Researchers have developed a faster way to acquire 3D endoscopic optical coherence tomography (OCT) images. With further development, the new approach could be useful for early detection and classification of a wide range of diseases.
The new method uses computational approaches that create a full 3D image from incomplete data. In the paper "Three-dimensional endoscopic OCT using sparse sampling with a miniature magnetic driven scanning probe," published in the Optical Society journal Applied Optics, the researchers report that useful 3D images could be constructed using 40% less data than traditional 3D OCT approaches, which would decrease imaging time by 40%.
OCT is a biomedical imaging technique that has seen expanding clinical use in recent years thanks to its ability to provide high-resolution images of tissue microstructures. Today, endoscopic OCT imaging is routinely used to classify plaques and lesions in the blood vessels and is finding more use in diagnosing gastrointestinal diseases.
"Although 3D OCT images are very useful for medical diagnosis, the significant amounts of imaging data they require limits imaging speed," says research team leader Jigang Wu from Shanghai Jiao Tong University. "Our new method solves this problem by forming 3D images from much less data."
Getting the whole picture
Creating 3D OCT images with current methods requires a data-intensive process of stitching together a series of 2D images taken at equal measurements. In the new work, the researchers used a method known as sparse sampling to acquire considerably fewer 2D images and then applied compressive sensing algorithms to fill in the missing information needed to create 3D images.
The researchers tested the new method using a magnetic-driven scanning OCT probe to image inside of an extracted pigeon trachea. The probe, which the team developed previously, uses an externally driven tiny magnet to scan 360 degrees. The design minimizes the OCT scanning mechanisms enough to fit inside a device just 1.4 millimeters in diameter.
Creating 3D images of a 2-millimeter portion of the human trachea would typically require imaging every 10 microns to obtain 200 image frames. Using sparse sampling, the researchers acquired 120 frames at random positions ranging from 0 to 2 millimeters and then used the compressive sensing algorithms to create 3D images.
Moving toward the clinic
"Our tests verified that a greatly reduced amount of experimental data can be used to reconstruct reasonable 3D OCT images," says Wu. "After we perform enough experiments to demonstrate that our probe and imaging method are useful for observing malignant features, our technique will be ready for clinical trials."
The researchers plan to use their new approach to image additional biological samples related to specific diseases. They also plan to improve the endoscopic OCT probe so that it will be more robust in a variety of situations and in the context of repeated contact with biological tissues.
"This work is just one example of applying computational techniques to imaging applications," says Wu. "We expect that similar approaches may be helpful for improving the experiment designs and data acquisition for many imaging modalities."