The researchers, who work collaboratively at the Marine Biological Laboratory’s Whitman Center, published their results in the journal Optica.
“Everybody knows fluorescence imaging is inefficient in that the microscope only captures a portion of the light (spewing off the specimen),” says senior author Hari Shroff of the National Institute of Biomedical Imaging and Bioengineering. “In this paper, we showed you can not only capture that lost light, but use computation to fuse it to the existing image and make the image sharper.”
Developed by Yicong Wu, a staff scientist in Shroff’s lab, the new system achieved resolution of up to 235 x 235 x 340 nanometers, which is double the volumetric resolution of traditional fluorescence microscopy methods.
To collect more of the available light (which, in turn, provides more information about the specimen), the new microscope has three objective lenses acquiring views of the sample simultaneously. The views are then aligned and merged by a computational process known as deconvolution.
Those computations were worked out in collaboration with co-author Patrick La Rivière of the University of Chicago’s Radiology Department, who typically develops algorithms for improving “dose efficiency” in human-scale medical imaging, such as CAT scans.
“In medical imaging, we are always worried about dose, about capturing every X-ray [used on the patient to improve scan resolution]. We are concerned with ‘How can we do more with less?'” La Rivière says.
In microscopy, the amount of light used presents similar concerns. “If you use very intense illuminations to image something microscopic like a worm embryo, you might change its biology or even kill it. You need to be dose efficient with your light,” La Rivière says.
La Rivière and Shroff began collaborating at the MBL in 2014, initially on algorithms to improve Shroff’s diSPIM microscope (which has two objective lenses) and eventually on the new three-lensed microscope (called triSPIM).