CN-122023533-A - Anaesthetic puncture auxiliary positioning method based on artificial intelligence

CN122023533ACN 122023533 ACN122023533 ACN 122023533ACN-122023533-A

Abstract

The invention relates to the technical field of ultrasonic image processing, in particular to an artificial intelligence-based anesthesia puncture auxiliary positioning method. The method comprises the steps of firstly extracting resistance factors of acoustic wave transmission of pixel points, further capturing the accumulated deflection angle of each historical frame relative to a current frame based on an optical flow algorithm, then constructing a time-space aligned resistance data body, further obtaining a corrected resistance map, extracting an acoustic wave reaching the accumulated resistance map and a minimum resistance track, further projecting the minimum resistance track back into each time slice in the resistance data body, obtaining an instantaneous total resistance sequence of each deep pixel point, further obtaining a resistance-angle change gradient according to data fluctuation in the instantaneous total resistance sequence and combining the distribution of the accumulated deflection angle, and finally distinguishing real intervertebral space and artifacts in an ultrasonic image with low signal-to-noise ratio according to the acoustic wave reaching the accumulated resistance map, the resistance-angle change gradient and the instantaneous total resistance sequence, so as to generate reliable positioning guide prompt information.

Inventors

LV JUNTAO
Dang Zhimei
ZHANG CHANGLE
YAO QING

Assignees

西安大兴医院

Dates

Publication Date: 20260512
Application Date: 20260413

Claims (10)

1. An artificial intelligence-based anesthesia puncture auxiliary positioning method is characterized by comprising the following steps: Acquiring a current image sequence to be analyzed, acquiring a resistance factor of acoustic wave transmission of each pixel point based on a gray value of each pixel point in the image, capturing pixel displacement characteristics between adjacent frames based on an optical flow algorithm, and acquiring an accumulated deflection angle of each historical frame relative to a current frame by combining the gray characteristics of the pixel points; Acquiring a resistance data body, acquiring a correction resistance map by combining spatial distribution of pixel points according to the distribution of resistance factors in the resistance data body along a time axis, extracting a minimum resistance track of sound waves reaching a cumulative resistance map and each deep pixel point at the deep part of a preset image from the correction resistance map, projecting the minimum resistance track back into each time slice in the resistance data body, acquiring an instantaneous total resistance sequence of each deep pixel point, and acquiring a resistance-angle change gradient by combining the distribution of the cumulative deflection angle according to data fluctuation in the instantaneous total resistance sequence; Generating a puncture target probability distribution map according to the sound wave arrival accumulated resistance map and the resistance-angle change gradient, and generating positioning guide prompt information by combining the accumulated deflection angle based on the puncture target probability distribution map and the instantaneous total resistance sequence.
2. The anesthesia puncture auxiliary positioning method based on artificial intelligence according to claim 1, wherein the method for acquiring the accumulated deflection angle comprises the following steps: Acquiring a pixel level displacement field of each frame of image relative to the image of the frame before the serial number by using a dense optical flow algorithm; based on the displacement components of each pixel point in each frame of image in the horizontal direction, the rotation angle weighting weights and the preset conversion coefficients, acquiring the instantaneous rotation angle of each frame of image, and based on all the instantaneous rotation angles, acquiring the accumulated deflection angle of each historical frame relative to the current frame.
3. The anesthesia puncture auxiliary positioning method based on artificial intelligence according to claim 1, wherein the resistance data body acquisition method comprises the following steps: And according to the accumulated deflection angle corresponding to each historical frame, carrying out inverse rotation transformation on the resistance factors of the pixel points in the historical frames by taking a preset rotation axis as a reference to obtain aligned resistance slices aligned with a space coordinate system of the current frame, and constructing three-dimensional resistance data bodies according to time index sequences by using the slices formed by the resistance factors of the current frame and the aligned resistance slices corresponding to all the historical frames.
4. The anesthesia puncture auxiliary positioning method based on artificial intelligence according to claim 1, wherein an acoustic wave arrival accumulation resistance map is obtained by using a fast traveling algorithm based on the correction resistance map, and gradient descent backtracking is performed on the acoustic wave arrival accumulation resistance map for each deep pixel point to obtain a minimum resistance track.
5. The method for assisting positioning of anesthesia puncture based on artificial intelligence according to claim 1, wherein the method for acquiring the gradient of resistance-angle change comprises the steps of: Acquiring the angular range of the accumulated deflection angle in the image sequence, and when the angular range is smaller than a preset angular threshold value, adopting the resistance-angle change gradient at the previous moment; when the angular range is greater than or equal to a preset angular threshold, according to the difference of two adjacent data in the instantaneous total resistance sequence of each deep pixel point, combining the difference of two adjacent accumulated deflection angles to obtain a resistance-angle change gradient.
6. The anesthesia puncture auxiliary positioning method based on artificial intelligence according to claim 1, wherein the puncture target probability distribution map acquisition method comprises the following steps: The method comprises the steps of respectively obtaining statistical means of the sound wave arrival cumulative resistance diagram and the resistance-angle change gradient in a statistics mode, obtaining accessibility probability based on the sound wave arrival cumulative resistance diagram and the statistical means thereof, obtaining stability probability based on the resistance-angle change gradient and the statistical means thereof, and carrying out fusion calculation on the accessibility probability, the stability probability and a preset depth mask to generate a puncture target probability distribution diagram.
7. The method for assisting in positioning anesthesia puncture based on artificial intelligence according to claim 1, wherein the method for generating positioning guidance prompt information comprises the following steps: Determining a target guide point in the puncture target probability distribution map, determining a minimum resistance moment according to the instantaneous total resistance sequence corresponding to the target guide point, and generating positioning guide prompt information based on the accumulated deflection angle corresponding to the minimum resistance moment.
8. The anesthesia puncture auxiliary positioning method based on artificial intelligence according to claim 7, wherein a correlation coefficient of one frame image and a current frame image of the minimum resistance moment is obtained, and when the correlation coefficient is larger than or equal to a preset recommended threshold value, a difference value between the accumulated deflection angle corresponding to the minimum resistance moment and the accumulated deflection angle of the current frame is calculated and used as a reset guiding deflection angle; and generating positioning guide prompt information according to the distribution of the reset guide deviation angle relative to a preset dead zone range.
9. The method for assisting positioning of anesthesia puncture based on artificial intelligence according to claim 1, wherein the method for acquiring the corrected resistance map comprises the steps of: And acquiring a correction resistance map by combining the transverse deviation of each pixel point and a preset central axis based on the average acoustic wave resistance of each pixel point in the average acoustic wave resistance distribution map.
10. The method for assisting positioning of anesthesia puncture based on artificial intelligence according to claim 1, wherein the method for acquiring the resistance factor comprises the following steps: When the gray value of the pixel point is smaller than or equal to the preset first gray threshold value, the positive correlation mapping of the gray value is carried out to obtain the resistance factor.

Description

Anaesthetic puncture auxiliary positioning method based on artificial intelligence Technical Field The invention relates to the technical field of ultrasonic image processing, in particular to an artificial intelligence-based anesthesia puncture auxiliary positioning method. Background In clinical practice in anesthesia, intrathecal penetration procedures rely on the accurate positioning of the intervertebral space. For obese patients, the superficial bony landmarks are difficult to reach due to the significant thickening of the subcutaneous fat layer, usually with ultrasound guidance. However, the strong scattering and attenuation of ultrasound waves by adipose tissue results in extremely low signal-to-noise ratio of deep anatomical imaging, the image is generally "foggy" blurred, the actual intervertebral space often appears as a weak hypoechoic channel, and the sound and multiple reflection artifacts generated by bone edges, calcifications or fibrous spaces also appear as morphologically similar dark areas, which are difficult to reliably distinguish only by a single frame of still image. Clinical experience shows that by dynamic sector scanning, the difference of the response of the real anatomical structure and the artifact to the change of the angle of the probe can be utilized to distinguish, namely, the real gap is kept relatively stable under the fine adjustment of the angle, and the artifact easily flashes severely or disappears along with the change of the angle. However, the method is highly dependent on the experience of an operator and the coordination ability of hands and eyes, and the optimal imaging visual angle is often about to be elapsed, so that the probe is difficult to accurately memorize and reproduce while being stable. Disclosure of Invention In order to solve the technical problem that the true intervertebral space is difficult to accurately position due to static feature confusion in the low signal-to-noise ratio ultrasonic image, the invention aims to provide an artificial intelligence-based anesthesia puncture auxiliary positioning method, which adopts the following technical scheme: Acquiring a current image sequence to be analyzed, acquiring a resistance factor of acoustic wave transmission of each pixel point based on a gray value of each pixel point in the image, capturing pixel displacement characteristics between adjacent frames based on an optical flow algorithm, and acquiring an accumulated deflection angle of each historical frame relative to a current frame by combining the gray characteristics of the pixel points; Acquiring a resistance data body, acquiring a correction resistance map by combining spatial distribution of pixel points according to the distribution of resistance factors in the resistance data body along a time axis, extracting a minimum resistance track of sound waves reaching a cumulative resistance map and each deep pixel point at the deep part of a preset image from the correction resistance map, projecting the minimum resistance track back into each time slice in the resistance data body, acquiring an instantaneous total resistance sequence of each deep pixel point, and acquiring a resistance-angle change gradient by combining the distribution of the cumulative deflection angle according to data fluctuation in the instantaneous total resistance sequence; Generating a puncture target probability distribution map according to the sound wave arrival accumulated resistance map and the resistance-angle change gradient, and generating positioning guide prompt information by combining the accumulated deflection angle based on the puncture target probability distribution map and the instantaneous total resistance sequence. Further, the method for acquiring the cumulative deflection angle comprises the following steps: Acquiring a pixel level displacement field of each frame of image relative to the image of the frame before the serial number by using a dense optical flow algorithm; based on the displacement components of each pixel point in each frame of image in the horizontal direction, the rotation angle weighting weights and the preset conversion coefficients, acquiring the instantaneous rotation angle of each frame of image, and based on all the instantaneous rotation angles, acquiring the accumulated deflection angle of each historical frame relative to the current frame. Further, the method for acquiring the resistance data volume includes: And according to the accumulated deflection angle corresponding to each historical frame, carrying out inverse rotation transformation on the resistance factors of the pixel points in the historical frames by taking a preset rotation axis as a reference to obtain aligned resistance slices aligned with a space coordinate system of the current frame, and constructing three-dimensional resistance data bodies according to time index sequences by using the slices formed by the resistance factors of the current frame a