C11649: A Real-Time Noise Subtraction for FD-OCT using the GPUNovelty:
The invention is a method to perform real-time reference spectrum (noise) subtraction from Fourier-Domain Optical Coherence Tomography (FD-OCT) without an initial reference measurement.
Value Proposition:
Fourier-domain optical coherence tomography (FD-OCT) is used in a number of medical applications including Dental imaging devices. FD-OCT, however, is inhibited by the presence of noise that typically appears as horizontal streaks on the image. To eliminate this noise a measurement is made without a sample present, and is subtracted from the image when a sample is present. This calibration step is fairly simple to perform, but the noise does not stay constant. Additionally, some applications would require frequent calibrations to achieve the best image quality, which is really time consuming. This invention eliminates this step all together, by finding consistency in the noise, and subtracting it out in real time. The invention has applications across the entire OCT imaging market. However, it is worth noting that it does require specific hardware, namely a GPU. Use of a GPU in OCT was developed at JHU and can be found in C11377.
• Real-time correction during 3D imaging process.
• Easy integration with present computer and imaging software system.
• A less tedious, time saving, and cost-effective model.
Technical Details:
Johns Hopkins researcher have developled a method to perform real-time reference spectrum subtraction from Fourier-domain optical coherence tomography (FD-OCT) without a reference measurement. This removes fixed pattern noise from images produced by an FD-OCT system, without the need for a calibration step (required by traditional systems). The reference spectrum is obtained instead utilizing a Minimum Variance Mean-line Subtraction (MVMS) method. Here each line in the the image is broken up into segments. With a high number of samples, multple averages can be made of each segment. The means with the smallest variance are selected for each segment are selected, and used to reconstruct the reference spectrum for subtraction. With the use of a graphics processing unit (GPU) in an FD-OCT System, JHU Scientists were able to process images at a rate of 70 frames/s with a frame size of 1000 (lateral) by 1024 (axial) pixels.
Looking for Partners:
To develop and commercialize the technology.
Stage of Development:
Prototype tested continuously for six hours achieve high speed and high quality real-time noise corrections in FD-OCT.
Data Availability:
Under CDA / NDA
Publications/Associated Cases:
C11377
