There’s no question that machines are getting better at it all the time. A device to detect and identify bacterial presence in wounds or other surfaces within seconds could be ready in three years, and an inexpensive, easy to use sensor can fairly reliably identify lung cancer, even at early stages, from a patient’s breath (though it requires more work before it can be ready for general clinical use.)

Our own institution is helping to develop a miniature Raman spectroscope that can instantly distinguish malignant from non-malignant tissue in patients undergoing surgery.

* * *

Differences in protein and other content within cells of the same type could help determine the likely course of a cancer. Imaging techniques being developed to show those differences could ultimately enable doctors to make fast and accurate prognoses, and thus make better treatment decisions.

Speaking of fast, within two years MRI machines could be 10 times faster than they are today, making them even more competitive with CT.

Source article

Physicians are increasingly inundated with patient physiological data from modern diagnostic devices and tests. Discerning a particular ripple in a maelstrom of data – a ripple that indicates, say, a heart problem or an imminent epileptic seizure – is increasingly difficult without the help of computers that perceive patterns faster and more reliably than humans can.

A seizure detector was developed in 2004 at MIT and Children’s Hospital of Boston. The first detectors used brain waves of children with epilepsy before, during, and after seizures to train a classification algorithm that could then tell whether a given waveform indicated an impending seizure in the particular patient.

The team has now incorporated its earlier work into a device that automatically activates an implant that stimulates the Vagus nerve, resulting in a system that both detects and prevents an imminent seizure. Described by Technology Review writer Jennifer Chu as “essentially a bathing cap of electrodes that fits over the scalp,” as of Spring 2007 the device was being tested on a handful of patients at Beth Israel Deaconess Medical Center.

Some of the same MIT researchers are also devising algorithms to analyze EKG readings. The immediate goal is to identify key similarities and differences between the EKGs of those who survive a heart attack and those who do not. The idea is to let the computer find significant patterns, rather than telling it what to look for, and in the process perhaps uncover unexpected relationships.

Bacterial Light Detector

Source article

Scientists at Sheffield University in the UK have developed custom polymer molecules that change shape and emit light when bound to bacteria. With funding from the Ministry of Defence, they are now developing their discovery into a portable kit that could be used in wound healing, counter-terrorism, MRSA detection, and other applications.

The kit, which could be available in three years, could be used by front-line medics to determine whether a wound is infected and identify the bacterium involved. Sending samples to a laboratory can take days, but the kit would produce results almost instantly.

Breath Test for Lung Cancer

Source article

Lung cancer cells give off chemicals called volatile organic compounds (VOCs). Cleveland Clinic researchers have developed a sensor able to identify, with about 75 percent sensitivity, VOCs in the breath of people with lung cancer, including early-stage cancer. The sensor is the size of a large coin, inexpensive, and easy to use. Gas chromatography and mass spectroscopy can detect VOCs with extreme accuracy, but the machines are expensive and require considerable expertise.

Spots on the sensor change color according to the chemicals with which they come into contact.

The test will require more development before it could become available clinically.

Diagnostics – Virtual Biopsy with Raman Spectroscopy

Source article

Physicians at Children’s Hospital of Michigan, a part of the Detroit Medical Center, are validating a “virtual biopsy” technology developed at Wayne State University’s Smart Sensors and Integrated Microelectronics lab. The technology uses Raman spectroscopy to determine whether tissue is cancerous without unnecessary and painful biopsies and leaving the maximum, uninfected portion of an organ to remain intact.

A miniaturized Raman spectroscope – currently the size of a dime but eventually the width of a hair — fitted to the end of a surgical instrument instantly determines the tissue’s chemical makeup by shining light on the tissue and measuring the shift in wavelengths of the reflected light. Different materials reflect different shifts. By identifying the wavelength shift, one can identify the tissue and determine whether it is normal, benign, or malignant.

The results as of February were described by Eric Morath of the Detroit News as “encouragingly accurate.” The Raman spectroscope was about as good as pathologists at determining if tissue taken from biopsies was cancerous. The next step, to be taken later this year, is to use the technology in the operating room for real-time cancer detection while patients are in surgery.

“It’s phenomenal … you can literally tell in a few seconds,” the hospital’s chief surgeon told Morath. “You hold it against tissue and it tells you if it’s cancer, so right in the operating room you can know what is malignant and what is not.”

The same technology could be applied to determining quickly whether burn victims will need skin grafts, or to tell instantly if a skin lesion is malignant. It could eventually be used by in family doctors’ offices, if the researchers succeed in bring the cost of the spectroscopes down from their present US$200,000 to a few thousand dollars.

SSIM already has a prototype of a hair-thin scope that could be used in areas such as the lung.

Beaumont Hospital is collaborating with SSIM on lung-cancer specific studies.

Cellular Level Cancer Diagnostics

Source article

Because of the difficulty of imaging at the cellular level, cellular-level diagnostics have had to rely on cellular population-level data – statistical averages – rather than on individual cell data. A single cell can produce thousands of different proteins, lipids, hormones, and metabolites, but in order to see those elements we have to design “affinity reagents” such as anti­bodies that attach to specific proteins. In order to design the reagents, we have to know what the element looks like. This presents a classic chicken-and-egg problem. It means we can only see what we already know to exist, and it means that much of the content and processes within the cell remain a mystery.

An MIT researcher who helped develop laser-based DNA sequencers for the Human Genome Project has developed a similar laser- and microfluidics-based technique to detect proteins, lipids, and carbohydrates in individual cells. It produces graphic displays of the various amounts of different sized proteins inside individual cells. Although it does not identify the specific proteins, the analyzer’s unprecedented sensitivity and ability to show potentially critical differences between cells could be useful in determining accurate prognoses for cancer patients. It appears that as a cancer progresses, cells of the same type diverge more and more widely in their protein content. If this is so, differences between cells would show whether a cancer is more likely to spread.

The researcher told Technology Review writer Jon Cohen that “This is way early-stage, but hopefully, in 10, 20, or 30 years, people will look back and say those were interesting baby steps.”

Faster MRI

Source article

A new image-processing algorithm developed at Rice University called “compressive sensing” could make digital cameras, medical scanners, and other digital imagers “smaller and faster,” and “let them take incredibly high-resolution pictures,” writes Kate Greene in Technology Review .

The Rice researchers have developed a camera with a single image sensor that collects “just enough” information to let a novel algorithm called “compressive sensing” reconstruct a high-resolution image. It needs only a small percentage of the data that today’s digital cameras must collect in order to build a comparable picture. The algorithm “fills in the blanks” to re-create complete pictures from incomplete information.

Within two years, compressive sensing could be in MRI systems that capture images up to 10 times faster than today’s scanners. In five to 10 years, the technology could find its way into cell-phone cameras that could then produce high-quality, poster-size images.