Method and system for stabilizing a series of intravascular ultrasound images and extracting vessel lumen from the images

A computer program product, comprising:

• a computer readable medium having executable instructions recorded thereon for stabilizing ultrasonic images, wherein said instructions when executed by a processor cause the processor to:
• receive an input of reflected ultrasonic signals from a tubular environment, wherein said reflected ultrasonic signals contain information about reflective surfaces of said tubular environment;
• convert said input into a series of images;
• generate a raw circumferential image depicting a cross-section of said tubular environment from each of said images in said series;
• filter said series of images to increase signal to noise and obtain a series of refined images of said cross-sections of said tubular environment;
• detect the edges of each of said refined images using a predictive model, wherein said predictive model includes setting a probability threshold for classifying a pixel as “lumen” or “non-lumen” and calculating the probability that a pixel is “lumen-based” or “non-lumen-based”;
• construct from said detected edges a binary image of said each of said refined images, wherein each binary image corresponds to one of said circumferential raw images;
• calculate a centroid of said binary image;
• align the geometric center of each said raw image frame with its corresponding centroid.

A carrier wave embodying a computer data signal representing computer program instructions for stabilizing ultrasonic images, which when executed by a processor cause the processor to:

• receive an input of reflected ultrasonic signals from a tubular environment, wherein said reflected ultrasonic signals contain information about reflective surfaces of said tubular environment;
• convert said input into a series of images;
• generate a raw circumferential image depicting a cross-section of said tubular environment from each of said images in said series;
• filter said series of images to increase signal to noise and obtain a series of refined images of said cross-sections of said tubular environment;
• detect the edges of each of said refined images using a predictive model, wherein said predictive model includes setting a probability threshold for classifying a pixel as “lumen” or “non-lumen” and calculating the probability that a pixel is “lumen-based” or “non-lumen-based”;
• construct from said detected edges a binary image of said each of said refined images, wherein each binary image corresponds to one of said circumferential raw images;
• calculate a centroid of said binary image;
• align the geometric center of each said raw image frame with its corresponding centroid.

A system for generating stabilized images, comprising:

• a storage device;
• and
• a processor in communication with said storage device and adapted to be in communication with an ultrasonic data module, said ultrasonic data module having a transmitter for transmitting a plurality of ultrasonic signals and a receiver for receiving a plurality of reflected ultrasonic signals, said reflected signals containing information about a tubular reflecting environment;
• wherein said storage device stores a program for controlling said processor, said processor is operative with said program to:
• receive an input of reflected ultrasonic signals from a tubular environment, wherein said reflected ultrasonic signals contain information about reflective surfaces of said tubular environment;
• convert said input into a series of images;
• generate a raw circumferential image depicting a cross-section of said tubular environment from each of said images in said series;
• filter said series of images to increase signal to noise and obtain a series of refined images of said cross-sections of said tubular environment;
• detect the edges of each of said refined images using a predictive model, wherein said predictive model includes setting a probability threshold for classifying a pixel as “lumen” or “non-lumen” and calculating the probability that a pixel is “lumen-based” or “non-lumen-based”;
• construct from said detected edges a binary image of said each of said refined images, wherein each binary image corresponds to one of said circumferential raw images;
• calculate a centroid of said binary image;
• align the geometric center of each said raw image frame with its corresponding centroid.