The two-day picture
Share
On 1 October 1971 a patient at Atkinson Morley Hospital in Wimbledon went into a machine that had not yet become ordinary. Godfrey Hounsfield, an engineer at EMI, had been building a scanner that could reconstruct a slice of the head from X-ray measurements taken all the way around it. Radiologist James Ambrose was the clinician who agreed to try it on a person. The patient had a suspected brain tumor.
Eighty by eighty is a humiliating resolution if you are used to a photograph. It was not a photograph. Each number in the grid was a calculation: how much the X-ray beam had been attenuated along many paths, inverted into a map of density inside the skull. A conventional skull X-ray stacks the whole head into one shadow. Hounsfield's trick was to stop stacking. Take the measurements as a set of projections, then let a computer argue its way back to a slice. The 1971 machine was the point at which that math was allowed into a hospital and pointed at a living head.
It was slow in a way that is hard to remember now. The scan itself was on the order of minutes, not hours, which was already a feat. The picture was not. Reconstruction ran on a time-shared mainframe away from the hospital. Ambrose did not see the images that day. Accounts of the first clinical scan put the wait at about two days. When the pictures arrived, they showed more than the earlier test images had led him to expect. The suspected tumor was visible. Surgery later confirmed it. Hounsfield and Ambrose, by Ambrose's telling, reacted like people who had just watched a goal: they jumped.
The grid is the part worth staying with. Medicine had looked inside the body before — surgery, and X-rays since 1895 — but those views were either invasive or flat. An 80×80 matrix of a brain slice, produced without opening the skull, was a different kind of seeing. The coarseness is not a flaw in the anecdote. It is the anecdote. You could still see a lesion that mattered.
Two myths stick to this machine, and both are worse than the true story. One is that it sprang finished from a single afternoon. Hounsfield had been on the problem for years. The other is that Beatles record sales paid for it. EMI had the money of a company that had sold a lot of records; the development that put scanners in hospitals leaned on the British health department as well as on EMI. The interesting fact is the wait: a clinical picture that existed only after a computer had finished the arithmetic.
Hounsfield and Allan Cormack, who had worked the mathematics from the other direction, shared the 1979 Nobel Prize in Physiology or Medicine. The first picture did not look like the scans hospitals expect now. It looked like a decision, rendered as a coarse grid, that a shadow could be turned back into a slice if you were willing to measure the head from every angle and then go away and calculate.
There is no tee for that grid in the shop. A brain pun would be a different joke. This note is the picture, and the two-day delay, and a resolution coarse enough that the milestone is still visible in the pixels.