Researchers tackle cervical cancer with stereovision

Kevin Robinson

Each year, more than half a million women are diagnosed with cervical cancer — the second most common cancer in women. As with most cancers, catching it early is the main key to successful treatment. The gold standard for screening is the Pap test, which is not readily available in developing countries. Researchers at Hong Kong University of Science and Technology are working on an optical method for detecting and analyzing cervical neoplasias that often lead to the disease. The method relies on using the acetowhitening effect and a motion canceling stereovision system.

“Colposcopy is the standard technique to detect early neoplastic growth in cervical tissues,” explained Jianan Y. Qu, who, with Tao T. Wu, published details of the technique in the Aug. 6 issue of Optics Express. “A fundamental part of colposcopy is the use of acetic acid to induce [the] acetowhitening effect to produce the contrast between the normal tissues and precancerous lesions,” Qu explained. The problem, he said, is that colposcopy is good at contrasting abnormal from normal tissue (about 80 to 99 percent effective) but less effective at determining whether the abnormality is cancerous.

Imaging acetowhitening

By adding image collection to the colposcope, Qu and Wu can examine the effects of actoewhitening more closely. “It has been shown that the quantitative measurement of acetowhitening kinetics at the cervix can potentially improve the diagnostic accuracy of colposcopy,” Qu explained. However, normal breathing, digestion and other automatic functions create enough movement that some sort of motion tracking system is necessary to collect the images. “Without image registration, the measurement of acetowhitening kinetics using the time-sequenced images will generate false diagnostic information because of the patient’s motion,” he added.

The researchers based their imaging setup on a standard colposcope, which is a low-power microscope, and added a channel for projecting a structured grid pattern onto the cervix. They also used the existing imaging channel to collect an image of the pattern to be reconstructed into a 3-D image. By measuring the 3-D surfaces at different times, they can track the motion of the object.

The scientists used a holographic grating and collimated light from a 100-mW laser diode operating at 660 nm, both from StockerYale Inc. The grating creates 33 stripes that are projected onto a beamsplitter, and one image is rotated 90° by a pair of mirrors.

For image collection, they employed a three-CCD color camera from Sony and a Matrox fast frame grabber. For white light imaging, they used a metal halide illuminator from Dolan-Jenner Industries Inc. and placed two polarizers configured for cross polarization in front of the light to eliminate specular reflection.

Using a human cervix model mounted on a multidimensional translational and rotational stage that can move in all three dimensions, they evaluated the system’s performance at tracking 26 “landmarks” on the cervical model as it moved a known amount in a known direction.

Using a CCD camera and a laser diode to project a structured grid on the cervix, researchers at Hong Kong University of Science and Technology can track motion ofthe cervix that then can be used to register images of a technique used to diagnoseprecancerous tissue. Courtesy of Jianan Y. Qu.

Qu explained that the information about motion tracking is used to register the time-sequenced images with each other throughout the acetowhitening process. The ratio of the registered image to the images recorded after the application of the acetic acid provides a map of the acetowhitening.

“The results show that the error of 2-D image registration is 0.8 pixels, equivalent to the motion tracking error of 0.05 mm in the field of view,” he said. “The accuracy is adequate for the measurement of temporal kinetics of the acetowhitening process from a cervical intraepithelial neoplasia.”

Qu said that, to his knowledge, this is the first application of this technique to cervical imaging, and that the technique is still in its early stages. The cervical model is widely accepted, he said, so the results will likely be applicable in actual human studies. However, the software requires 100 s to register each image, which might hamper its clinical adoption.

Future work will be aimed at developing a system that can be used in clinical studies to determine if the acetowhitening kinetics prove useful for improving the specificity of colposcopy. To reach this goal, the researchers plan to focus on improving the computer algorithms that track motion and register the images and “to develop parallel computing for fast implementation of the algorithms on a vision processor.”