisciplinary arena of the modern life sciences’.