Soon after Sokoloff's team successfully demonstrated the 2-DG method, Martin Reivich (who had worked with Sokoloff on his initial attempt to develop a model) urged them to adapt it for use in humans. While the theoretical basis could be applied without any problem, obviously the autoradiography of postmortem brain slices could not. They would need a non-invasive way to measure local tissue concentrations of the tracer isotope. Fortunately, several researchers in the U.S. and Europe had been developing computerized three dimensional scanning technologies for clinical use. David Kuhl, then in the Department of Radiology at the University of Pennsylvania, had built a single-photon scanner that could measure local concentrations of gamma-emitting isotopes in cross sections of human brain by rotating radiation detectors (scintillation probes) around the head. Kuhl was working with Michael Phelps and Edward Hoffman, who had recently moved to Penn from Michael TerPogossian's lab at Washington University in St. Louis, where they pioneered development of the first positron emission tomography (PET) scanner. Reivich brought them in to the project.

To use 2-DG with such a scanner, however, it had to be tagged with a gamma-emitting isotope, rather than 14C. When radioactive isotopes decay, they give off alpha, beta, or gamma radiation, or positrons. (Alpha particles are protons and neutrons; beta particles are electrons; gamma radiation is light waves or photons; positrons are electron-sized particles that have a positive charge.) In a tracer substance containing a gamma-emitting isotope, positrons escape from the nuclei of the decaying radioactive atoms with a kinetic energy that drives them through the tissue. Moving through the tissue, they soon interact with the nearest negatively charged electrons. The two types of particles annihilate each other, and each collision produces two gamma rays. The two gamma rays move away from the point of annihilation with the speed of light in almost exactly opposite directions, at approximately 180 degrees with respect to each other. By placing two shielded detectors opposite each other in a line and counting the gamma rays that reach them in at the same moment, the radioactivity distributed along the line can be measured. If many such detectors are rotated around the head, then the counting data they collect can be put into a computer program that can construct images that represent quantitatively the distribution of the radioactivity within the structures of the brain.

None of the components of glucose (hydrogen, oxygen, and carbon) form gamma-emitting isotopes, so the 2-DG tracer would have to be modified with some element that did. Sokoloff enlisted the help of Alfred Wolf and Joanna Fowler, radiochemists at Brookhaven National Laboratories, who synthesized a gamma-emitting version of 2-DG, tagged with radioactive fluorine: 2[18F]fluoro-2-deoxy-D-glucose ([18F]FDG.) This tracer was used with Kuhl's single-photon scanner to image the human brain in 1977. Soon after, Kuhl, Phelps, and Hoffman moved to the University of California, Los Angeles, where they refined the early PET technology and brought it into mainstream clinical use. PET offered better spatial resolution and accuracy than Kuhl's first scanner, and the UCLA group adapted the [18F]FDG technique for use with PET. PET imaging made it possible to map all kinds of activity in the human brain with greater certainty, from the processes of thinking, sleeping, dreaming, and physical movement, to the actions of drugs and hormones. It also proved useful in clinical diagnosis, e.g., for Alzheimer's disease, Huntington's disease, and seizure disorders, and later, for identifying and tracking the treatment of many types of tumors.

Sokoloff's 2-DG method was enthusiastically received by neuroscientists in the late 1970s. As other researchers attempted to replicate the work, however, a controversy developed about the validity of the 2-deoxyglucose tracer method. In developing the DG method, Sokoloff's group had carefully accounted for the biochemical activity involved in cerebral glucose metabolism as far as possible. This included the possibility that the enzyme glucose-6-phosphatase could convert 2-DG-6-phosphate (the compound they measured) back to 2-DG, which would skew their estimation of the glucose utilization rate. But they determined that this enzyme's activity in the brain was very low, and would not significantly affect the amount of 2-DG-6-phosphate retained in the brain. Other researchers argued that it could and did in their experiments. Sokoloff and his colleagues spent nearly a decade picking apart the flawed assumptions and methods of the critics, and demonstrated that the 2-DG method was sound. Although Sokoloff was annoyed with the extra work the controversy imposed, the analyses actually increased his group's understanding of brain metabolism and helped them revise and improve the 2-DG method.

During the next twenty years, Sokoloff's group continued using 2-DG tracers to map brain regions to specific functions, and to explore the brain's metabolic operation at the cellular level. They also investigated the effects of a wide range of drugs that influence the various neurotransmitter pathways, and how genetic modifications in genetically engineered mice affect cerebral metabolism.

In his later years, Sokoloff became discouraged by the changing culture of basic science research. When he began his career, scientists had much more freedom to enjoy the process of inquiry and learning, to consider their projects carefully, and publish only when they had finished and had some solid conclusions to report. As science became more dependent on government grant funding to universities, scientists spent more time applying for grants and less time on research and training new scientists. The result was, he noted, that science had become dominated by the drive for money, power, and prestige; the scientific literature had become boring recitals of experiments; and it was increasingly difficult to find really exceptional research fellows for his NIH lab. By 2003, NIH was relocating Sokoloff's lab to much smaller quarters. Feeling that his work was losing support, and also feeling the effects of his wife's death and his own declining health, Sokoloff decided to retire in 2004. He was appointed Scientist Emeritus, and continued to collaborate on neuroscience projects for another ten years.

The Laboratory of Cerebral Metabolism, which he had headed for many decades, was closed, though some of his group continued as a new section in the NIMH to study neuro- adaptation and protein metabolism (under Dr. Carolyn Smith.) Sokoloff died on July 30, 2015, in Washington, DC, following a brief illness. His legacy was not only in his brilliant basic research, which both transformed neuroscience and translated into valuable clinical brain scanning techniques, but in the scientific values his work embodied.