Synchronized image capture gets to the heart of the matter

Marie Freebody, marie.freebody@photonics.com

The average human heart beats 60 to 70 times per minute, but that of a mouse can reach up to 600 to 700 beats per minute. Capturing this rapid movement on video or with still images requires some of today’s most sophisticated imaging and computer technologies.

Now, scientists funded by the UK Biotechnology and Biological Sciences Research Council and the British Heart Foundation have combined existing off-the-shelf technologies found in standard cameras and digital movie projectors to do both jobs in one, and at a price suitable for the consumer market.

“Affordable access to high-resolution and high-speed optical data, especially in a format that is compatible with existing image- and video-storage technologies, has clear potential, both for customer and specialist applications,” said Dr. Peter Kohl, who heads up the Cardiac Mechano-Electric Feedback Group at the University of Oxford. “Our work has received absolutely overwhelming attention and several companies have been in contact.”

Everyday consumer applications range from closed-circuit television to sports photography, but in the medical field, rapid image capture of cell processes could reveal more about what is happening inside a cell.

It is difficult to take sharp images or video of high-speed scenes such as a soccer match or a car race, even with a fairly decent digital single-lens reflex, because movement causes blurring. The same problem occurs in science, where vital information can be missed, including very rapid changes in intensity of light from fluorescent molecules.

The team’s work involves the use of molecular fluorescence markers to measure dynamic changes in biological functions at microscopic levels.

The beating of heart cells, for example, requires a rapid spike in calcium concentration, visible as a “calcium transient” with certain fluorescent techniques. Given the dynamics of this transient in heart cells, it must be monitored at approximately 250 Hz to get reliable data. Other processes, such as electrical potential changes, occur at even faster rates. But fast optical mapping has been something most academic research teams have been unable to afford, until the recent work by the Oxford laboratory.

Sixteen cameras in one

The team’s method was published in Nature Methods on Feb. 14, 2010, and is based on an ingenious idea by Dr. Gil Bub, a senior research fellow in the university’s department of physiology, anatomy & genetics and a member of the feedback group, who developed a new take on existing technology.

“If you wanted to take 400-fps movies using 25-fps camera technology, you could align 16 of these cameras side by side and arrange for each camera to take its frames with a defined but very short delay, one after the other. You could then take apart the 16 movies recorded, rearrange frames in the sequence they were taken, and have a 400-fps view,” Bub explained.

But, instead, the team found a solution, where pixels of a single megapixel detector are treated as if they were from separate cameras.

A pixel multiplexing camera exposes different pixels at various times during the normal frame integration time (t_i). Here, a 2 x 2-pixel grid is used (see yellow outline, a). All pixels #1 are exposed for a duration t_e (expiration time) 1/4 t_i, then all pixels #2 are exposed, and so on (b). The trick is that all the pixels are eventually exposed during t_i to record a high-resolution still image (top panel in c). In addition, you can regroup co-exposed pixels (1 … 4) into a series of smaller images that can be played back as a movie (bottom panel in c). Because the information in “still” and “movie sequence” is identical, you can obtain and store this data on a single device, using the same optics, data bus and memory as conventional cameras. Processing for video playback is straightforward and does not involve complex processing algorithms. Courtesy of Dr. Gil Bub, Cardiac Mechano-Electric Feedback Lab, Oxford University.

“By addressing individual pixels in nonoverlapping 4 x 4 patterns all over one large chip, we emulate 16 cameras in one device,” he said. “Corresponding pixels in these groups are then allowed to take their part of the bigger picture in quick succession, so that the same result – 400-fps movies from 25-fps hardware – can be achieved, with the added benefit that you also get a high-resolution still image.”

“Doing all of this on a single detector reduces both setting-up headaches and postprocessing needs,” Kohl said. “And we get the data at a fraction of the cost that solutions with similar capability normally retail at.”

The move from optical bench to a consumer-friendly package already is under way. Dr. Mark Pitter, senior research fellow at the University of Nottingham, is planning to implement the basic concept “on chip” to make the technology compact and portable. Commercialization is being led by Oxford University’s intellectual property agent Isis Innovation Ltd. of Summertown.