Winfree and New York University’s Ned Seeman, which is focused on DNA computing and nanotech.

While these DNA projects have received a great deal of attention in scientific journals and even in the mainstream press, other, less well-known approaches to molecular computing might lead to entirely new computing paradigms, say researchers. One example is the work of Andrew Adamatzky, professor of unconventional computing in the department of computer science at the University of the West of England. Adamatzky’s research focuses on reaction-diffusion computing, in particular on a chemical reaction called the Belousov-Zhabotinsky (BZ) reaction, which causes waves of ions to propagate through an environment. By controlling the BZ propagation pattern, Adamatzky has shown it is possible to produce biological logic gates.

Given the proper environment, the BZ reaction can operate much like a parallel processor in which each point on the wave front, mapped to a particular grid, can serve as a point of calculation. To create a Boolean logic gate, for example, Adamatzky represents True and False by the presence or absence of a wave fragment. When two or more wave fragments collide, they fuse, dissipate, generate new wave fragments, or change their trajectory or velocity, representing the Boolean variables and implementing the computation. The trajectories of the traveling wave fragments can be changed dynamically, and adjusted and tuned by colliding other wave fragments against them, making for complex interactions. “The medium can implement such sophisticated tasks as computation of the shortest collision-free path, approximation of Voronoi diagrams of arbitrary geometrical objects, and development of a skeleton of a planar shape,” he says. “We proved the computational universality of the BZ medium by constructing a set of functionally complete logical gates in laboratory experiments. Our results indicate that the BZ system is a general-purpose parallel computer.”

But as with other work in molecular computing, several problems must be addressed in reaction-diffusion computing. Sensitivity of the BZ reaction is one of the major issues facing Adamatzky. “We design logical circuits according

future applications
include environment-
ally friendly systems
that can automatically
decompose and drug
delivery systems that
can be embedded in
human bodies.

to principles of collision-based computing, where information is represented by traveling wave fragments,” he says. “ Unfortunately, these localized excitations are unstable; they collapse or expand after some period of time.” But even with unsolved problems and unanswered questions, the experiments are proving to be useful for other fields. The core ideas of reaction-diffusion computing already have spread to those working in massively parallel computing; and even some conventional silicon processors execute wave-based algorithms.

These projects represent a small cross section of the ongoing developments in molecular computing. Unlike quantum-computing projects that require sizeable machinery and clean-room-style environments, the promise of molecular computing is that it can operate in natural environments, without electrical power. Whether scientists are able to achieve Feynman’s vision of nanoscale computers consisting of circuits and switches made of a handful of atoms remains to be seen, but researchers today are hopeful that paradigms developed through work in molecular computing will lead to entirely new applications, such as environmentally friendly systems that automatically decompose, energy-efficient machines that generate minimal amounts of heat, and biocompatible communication systems that can be embedded in human bodies.

 

based in Los angeles, Kirk L. Kroeker is a freelance editor and writer specializing in science and technology. Tatsuya suda, National science foundation, contributed to the development of this article.

Medicine
Detecting
Breast
Cancer

a large-scale british study has found that one radiologist aided by a computer is as accurate as two radiologists when it comes to detecting breast cancer in a mammogram, according to an article published online in the New England Journal of Medicine.

Mammograms are used to screen women for early signs of breast cancer, but the tests are not perfect. in the u.s. most mammograms are read by a single radiologist, while in britain, they are read by two radiologists or technicians. The british researchers, led by radiologist Fiona J. gilbert of the university of aberdeen, analyzed the results from a randomized study of 31,000 british women. it found that a radiologist aided by a computer detected 198 cancers out of 227, while a pair of radiologists detected 199 cancers.

computer-aided detection (caD) systems use computer algorithms to analyze digital mammogram images and to pick out and mark suspicious areas. caD systems are used in approximately 25% of mammogram readings in the u.s., and this percentage should increase as more medical centers switch from film x-rays to digital images.

“in the united states, it’s just not practical in most practices to do double readings by physicians,” said carol h. lee, a radiologist at Memorial sloan-Kettering cancer center in New york and head of the american college of radiology’s breast imaging commission. “These results are reassuring to me that a single reading with caD can achieve that same sensitivity.”

Double reading is the standard practice in at least 12 european countries. “up to now double reading has been the gold standard of mammography,” said Dr. Marco rosselli del Turco, president of the european society of breast cancer specialists. “We have been waiting for a well-designed, prospective, randomized study to establish the role of caD. This study provides a definitive answer about the value of adding caD to single reading, and is likely to lead to a change in european guidelines.”