Temporal unmixing Optoacoustics – Machine learning to enable routine whole animal Optoacoustic imaging of genetically encoded photo-modulatable labels.
Bioinformatics and Theoretical Biology
Final Report Abstract
Bio(medical) imaging of intact opaque organisms in vivo allows unique insights into anatomy, physiology and the cause and treatment of diseases alike. Often such imaging technologies are closely related to their counterparts in diagnostic medical imaging, and findings from the research on model organisms, and increasingly organoids, are a first step towards a translation to medical applications with societal benefit. So called photo- or optoacoustic imaging is a unique imaging methods combining deep penetration in tissue with high resolution. Its limitations regarding targetable contrast, e.g. for specific cell types, have been recently challenged by us and others with the introduction of photoswitching optoacoustic imaging. The method of photoswitching optoaocustics relies on contrast agents providing a light driven modulation that allows to supress the strong background inherent to optoacoustic and allows to also visualize lower numbers of labelled cells deep in the tissue. The funded research seeks to advance on the initial data analysis concept which extracts the modulation from the data as well as better understanding a number of artefacts that arose in first implementations. The work succeeded in laying the foundations for a standardized data acquisition that allows to concomitantly acquire highly correlated ground truth data. This ground-truth data, orthogonal to the in vivo photoswitching optoacoustic imaging, is acquired via fluorescence imaging on cryosection of the sacrificed animal. Such data is essential to benchmark an imaging method and understand what one can see and what not, to better understand artefacts and as training data for advanced AI driven analysis concepts. Furthermore, using so called tissue-mimicking phantoms with highly defined conditions a number of confounders like the impact of out of focus signals from photoswitching contrast agent could be better understood. Lastly, an incrementally improved analysis version was established and employed in the unmixing of data from a targetable injectable agent for photoswitching optoacoustics, which provides a potential first step towards translation to the diagnostic use of the photoswitching concept. In summary, the funded researched advanced the routine application of photoswitching optoacoustics and lays the foundations for a true improvement of the analysis methodologies aiming to visualize also small numbers (<100 cells) deep in the animal with high resolution towards live insights into the living organism.
Publications
-
A practical guide to photoswitching optoacoustics tomography. Methods in Enzymology, 365-383. Elsevier.
Stankevych M., Mishra K., Ntziachristos V. & Stiel A.C.
-
Alginate beads as a highly versatile test-sample for optoacoustic imaging. Photoacoustics, 25, 100301.
Fuenzalida-Werner, Juan Pablo; Mishra, Kanuj; Stankevych, Mariia; Klemm, Uwe; Ntziachristos, Vasilis & Stiel, Andre C.
-
Controlling the sound of light: photoswitching optoacoustic imaging. Nature Methods, 21(11), 1996-2007.
Stiel, Andre C. & Ntziachristos, Vasilis
-
Photoswitching protein-XTEN fusions as injectable optoacoustic probes. Acta Biomaterialia, 195, 536-546.
Huang, Yishu; Stankevych, Mariia; Gujrati, Vipul; Klemm, Uwe; Mohammed, Azeem; Wiesner, David; Saccomano, Mara; Tost, Monica; Feuchtinger, Annette; Mishra, Kanuj; Bruns, Oliver; Geerlof, Arie; Ntziachristos, Vasilis & Stiel, Andre C.
