Surgery – Diagnostic testing – Detecting nuclear – electromagnetic – or ultrasonic radiation
Reexamination Certificate
2003-04-07
2004-11-16
Jaworski, Francis J. (Department: 3737)
Surgery
Diagnostic testing
Detecting nuclear, electromagnetic, or ultrasonic radiation
C382S128000
Reexamination Certificate
active
06817982
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field
This invention pertains to methods and apparatuses related to measuring the intima-media thickness of carotid arteries to diagnose arterial sclerosis using digitally captured images from an ultrasound device inputted into and analyzed by a computer algorithm. In particular it pertains to a computer device digitally analyzing the inputted digital ultrasound image via various pixel algorithms.
2. State of the Art
A number of manual and automated measuring methods of intima-media thickness of carotid arteries to diagnose arterial sclerosis systems are known. These devices employ an ultrasound device for generating digital and/or analog images of the carotid arteries, which are then manually measured or automatically analyzed. The manual measurements are time consuming and require a trained analyst to perform the analysis. Hence, the need for a device to automatically measure and analyze the digital ultrasound images.
Masao et al., U.S. Pat. No. 6,132,373 discloses an automated intima-media thickness computer measuring apparatus and arterial sclerosis diagnosing system of the digital pixel data of ultrasound images of the carotid artery. Masao et al has the following shortcomings, which are overcome by the present invention, as described below:
Ultrasound machine requires a digital capture card that interfaces with a corresponding PCI adapter card on the PC to transmit digital images. The present invention does not require a digital capture card. It only requires some mechanism to capture images from an ultrasound machine. These mechanisms could include recording to video tape, sending the video out to a analog capture device, or sending data directly to a computer via a serial cable, USB cable, a 1394 cable, a wired or wireless network, or any other standardized communications medium.
Searching for adventitia starts from a user-specified position and only searches a pre-determined number of pixels. The thickness of the adventitia can be quite large. Searching only a predetermined number of pixels could stop short of the actual adventitia for a thick intima-media layer. The present invention starts by searching for the adventitia first by looking for the brightest portion of the image. Its search region is not limited to a predetermined number of pixels.
The Masao et al algorithm has no mention of a calibration process. This implies that the algorithm Masao et al use can only be used at a single resolution and only on the machine that the algorithm is written for. The present invention requires the user to calibrate the pixel size to ensure that any image from any ultrasound machine can be used to accurately measure IMT. Additionally, the present invention can automatically calibrate the pixel size if the image comes from a recognized ultrasound device. This is accomplished by looking at precise known locations for calibration markers that exist on images output from certain ultrasound devices. If these exact patterns are found, the calibration process is automatic, precise and error free.
The Masao et al algorithm locates the intima using a second maximum. The second highest maximal peak of image values from the lumen to the adventitia does not work if the intensity values do not have local maximum points at all, but rather gradually slope upward from the lumen to the adventitia. Applicants have seen many instances where this is the case. The present invention's algorithm solves this by looking for the largest local gradient to determine the lumen/intima boundary.
The Masao et al algorithm is very sensitive to where the user specifies the vertical location of the base position. If the user specifies a too high base position, the predetermined search distance will not search down far enough to find the adventitia. If the user specifies too low of a base position, the lumen/intima boundary may not be found correctly. The present invention solves this issue by looking first for the brightest spot in the image in a relatively large region about the user specified location. This significantly reduces the criticality of the user specified vertical location.
If the second maximal intensity peak (the peak corresponding to the intima location) is not found for a given column, the corresponding column of pixels is ignored under the Masao et al algorithm. By doing this, their algorithm could ignore a crucial location of the media/adventitia boundary, and information is lost. By throwing away information, one reduces the degree of accuracy of the measured IMT. The present invention essentially fills in these gaps at the lumen/intima boundary using neighboring column information for poor contrast regions to allow a column of data to not be ignored. This allows the media/adventitia boundary information to still be used if its data is of good quality.
Masao et al's regression curve fit algorithm uses a cubic polynomial curve fit across the entire measurement range of data. If the measurement range is small, this is fine. But if the measurement range spans multiple millimeters, this curve fit falsely restricts the shape of the intima-media layer. The present invention uses a piecewise curve-fit interpolation to allow an unbounded measurement range while still smoothing the interface boundaries and not artificially limiting the potentially tortuous nature of the intima-media layer.
Masao et al allows no user intervention for poor quality images. The present invention allows the user to intercede in the computer's behalf for poor quality images to allow accurate measurements to still take place by using the advantage of the human's visual pattern matching abilities to limit where lumen/intima and media/adventitia boundaries can lie.
Masao et al uses a single column of pixels independently to determine the location of the intima, adventitia, media/adventitia boundary and the lumen/intima boundary locations are. The present invention's algorithm finds the location of the adventitia by using adjacent column's pixel intensity information. This significantly reduces false adventitia location errors due to noise and/or poor contrast in the digital image. It also tracks the bottom of the media layer using neighboring pixel information to assist in segmenting the search region for the lumen/adventitia boundary from the media/adventitia boundary.
Masao et al determines the IMT by averaging only three values: the maximum IMT value, and two other non-maximal IMT values. The present invention uses the distance from the lumen/intima boundary to the media/adventitia boundary for every column in the measurement region to more accurately determine the true value of the IMT. This approach takes into account all portions of the tortuous shape of the intima-media along the entire measurement range, which directly improves the measurement accuracy.
Masao et al addresses none of the 8 operator associated factors. Without compensating for these factors, the ultimate accuracy of IMT measurements is significantly reduced. The present invention discusses these factors and describes how to minimize the effects.
Pitsillides et al., U.S. Pat. No. 5,544,656 discloses a closed-loop single-crystal ultrasonic sonomicrometer capable of identifying the myocardial muscle/blood interface and continuously tracking this interface through the cardiac cycle using a unique piezoelectric transducer. The echoes from the transducer are amplified and amplified and applied to a Doppler decoder, which analyzes the signals to determine myocardial wall thickness throughout the cardiac cycle. Criton et al., U.S. Pat. No. 5,800,356 discloses an ultrasonic diagnostic imaging system with Doppler assisted tracking of tissue motion providing a method for tracing the border of tissue through temporarily acquired scan lines comprising the steps of reducing noise in the scan lines, producing a map of tissue edges from the scan lines, denoting a tissue border to be traced, and using velocity information corresponding to tissue edges to trace the denoted border. Sumanaweera et al., U.S. Pa
Fritz Helmuth
Fritz Terry
Fulbright & Jaworski L.L.P.
Jaworski Francis J.
Sonosite, Inc.
LandOfFree
Method, apparatus, and product for accurately determining... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method, apparatus, and product for accurately determining..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method, apparatus, and product for accurately determining... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3346571