CN107231142A - A kind of thrombelastogram instrument Adaptive Signal Processing Algorithm - Google Patents

Abstract

The invention includes an adaptive signal processing algorithm of a thrombus elastography instrument and its realization. The adaptive algorithm provided by the invention includes an adaptive filtering algorithm for coagulation signals and an adaptive smoothing algorithm for coagulation curve signals. The blood coagulation signal is sequentially processed by the above two algorithms, and the processed blood coagulation curve ensures the accuracy of the results of multiple blood coagulation parameters obtained according to the established formula. The algorithm of the present invention has the advantage of automatically identifying interference without human intervention, only filtering the coagulation signal with interference, smoothing the coagulation curve with interference, reducing the utilization rate of the processing unit, thereby improving the efficiency of the overall instrument. In addition, the algorithm of the present invention performs space compression in multiple places to reduce memory consumption.

Description

An Adaptive Signal Processing Algorithm for Thromboelastography

technical field

The invention relates to the realization of an adaptive signal processing algorithm of a thrombus elastography instrument.

Background technique

For the processing of the coagulation cycle signal, the traditional method has a sliding filter to filter out the noise interference in the signal. Since the blood coagulation cycle signal is a certain frequency within a bandwidth range (one of the fixed frequencies will be selected for different detection types or different detection substances), for the above signal characteristics, although the sliding filter will not be affected by different frequencies, it only It can effectively deal with small mutation noise, but it can’t do anything for long-term noise. When the noise interference is relatively large, it will cause signal distortion, which will affect the extraction of coagulation curve in the next step, and seriously lead to wrong calculation of coagulation parameters and provide wrong diagnostic information. . The present invention proposes an adaptive filter for the blood coagulation cycle signal with a range of frequencies used. The principle is to adaptively identify the frequency used by the current user and perform a corresponding filter with fixed coefficients. For signals with adjustable frequency, superimpose noise interference signals with different characteristics introduced by various noise sources in different environments, identify the frequency value in real time, then judge whether the noise interference exists, quantify the noise interference level, and give relevant prompts if necessary.

The purpose of the blood coagulation signal after the above processing is to extract the blood coagulation curve in the next step.

For the processing of coagulation curve, traditional methods include sliding filter and curve fitting method, but both of them have deficiencies. Sliding filtering cannot filter out large interference, and performing multiple sliding filtering may lead to curve distortion and increase computing overhead. Compared with the sliding filter, the curve fitting method has a certain improvement, but the obtained coagulation curve will change its shape according to the difference of the blood itself or the detection type, and the fitting polynomial cannot be better determined. Therefore, the coagulation curve smoothing algorithm designed in the present invention can perform smoothing according to the characteristics of the coagulation curve itself. Specifically, according to the difference between the slope of the curve and the absolute time, the number of regression analysis points is selected for smoothing.

Contents of the invention

(1) Adaptive coagulation signal filtering algorithm

1. The flow chart of the algorithm is shown in Figure 1 in the accompanying drawings.

2. Execution of the program After obtaining the blood coagulation signal collected by the AD, first identify the frequency of the signal, and then judge whether there is interference through the interference identification method 1, and then perform related processing.

The steps of adaptive filtering of blood coagulation signal are as follows:

1) Swipe to acquire a certain length of signal for buffering;

2) Adaptive identification signal frequency;

3) Identify signal interference method 1;

4) Select the corresponding fixed coefficient filter.

The steps of the blood coagulation signal interference identification method 1 are as follows:

1) Using the above-mentioned sliding buffer signal as a reference signal;

2) Calculate the energy of the signal in the cache;

3) If the energy increases or decreases steadily, it is determined that non-noise interference exists; if the energy suddenly increases or decreases, it is determined that noise interference exists;

4) Quantify according to the degree of sudden increase or decrease in capability to obtain the degree of noise interference D1, and provide corresponding prompts when necessary (D1 is greater than the preset interference threshold TH).

(2) Adaptive coagulation curve filtering algorithm

1. The algorithm flow chart is shown in Figure 2 in the accompanying drawings of the specification.

2. Execution of the program After obtaining the coagulation curve, first carry out interference identification to obtain the degree of interference D2, and then combine the real-time slope S of the curve and the absolute time T to select a nonlinear regression method, specifically to determine the size of the neighborhood in the regression method. The steps of adaptive smoothing of the coagulation curve are as follows:

1) Swipe to acquire a curve of a certain length for caching;

2) Identify signal interference method 2, quantify the interference degree D2;

3) Obtain the real-time slope S of the curve;

4) Acquire the absolute time T of the curve;

5) Select the corresponding regression analysis method in combination with D2, S, and T.

The specific selection process in the above step 5 is as follows. The purpose is to select the neighborhood for nonlinear regression analysis. The larger the neighborhood, the better the smoothing effect, but at the same time it will increase memory consumption. The present invention does not open the cache in the case of non-interference, but depends on the degree of interference in the case of interference. The specific decision is as follows:

1) Calculate the parameter P according to the three variables D2, S, and T;

2) Set two thresholds Pth1 and Pth2;

3) When P is less than or equal to Pth1, the neighborhood of nonlinear regression analysis is K1;

4) When P is greater than Pth1 and less than or equal to Pth2, the neighborhood of nonlinear regression analysis is K2;

5) When P is greater than Pth2, the neighborhood of nonlinear regression analysis is K3.

(3) The algorithm has low memory consumption and high computing efficiency

Specifically reflected in:

1. Adaptive blood coagulation signal filtering and adaptive blood coagulation curve filtering algorithm, both use a sliding window to cache signals or curves, and only filter or smooth signals with interference rather than the entire signal or curve, reducing The computing efficiency is improved on the basis of memory consumption.

2. The self-adaptive coagulation curve filtering algorithm is processed on the neighborhood selected for nonlinear regression analysis, and the degree of interference is adaptively identified, thereby dynamically selecting the size of the neighborhood.

Description of drawings

Fig. 1 is a flowchart of an adaptive coagulation signal filtering algorithm.

Fig. 2 is a flowchart of an adaptive coagulation curve filtering algorithm.

Claims (4)

1. a kind of thrombelastogram instrument Adaptive Signal Processing Algorithm design, it is characterised in that including：Adaptive coagulation signal filter Wave method, adaptive coagulation curve smoothing method.

2. this adaptive signal processing algorithm, operation efficiency is improved while relative saving memory consumption.

3. algorithm according to claim 1, adaptive coagulation signal filtering method characterized in that,

(1) AD signals are obtained, signal frequency is recognized.

(2) interference identification method 1.

(3) quantizing noise annoyance level D1, carries out related prompting if necessary.

(4) corresponding wave filter is selected to be filtered.

4. algorithm according to claim 1, its smooth feature of adaptive coagulation curve in,

(1) filtered coagulation signal is obtained, coagulation curve extraction is carried out.

(2) interference identification method 2.

(3) quantizing noise annoyance level, parameter P is obtained with reference to annoyance level D2 and the real-time slope S of curve, curve absolute time T.

(4) according to the size of parameter P values, corresponding nonlinear regression method is selected to be smoothed.