CN115299946B - Self-adaptive input signal channel screening circuit introducing detection result feedback - Google Patents
Self-adaptive input signal channel screening circuit introducing detection result feedback Download PDFInfo
- Publication number
- CN115299946B CN115299946B CN202211028703.6A CN202211028703A CN115299946B CN 115299946 B CN115299946 B CN 115299946B CN 202211028703 A CN202211028703 A CN 202211028703A CN 115299946 B CN115299946 B CN 115299946B
- Authority
- CN
- China
- Prior art keywords
- screening
- signal
- unit
- detection
- calculation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 100
- 238000012216 screening Methods 0.000 title claims abstract description 95
- 238000004364 calculation method Methods 0.000 claims abstract description 68
- 238000000605 extraction Methods 0.000 claims abstract description 15
- 238000012163 sequencing technique Methods 0.000 claims abstract description 9
- 239000013598 vector Substances 0.000 claims description 16
- 238000010606 normalization Methods 0.000 claims description 12
- 230000003044 adaptive effect Effects 0.000 claims description 8
- 239000000872 buffer Substances 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 5
- 238000009825 accumulation Methods 0.000 claims description 3
- 230000005587 bubbling Effects 0.000 claims description 3
- 230000001174 ascending effect Effects 0.000 claims description 2
- 230000000694 effects Effects 0.000 abstract description 5
- 230000001419 dependent effect Effects 0.000 abstract 1
- 230000001037 epileptic effect Effects 0.000 description 20
- 238000010586 diagram Methods 0.000 description 6
- 238000005070 sampling Methods 0.000 description 6
- 238000013527 convolutional neural network Methods 0.000 description 4
- 206010015037 epilepsy Diseases 0.000 description 4
- 230000008901 benefit Effects 0.000 description 2
- 210000004556 brain Anatomy 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 238000011895 specific detection Methods 0.000 description 2
- 206010010904 Convulsion Diseases 0.000 description 1
- 208000019914 Mental Fatigue Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/16—Devices for psychotechnics; Testing reaction times ; Devices for evaluating the psychological state
- A61B5/165—Evaluating the state of mind, e.g. depression, anxiety
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/16—Devices for psychotechnics; Testing reaction times ; Devices for evaluating the psychological state
- A61B5/168—Evaluating attention deficit, hyperactivity
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/369—Electroencephalography [EEG]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/369—Electroencephalography [EEG]
- A61B5/372—Analysis of electroencephalograms
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/40—Detecting, measuring or recording for evaluating the nervous system
- A61B5/4076—Diagnosing or monitoring particular conditions of the nervous system
- A61B5/4094—Diagnosing or monitoring seizure diseases, e.g. epilepsy
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Public Health (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Molecular Biology (AREA)
- Psychiatry (AREA)
- Veterinary Medicine (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Artificial Intelligence (AREA)
- Psychology (AREA)
- Physiology (AREA)
- Neurology (AREA)
- Developmental Disabilities (AREA)
- Child & Adolescent Psychology (AREA)
- Educational Technology (AREA)
- Neurosurgery (AREA)
- Hospice & Palliative Care (AREA)
- Social Psychology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Complex Calculations (AREA)
Abstract
The invention discloses a self-adaptive input signal channel screening circuit introducing detection result feedback, and belongs to the field of signal processing. The invention comprises a feature extraction unit, a similarity calculation unit and a sequencing unit; the channel screening circuit is used for carrying out channel screening on input signals and sending channel screening results to the signal detection unit so that the signal detection unit only carries out two kinds of detection tasks on the signals of the screened channels, meanwhile, the detection results of the signal detection unit are introduced, and the channel screening results are adaptively adjusted based on the detection results so as to achieve better screening effects. The invention feeds back the detection result to the channel screening part, so that the information provided by the detection result can be utilized when the channel screening is carried out, the calculation amount is not required to be additionally increased, the channel screening is more dependent, the channel screening is more accurate, and the influence on the detection accuracy is minimized. Therefore, the signal detection unit is reduced in calculation amount for performing classification detection tasks, and power consumption is reduced.
Description
Technical Field
The invention belongs to the field of signal processing, and particularly relates to a self-adaptive input signal channel screening circuit for introducing detection result feedback.
Background
Signal detection is a common signal processing task, such as epileptic detection based on electroencephalogram signals, and electrocardiographic classification based on multi-lead electrocardiograph signals all involve multiple input signal channels, and in the past, all channels are directly used for detection, but not all channels contain effective information of a specific detection task, and the direct use of all channels increases complexity but does not necessarily improve detection effect, and in the following, epileptic detection is taken as an example.
Because of the burstiness and unpredictability of epilepsy, the seizure frequency of epilepsy varies from years to times a day, so that in order not to miss any epilepsy which may occur at any time, the detection must be monitored in real time for a long period of time, which requires hardware for realizing the detection function not only to have high detection accuracy but also to have low power consumption characteristics that can be detected for a long period of time. The number of channels for acquiring the electroencephalogram signals can be as large as 36, and if all channel data are directly used for epileptic detection without distinction, the two problems are brought. First, the brain electrical signals are a concentrated reflection of the massive neuronal cell re-discharge activity conducted to the cortex, and different acquisition channels correspond to electrodes placed at different locations of the cortex. There may be redundancy in physiological information reflected by adjacent electrodes in some states, or the amount of information contained by electrodes in areas where some neurons are not active is inherently weak. Screening for effective channels is important because if these channels are taken directly for epileptic detection without distinction, the detection effect may be rather reduced. Secondly, taking all channels to perform epileptic detection inevitably increases the calculation amount of detection greatly, which is contrary to the requirement of low power consumption, so that eliminating redundant channels as much as possible on the premise of ensuring detection accuracy is important for realizing long-time monitoring with lower power consumption.
From the above, efficient and intelligent channel screening is of great importance for signal detection.
Disclosure of Invention
The invention aims at: aiming at the problems, the self-adaptive input signal channel screening circuit which introduces the feedback of the detection result is provided, so that partial channels are screened out on the premise of ensuring the follow-up detection accuracy, the calculated amount of detection is reduced, and the power consumption is reduced.
The invention adopts the technical scheme that:
An adaptive input signal path screening circuit incorporating detection result feedback, comprising: the device comprises a feature extraction unit, a similarity calculation unit and a sequencing unit; the self-adaptive input signal channel screening circuit provided by the invention is used for carrying out channel screening on input signals and sending the channel screening result to the signal detection unit so that the signal detection unit only carries out two kinds of detection tasks on the signals of the screened channels, and meanwhile, the self-adaptive input signal channel screening circuit introduces the detection result of the signal detection unit and self-adaptively adjusts the channel screening result based on the detection result so as to achieve a better screening effect; the self-adaptive input signal channel screening circuit provided by the invention can be matched with various detection units, so that partial channels are screened out on the premise of ensuring the follow-up detection accuracy, the calculation amount of the signal detection units for executing classification detection tasks is reduced, and the power consumption is reduced.
Wherein the input signal of the feature extractor unit comprises a plurality of channels, the signal of one channel is input at a time, and the input signal is a positive sample signal or a negative sample signal (namely, the positive and the negative of the sample signal are used for representing two categories of the classification detection task of the signal detection unit);
The characteristic extractor unit comprises multiple paths of calculation paths, a switch unit and a multiplexer, wherein the switch unit is respectively connected with each path of calculation path, the output end of each path of calculation path is connected with the multiplexer, the switch unit is used for controlling the number of samples of a continuous input positive sample signal and the number of samples of a continuous input negative sample signal, each path of calculation path is used for calculating a time-frequency domain characteristic of the input signal, so that the time-frequency domain characteristic vector of the input signal is obtained through the output of all calculation paths, and the time-frequency domain characteristic vector of the input signal is respectively registered to a designated storage unit through the multiplexer according to the time-frequency domain characteristic vector of the positive sample and the time-frequency domain characteristic vector of the negative sample; triggering a similarity calculation unit to calculate the screening weight of the channel when the number of the samples of the input positive and negative sample signals meets the requirement;
The similarity calculation unit reads all positive sample time-frequency domain feature vectors and all negative sample time-frequency domain feature vectors from the storage unit, and obtains a positive sample feature template and a negative sample feature template based on the average value of all positive and negative sample time-frequency domain feature vectors; the screening weight of the current channel of the input signal is represented and cached based on the distance between the positive and negative sample feature templates, and after the screening weight of all channels of the input signal is obtained, the screening weight is sent to the sequencing unit;
The sorting unit is used for sorting the screening weights of all the channels in ascending order, taking channel identifiers corresponding to the first K screening values as channel screening results, and sending the channel identifiers to the signal detection unit;
the signal detection unit reads signal data of a channel corresponding to a signal to be detected based on a current channel screening result, performs a two-class detection task, feeds back the detection result to the feature extractor unit, and the switching unit of the feature extractor unit controls whether signals of all channels of the current signal to be detected are sequentially input to all paths of calculation paths as input signals based on the detection result: if the category of the detection result is matched with the category requirement to be input into the calculation path, the switch unit is turned on, otherwise, the switch unit is turned off; and the multiplexer of the feature extractor unit determines the storage unit of the time-frequency domain feature vector of the input signal based on the current detection result.
Further, when the signal detection unit performs the two classification detection tasks based on the neural network, the channel screening result includes a channel number K that matches the input channel number of the input layer of the neural network.
Further, the feature extractor unit comprises four paths of calculation paths for calculating four time-frequency domain features of zero crossing points, areas, attenuation and line lengths of the input signals respectively.
Further, the four paths of computation paths of the feature extractor unit are respectively:
The zero crossing calculation path sequentially comprises two comparators, an exclusive-OR gate, a multiplier and an accumulator, and is used for carrying out point-by-point calculation and accumulation on each sampling point of a current input signal to obtain zero crossing point characteristics of the current sample signal (positive sample signal or negative sample signal), one comparator is used for comparing the input data at the current moment with 0, the other comparator is used for comparing the input data at the previous moment with 0, if the comparison result of the comparator is greater than 0, the output result of the comparator is 1, if the comparison result of the comparator is less than 0, the output result of the comparator is 0, the exclusive-OR result is 1, the data at the current moment and the data at the previous moment are represented by the exclusive-OR result, namely, a zero crossing point exists, the exclusive-OR result is multiplied by a preset normalization coefficient alpha 1, and then accumulation is carried out, and the zero crossing point characteristics of the current sample signal are obtained after all sampling points in the current sample signal (positive sample signal or negative sample signal) are calculated;
The area calculation path sequentially comprises an adder, a multiplier and an accumulator, wherein the adder is used for adding the input data at the current moment and the input data at the previous moment, the multiplier is used for multiplying the added result by a normalization coefficient alpha 2, the accumulator is used for accumulating all sampling points, and the area characteristics of the sample signal are obtained after all sampling points in the sample signal are calculated;
The attenuation calculation path sequentially comprises a comparator, a multiplier and an accumulator, wherein the comparator is used for comparing the input data at the current moment with the input data at the previous moment, multiplying the comparison result by a normalization coefficient alpha 3 and then accumulating, and the attenuation characteristics of the sample signal are obtained after all sampling points in the sample signal are calculated;
The line length calculation path sequentially comprises a subtracter, an absolute value calculation unit, a multiplier and an accumulator, wherein the input data at the current moment is subtracted from the input data at the previous moment, the absolute value of the subtraction result is calculated, the subtraction result is multiplied by a normalization coefficient alpha 4, and the subtraction result is accumulated, so that the line length characteristics of the sample signal are obtained after all sampling points in the sample signal are calculated.
Further, the similarity calculation unit calculates the screening weight of each channel through the cosine distance between the positive sample characteristic template and the negative sample characteristic template.
Further, the similarity calculation unit comprises a plurality of similarity sub-calculation paths, and is used for cosine distance calculation between each dimension feature of the positive and negative sample feature templates of the current channel, and the cosine distance calculation result of each similarity sub-calculation path is accumulated to obtain the screening weight of the current channel.
Further, each similarity sub-calculation path sequentially comprises a subtracter and a multiplier; each dimension characteristic of the positive and negative sample characteristic templates is subtracted by a subtracter, and then square is obtained by a multiplier, namely the cosine distance calculation of different channels is time-division multiplexed by the same set of cosine distance calculation logic.
Further, the sorting unit sorts the screening weights of the channels in a bubbling sorting mode.
Further, the sorting unit comprises two screening weight caches (when cosine distances are adopted to calculate screening weight values of all channels, the two caches are cosine distance caches), a comparator, three multiplexers and a decomposer, wherein the first screening weight cache is used for storing cosine distances of all channels, the second screening weight cache is used for storing channel screening results, the two-in-one multiplexers select adjacent elements from the first screening weight cache to enter the comparator for comparison, the comparison results respectively enter the two-in-one selectors, the two-in-one selectors respectively select one maximum element to enter the two-in-one decomposer, and the output end of the two-in-one decomposer is connected to the second screening weight cache.
When the sorting unit performs bubble sorting, adjacent elements are compared, if the first one is larger than the second one, the two elements are exchanged, and the same work is performed on each pair of adjacent elements, namely, the last pair from the first pair to the end; the last element should be the largest number. The above steps are repeated for all elements except the last one. The above steps continue to be repeated for fewer and fewer elements at a time until no pair of numbers need to be compared.
The technical scheme provided by the invention has at least the following beneficial effects:
(1) The influence of channel screening on detection accuracy is minimized: by feeding the detection result back to the channel screening part, the information provided by the detection result can be utilized when the channel screening is carried out, the calculation amount is not required to be additionally increased, the channel screening can be more dependable, the channel screening is more accurate, and the influence on the detection accuracy is minimized.
(2) The channel screening operand is low, and the power consumption of a detection system is reduced: four features which are very effective in channel screening are analyzed and selected, and the calculated amount for extracting the features is small; in addition, by exploring the most effective channel, the invention can reach the same level of detection accuracy with fewer channels, thus greatly reducing the calculation amount of the subsequent classifier, and having more remarkable power consumption benefit for the whole system; the invention also optimizes the power consumption of the hardware implementation of channel screening. This results in a relatively low power consumption of the channel screening hardware according to the invention.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic diagram of a hardware architecture of an adaptive channel screening circuit of the present invention for introducing result feedback in a specific embodiment;
FIG. 2 is a schematic diagram of a feature extraction unit according to the present invention in an embodiment;
FIG. 3 is a schematic diagram of a similarity calculation unit according to the present invention in an embodiment;
fig. 4 is a schematic diagram of a sorting unit according to the present invention in an embodiment.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the embodiments of the present invention will be described in further detail with reference to the accompanying drawings.
The embodiment of the invention provides a self-adaptive input signal channel screening circuit for introducing detection result feedback, which comprises a feature extraction unit, a similarity calculation unit and a sequencing unit, wherein the structural schematic diagram of the channel screening circuit is shown in fig. 1, the feature extraction unit introduces detection results for calculating four feature values of input signals (such as brain signals), then a cosine distance calculation unit calculates cosine distances of all channels by utilizing features obtained by the feature extraction unit, and finally the cosine distance sequencing unit is used for sequencing the cosine distances of all channels, and after sequencing is completed, partial channels in front are taken as final channel screening results.
The following describes in further detail the adaptive input signal channel screening circuit for introducing feedback of detection results provided by the embodiment of the present invention, taking multi-channel screening for epileptic detection as an example.
In the invention, the channel screening circuit needs to introduce the classification detection result of the signals so as to generate the corresponding channel screening result according to the detection result in a further self-adaptive manner.
That is, the channel screening result output by the channel screening circuit is sent to the signal detection unit (for detecting whether the current user is epileptic, of course, the signal detection unit related to the present invention can also be used for other two kinds of detection tasks (for example, whether to depression, whether to mental fatigue, whether to have a target object, etc.), and the setting can be adjusted based on the actual application requirement.
For the acquired signal to be detected, since the acquired signal comprises a plurality of channels, if all channel data are input to the signal detection unit for detection processing, higher computational complexity is required to be caused, so that when the two classification detection processing tasks related to the signal class data are faced, the channel screening circuit provided by the embodiment of the invention can determine the optimal channels, and the signal detection unit only needs to execute detection processing on the data of the optimal channels, so that the processing complexity of the detection task is reduced.
For the two-class detection task, the classification detection structure is set to be positive and negative, in the initial stage, the initial channel screening result is obtained by the channel screening circuit directly based on the sample signal (positive sample signal or negative sample signal) of the known class label, so that the signal detection unit can execute the two-class detection processing of the signal to be detected based on the data of the channel corresponding to the reading of the channel screening structure, and then the signal to be detected is set to be the positive or negative sample signal based on the detection result, thereby realizing the updating of the input sample and further obtaining the more applicable channel screening result in a self-adaptive manner.
In this embodiment, the feature extractor unit is configured to calculate four feature values of the electroencephalogram signal, and the structure diagram is shown in fig. 2, and fig. 2 is an example of channel screening for M channels. Each feature is computed for n samples and normalized to make the feature more representative. The four features are Zero-crossing (Zero-crossing), area (Area), attenuation (Decay), and Line length (Line-length), respectively, and the four features are calculated separately in four calculation passes. The zero-crossing calculation path comprises two comparators, an exclusive-OR gate, a multiplier and an accumulator, wherein the current input data and the last-moment input data are respectively compared in size of 0, the comparison result is 1 when the comparison result is larger than 0, the comparison result is 0 when the comparison result is smaller than 0, the output result is 0, the results output by the two comparators are exclusive-or, and the exclusive-OR result is 1, and the data at the current moment and the data at the last moment are different in number, namely, a zero-crossing point exists. And then multiplying the exclusive OR result by the normalization coefficient, accumulating, and obtaining the zero crossing point characteristic of the sample after all points in the sample are calculated. The area calculation path comprises an adder, a multiplier and an accumulator, the input of the current moment is added with the input of the previous moment, the added result is multiplied by the normalization coefficient, and then the summation is carried out, and the area characteristics of the sample are obtained after all points in the sample are calculated. The attenuation calculation path comprises a comparator, a multiplier and an accumulator, the current input data and the data at the last moment are compared in size, the comparison result is multiplied by the normalization coefficient and then accumulated, and the attenuation characteristic of the sample is obtained after all points in the sample are calculated. The linear length calculation path comprises a subtracter, an absolute value calculation unit, a multiplier and an accumulator, the current input data and the input data at the previous moment are subtracted, the absolute value of the subtraction result is calculated, the subtraction result is multiplied by a normalization coefficient and then accumulated, and the linear length characteristics of the sample are obtained after all points in the sample are calculated. In addition, the feature extraction unit further comprises a switch unit and a multiplexer, wherein the switch unit controls to enter a specified number of non-epileptic samples first and then enter a specified number of epileptic samples (or enter a specified number of epileptic samples first and then enter a specified number of non-epileptic samples) so as to average the features of a plurality of samples of the same type. The multiplexer registers the characteristic calculation results according to the current sample types respectively according to epilepsy and non-epileptic. Alpha 1、α2、α3 and alpha 4 in fig. 2 are normalized coefficients, and the preset value is specific to the actual application scenario.
The similarity calculation unit is used for calculating screening weights of all channels, and the structure of the similarity calculation unit is shown in fig. 3 and comprises four subtractors, four adders and an accumulator. The screening weight of each channel is characterized by cosine distances among four characteristics of two types of epileptic and non-epileptic samples, after the cosine distance calculation unit receives the characteristics of the epileptic and non-epileptic samples extracted by the characteristic extractor unit, the cosine distances among the four characteristics of the epileptic sample and the four characteristics of the non-epileptic sample are respectively calculated according to the channels, namely, the four characteristics of the epileptic and the non-epileptic are subtracted by a subtracter respectively, square is calculated by a multiplier, and then the square results of the four characteristics are accumulated to obtain the cosine distance of one channel, and the cosine distances of different channels are used as the screening weight of the channel, and the cosine distances of different channels are calculated to multiplex the same set of cosine distance calculation logic in a time sharing mode.
The sorting unit is configured to sort cosine distances of channels, as shown in fig. 4, where fig. 4 is an example of screening K channels (K < M) from M channels, and includes two cosine distance buffers, a comparator, and four multiplexers. The sorting adopts a bubbling sorting mode, and adjacent elements are compared. If the first is larger than the second, they are swapped. The same is done for each pair of adjacent elements, from the first pair to the last pair of the end. At this point, the last element should be the largest number. The above steps are repeated for all elements except the last one. The above steps continue to be repeated for fewer and fewer elements at a time until no pair of numbers need to be compared. And taking the previous K channels as channel screening results after the sorting is completed.
In the invention, the signal detection unit is used for carrying out classification detection on the signals, the detection result comprises a positive type and a negative type, and the detection result is fed back to the feature extraction unit. The invention can provide the best channel selection for the classifier and can be matched with a plurality of specific detection units, such as a common convolutional neural network. When the convolutional neural network is used as a classifier, the convolutional neural network only needs to screen out the optimal channels with the same number as the input channels of the neural network, and the convolutional neural network can extract the characteristics of the input samples through a plurality of convolutional layers, select and recombine the characteristics through a full connection layer and finish classification.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
What has been described above is merely some embodiments of the present invention. It will be apparent to those skilled in the art that various modifications and improvements can be made without departing from the spirit of the invention.
Claims (5)
1. An adaptive input signal path screening circuit incorporating feedback of a detection result, comprising: the device comprises a feature extraction unit, a similarity calculation unit and a sequencing unit;
the input signal of the characteristic extraction unit comprises a plurality of channels, the signal of one channel is input each time, the input signal is a positive sample signal or a negative sample signal, and the positive and negative of the sample signal are used for representing two categories of the two-category detection task of the signal detection unit;
The characteristic extraction unit comprises four paths of calculation paths, a switch unit and a multiplexer, wherein the switch unit is respectively connected with each path of calculation path, the output end of each path of calculation path is connected with the multiplexer, the switch unit is used for controlling the number of samples of a continuous input positive sample signal and the number of samples of a continuous input negative sample signal, each path of calculation path is used for calculating a time-frequency domain characteristic of the input signal, the time-frequency domain characteristic comprises zero crossing point, area, attenuation and line length, the time-frequency domain characteristic vector of the input signal is obtained based on the output of all calculation paths, and the time-frequency domain characteristic vector of the input signal is respectively registered to a designated storage unit through the multiplexer according to the time-frequency domain characteristic vector of the positive sample and the time-frequency domain characteristic vector of the negative sample; triggering a similarity calculation unit to calculate the screening weight of the channel when the number of the samples of the input positive and negative sample signals meets the requirement;
the four paths of calculation paths of the feature extraction unit are respectively as follows:
the zero crossing point calculation path sequentially comprises two comparators, an exclusive-OR gate, a multiplier and an accumulator, wherein one comparator is used for comparing the input data at the current moment with 0 in size, the other comparator is used for comparing the input data at the previous moment with 0 in size, if the comparison result of the comparator is greater than 0, the output is 1, if the comparison result of the comparator is less than 0, the output is 0, the results output by the two comparators are subjected to exclusive-OR, and the exclusive-OR result is multiplied by a preset normalization coefficient alpha 1 and accumulated to obtain the zero crossing point characteristic of the current sample signal;
The area calculation path sequentially comprises an adder, a multiplier and an accumulator, adds the input data at the current moment and the input data at the previous moment, multiplies the added result by a normalization coefficient alpha 2, and then performs accumulation to obtain the area characteristic of the current sample signal;
The attenuation calculation path sequentially comprises a comparator, a multiplier and an accumulator, compares the input data at the current moment with the input data at the previous moment in size, multiplies the comparison result by a normalization coefficient alpha 3 and then accumulates to obtain the attenuation characteristic of the current sample signal;
The line length calculation path sequentially comprises a subtracter, an absolute value calculation unit, a multiplier and an accumulator, wherein the input data at the current moment is subtracted from the input data at the previous moment, the absolute value of the subtraction result is calculated, and the subtraction result is multiplied by a normalization coefficient alpha 4 and then accumulated to obtain the line length characteristic of the current sample signal;
The similarity calculation unit reads all positive sample time-frequency domain feature vectors and all negative sample time-frequency domain feature vectors from the storage unit, and obtains a positive sample feature template and a negative sample feature template based on the average value of all positive and negative sample time-frequency domain feature vectors; the screening weight of the current channel of the input signal is represented and cached based on the cosine distance between the positive and negative sample feature templates, and after the screening weight of all channels of the input signal is obtained, the screening weight is sent to the sequencing unit; the similarity calculation unit comprises a plurality of similarity sub-calculation paths and is used for cosine distance calculation between each dimension feature of the positive and negative sample feature templates of the current channel, and the cosine distance calculation result of each similarity sub-calculation path is accumulated to obtain the screening weight of the current channel;
The sorting unit is used for sorting the screening weights of all the channels in ascending order, taking channel identifiers corresponding to the first K screening values as channel screening results, and sending the channel identifiers to the signal detection unit;
The signal detection unit reads signal data of a channel corresponding to a signal to be detected based on a current channel screening result, performs a two-class detection task, feeds back the detection result to the feature extraction unit, and the switching unit of the feature extraction unit controls whether signals of all channels of the current signal to be detected are sequentially input to all paths of calculation channels as input signals based on the detection result: if the category of the detection result is matched with the category requirement to be input into the calculation path, the switch unit is turned on, otherwise, the switch unit is turned off; and the multiplexer of the feature extraction unit determines the storage unit of the time-frequency domain feature vector of the input signal based on the current detection result.
2. The adaptive input signal path screening circuit for introducing feedback of detection results according to claim 1, wherein the path screening result includes a number of paths K matching the number of input paths of the input layer of the neural network when the signal detection unit performs the classification detection task based on the neural network.
3. The adaptive input signal path screening circuit for introducing test result feedback of claim 1, wherein each similarity sub-computation path comprises a subtractor and a multiplier in sequence; each dimension characteristic of the positive and negative sample characteristic templates is subtracted by a subtracter, and then square is obtained by a multiplier.
4. The adaptive input signal path screening circuit incorporating detection result feedback of claim 1, wherein the ranking unit ranks the screening weights of the paths in a bubbling ranking manner.
5. The adaptive input signal channel screening circuit for introducing feedback of detection results according to claim 4, wherein the sorting unit comprises two screening weight buffers, a comparator, three multiplexers and a demultiplexer, wherein a first screening weight buffer is used for storing cosine distances of all channels, a second screening weight buffer is used for storing channel screening results, a multi-selector of two multiple choices selects adjacent elements from the first screening weight buffer to enter the comparator for comparison, the comparison results respectively enter two selectors of one choice, each of the two selectors of one choice selects a largest element to enter the demultiplexer of two choice, and the output end of the demultiplexer of two choice is connected to the second screening weight buffer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211028703.6A CN115299946B (en) | 2022-08-25 | 2022-08-25 | Self-adaptive input signal channel screening circuit introducing detection result feedback |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211028703.6A CN115299946B (en) | 2022-08-25 | 2022-08-25 | Self-adaptive input signal channel screening circuit introducing detection result feedback |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115299946A CN115299946A (en) | 2022-11-08 |
| CN115299946B true CN115299946B (en) | 2024-05-28 |
Family
ID=83864156
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211028703.6A Active CN115299946B (en) | 2022-08-25 | 2022-08-25 | Self-adaptive input signal channel screening circuit introducing detection result feedback |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115299946B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104080397A (en) * | 2012-01-19 | 2014-10-01 | 脑科技医疗系统股份有限公司 | Diagnostic system for detection of fluid changes |
| CN109620148A (en) * | 2018-11-29 | 2019-04-16 | 西安交通大学 | A kind of epilepsy detection integrated circuit based on sparse extreme learning machine algorithm |
| CN110811609A (en) * | 2019-10-28 | 2020-02-21 | 杭州电子科技大学 | Intelligent epileptic spike detection method based on fusion of adaptive template matching and machine learning algorithm |
| CN111034041A (en) * | 2017-05-24 | 2020-04-17 | 安乐泰克有限公司 | Apparatus and method for controlling resonator |
| CN112949533A (en) * | 2021-03-15 | 2021-06-11 | 成都信息工程大学 | Motor imagery electroencephalogram identification method based on relative wavelet packet entropy brain network and improved version lasso |
| CN113936791A (en) * | 2021-10-18 | 2022-01-14 | 贵州师范大学 | Classification model training method combining active learning and transfer learning |
| CN114532993A (en) * | 2022-03-23 | 2022-05-27 | 电子科技大学 | Automatic detection method for electroencephalogram high-frequency oscillation signals of epileptic |
| CN114587381A (en) * | 2022-03-14 | 2022-06-07 | 杭州电子科技大学 | Spike detection method based on multi-channel electroencephalogram intelligent screening and weighted sample generation |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7630757B2 (en) * | 1997-01-06 | 2009-12-08 | Flint Hills Scientific Llc | System for the prediction, rapid detection, warning, prevention, or control of changes in activity states in the brain of a subject |
| IL155955A0 (en) * | 2003-05-15 | 2003-12-23 | Widemed Ltd | Adaptive prediction of changes of physiological/pathological states using processing of biomedical signal |
| GB0906029D0 (en) * | 2009-04-07 | 2009-05-20 | Nat Univ Ireland Cork | A method of analysing an electroencephalogram (EEG) signal |
| US9305559B2 (en) * | 2012-10-15 | 2016-04-05 | Digimarc Corporation | Audio watermark encoding with reversing polarity and pairwise embedding |
| US10130813B2 (en) * | 2015-02-10 | 2018-11-20 | Neuropace, Inc. | Seizure onset classification and stimulation parameter selection |
| US11682492B2 (en) * | 2019-06-07 | 2023-06-20 | Cornell University | Energy-efficient on-chip classifier for detecting physiological conditions |
-
2022
- 2022-08-25 CN CN202211028703.6A patent/CN115299946B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104080397A (en) * | 2012-01-19 | 2014-10-01 | 脑科技医疗系统股份有限公司 | Diagnostic system for detection of fluid changes |
| CN111034041A (en) * | 2017-05-24 | 2020-04-17 | 安乐泰克有限公司 | Apparatus and method for controlling resonator |
| CN109620148A (en) * | 2018-11-29 | 2019-04-16 | 西安交通大学 | A kind of epilepsy detection integrated circuit based on sparse extreme learning machine algorithm |
| CN110811609A (en) * | 2019-10-28 | 2020-02-21 | 杭州电子科技大学 | Intelligent epileptic spike detection method based on fusion of adaptive template matching and machine learning algorithm |
| CN112949533A (en) * | 2021-03-15 | 2021-06-11 | 成都信息工程大学 | Motor imagery electroencephalogram identification method based on relative wavelet packet entropy brain network and improved version lasso |
| CN113936791A (en) * | 2021-10-18 | 2022-01-14 | 贵州师范大学 | Classification model training method combining active learning and transfer learning |
| CN114587381A (en) * | 2022-03-14 | 2022-06-07 | 杭州电子科技大学 | Spike detection method based on multi-channel electroencephalogram intelligent screening and weighted sample generation |
| CN114532993A (en) * | 2022-03-23 | 2022-05-27 | 电子科技大学 | Automatic detection method for electroencephalogram high-frequency oscillation signals of epileptic |
Non-Patent Citations (4)
| Title |
|---|
| Classification of Epileptic Seizures using Artificial Neural Network with Adaptive Momentum;Hameed, A;《2020 International Conference on Data Analytics for Business and Industry》;20201231;第1-5页 * |
| 基于多特征融合和集成学习的癫痫脑电自动识别方法研究;黄璐弘;《中国优秀硕士学位论文全文数据库》;20201231;第1-58页 * |
| 基于神经网络的低功耗智能癫痫检测硬件设计;周勇;《中国优秀硕士学位论文全文数据库》;20220321;第1-50页 * |
| 自适应最近邻算法及在门禁人脸识别中的应用;王明达;《中国优秀硕士学位论文全文数据库》;20181231;第1-62页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115299946A (en) | 2022-11-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107822622B (en) | Electrocardiogram diagnosis method and system based on deep convolutional neural network | |
| CN114595874B (en) | Ultra-short-term power load prediction method based on dynamic neural network | |
| CN112232164B (en) | Video classification method and device | |
| Bing et al. | Electrocardiogram classification using TSST-based spectrogram and ConViT | |
| CN110796120A (en) | Time domain feature-based circuit breaker mechanical fault XGboost diagnosis method | |
| CN104616002A (en) | Facial recognition equipment used for judging age groups | |
| CN111785385A (en) | Disease classification method, device, equipment and storage medium | |
| CN114818837B (en) | Electroencephalogram signal intelligent processing circuit based on multistage neural network and block calculation | |
| CN115299946B (en) | Self-adaptive input signal channel screening circuit introducing detection result feedback | |
| CN117463506A (en) | Self-adaptive constant-current constant-voltage control high-voltage power supply | |
| CN109934248B (en) | Multi-model random generation and dynamic self-adaptive combination method for transfer learning | |
| CN116089820A (en) | Load identification method and system based on user cooperation, electronic equipment and medium | |
| CN114224354B (en) | Arrhythmia classification method, arrhythmia classification device, and readable storage medium | |
| CN113066544B (en) | FVEP characteristic point detection method based on CAA-Net and LightGBM | |
| KR102290243B1 (en) | Classification method for Electroencephalography motor imagery | |
| CN114611556A (en) | Multi-class motor imagery task identification method based on graph neural network | |
| CN110448273B (en) | Low-power-consumption epilepsy prediction circuit based on support vector machine | |
| CN112036551A (en) | Acceleration method and device of convolutional neural network | |
| CN115168016B (en) | Task scheduling method and related device, chip, device and medium | |
| CN113688802B (en) | Gesture recognition method, device and equipment based on electromyographic signals and storage medium | |
| CN112863520B (en) | Binary weight convolutional neural network module and method for identifying voiceprint by using same | |
| Liu et al. | A selective quantization approach for optimizing quantized inference engine | |
| CN114782340B (en) | Aortic dissection detection method, aortic dissection detection system, electronic equipment and storage medium | |
| CN117037287B (en) | Behavior recognition method, system and device based on 3D impulse neural network | |
| Nimitha et al. | Diagnosis and treatment recommendation engine for skin diseases using AI model VGG19 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |