CN110859599A - Cerebrovascular disease nerve injury patient cognitive function automatic screening system - Google Patents
Cerebrovascular disease nerve injury patient cognitive function automatic screening system Download PDFInfo
- Publication number
- CN110859599A CN110859599A CN201911176388.XA CN201911176388A CN110859599A CN 110859599 A CN110859599 A CN 110859599A CN 201911176388 A CN201911176388 A CN 201911176388A CN 110859599 A CN110859599 A CN 110859599A
- Authority
- CN
- China
- Prior art keywords
- central point
- coordinates
- point coordinate
- clock
- coordinate
- 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.)
- Granted
Links
Images
Classifications
-
- 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
- A61B5/4088—Diagnosing of monitoring cognitive diseases, e.g. Alzheimer, prion diseases or dementia
-
- 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
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Neurology (AREA)
- Public Health (AREA)
- Surgery (AREA)
- Veterinary Medicine (AREA)
- Physics & Mathematics (AREA)
- Physiology (AREA)
- Biophysics (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Psychiatry (AREA)
- Animal Behavior & Ethology (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Child & Adolescent Psychology (AREA)
- Developmental Disabilities (AREA)
- Hospice & Palliative Care (AREA)
- Artificial Intelligence (AREA)
- Psychology (AREA)
- Neurosurgery (AREA)
- Image Analysis (AREA)
Abstract
The invention belongs to the technical field of medical automatic screening, and particularly relates to an automatic screening system for cognitive functions of patients with cerebrovascular disease and nerve injury. The system of the invention comprises: the system comprises an image acquisition device and an intelligent analysis system; the intelligent analysis system comprises a first target detection network, a cutting module, a second target detection network, a grading module and a discrimination module. The system inputs the basic information of the subject and the clock image information for testing; the clock image information is a circular clock drawn by the subject, and is provided with an hour hand and a minute hand which are marked with 1-12 hours of Araba numbers; automatically analyzing by an intelligent analysis system, and finally outputting a screening result report for cognition of the testee; the invention can effectively reduce the labor input of professional medical personnel and keep the consistency of the screening result.
Description
Technical Field
The invention belongs to the technical field of medical automatic screening, and particularly relates to an automatic screening system for cognitive functions of patients with cerebrovascular disease and nerve injury.
Background
With the development of medical technology, the aging of social structures suggests that the diagnosis and treatment requirements of senile diseases need to be emphasized urgently. Traditional senile brain diseases such as stroke, epilepsy and the like also tend to be younger and younger. Early detection and screening of patients with cerebrovascular disease nerve injury for timely targeted intervention are key strategies for improving the prognosis capability of patients with cerebrovascular disease nerve injury and reducing the social burden. At present, the nerve injury of many cerebrovascular diseases is more hidden, pathological changes occur in the brain, but when the pathological changes are not serious, obvious clinical symptoms and effective detection means are lacked. For patients with cerebrovascular disease nerve injury, due to the existence of potential cranial nerve injury, the patients show paroxysmal behavioral abnormality and mental state abnormality, and are easily overlooked by family members or nursing personnel, if the patients do not discover and intervene in time, the brain function is further damaged, and even threaten the life.
At present, in order to improve the early screening capability of the cognitive function of the patients and follow up the change of the disease, various cognitive scales are usually adopted for behavior or cognitive function assessment in clinic. The commonly used screening scales include the simple mental state assessment scale (MMSE), the montreal cognitive assessment scale (McCA), the activities of daily living scale (ADL), and the like. However, the above scales tend to be only one-sided and require trained physicians or professionals to complete. Since such scales need to be queried and observed by others, the evaluations of different persons are different, and it is difficult to meet the requirement of evaluation consistency.
In order to solve the consistency problem and reduce the burden of medical staff and patients, the invention provides an automatic system for screening the cognitive function of the cerebrovascular disease nerve injury patients based on a computer algorithm. The system adopts a non-contact mode, only needs the patient to draw a clock, and does not bring great burden to the patient. The system is simple and clear in operation, does not need professional training of medical personnel to operate, and reduces the burden of the medical personnel. Because the system adopts a computer algorithm to carry out information analysis, the system is not influenced by other objective factors and has higher screening consistency.
Disclosure of Invention
The invention aims to provide an automatic screening system for cognitive functions of patients with cerebrovascular disease and nerve injury, which has good screening consistency and is simple and convenient to operate.
According to the automatic screening system for cognitive functions of the patient with the cerebrovascular disease and the nerve injury, provided by the invention, the cognitive function condition of the patient with the cerebrovascular disease and the nerve injury can be known only by drawing a clock image for testing by the patient, establishing a certain algorithm and carrying out information analysis by a computer.
The automatic screening system provided by the invention has the following input of two parts of information: basic information of the subject, and clock image information for the test. The basic information comprises information such as age, sex, school calendar, past medical history and the like, and is used for assisting automatic screening analysis; the picture information of drawing clock is a round clock drawn on paper by the subject, and the clock has hour hand and minute hand, and is marked with 1-12 hours of Araba number, and the hand points to 3 hours and 40 minutes; the basic information of the testee is used as assistance, combined with the information of the picture clock image, input into the system, automatically analyzed by the intelligent system, and finally output a screening result report for the cognition of the testee.
The system comprises: the system comprises an image acquisition device and an intelligent analysis system; the intelligent analysis system comprises an image preprocessing module, a first target detection network 1, a cutting module, a second target detection network 2, a grading module and a discrimination module; wherein:
the image acquisition device is used for shooting clock image information drawn by the subject; inputting the image into an intelligent analysis system;
the image acquisition device may be a cell phone camera, a computer camera, an ipad, or other type of camera-ready device.
The image preprocessing module is used for preprocessing the original image acquired by the image acquisition device, and roughly cutting off background patterns irrelevant to the surface; and carrying out local light supplement and contrast adjustment on the picture.
The first object detection network 1 is used for detecting the coordinates (x1, y1) of the upper left corner and the coordinates (x2, y2) of the lower right corner of the painted clock contour, so as to place the painted clock contour in a box enclosed by the coordinates (x1, y1) of the upper left corner and the coordinates (x2, y2) of the lower right corner, and the figure 3 is referred to.
And the cutting module is used for cutting the clock image obtained by the first target detection network. Considering that the source of the original clock image may be complex and the background may be cluttered, the clock contour outlined by the first object detection network 1 may be too small or too large compared to the actual one. The too large outline of the drawing clock has little influence on the next identification; but too small or even not fully framing the number can have a large impact on the subsequent recognition. Therefore, the coordinate values of the clipping are given according to the following coordinate transformation formula, and the coordinate system is determined according to the coordinate system regulation method commonly used in the digital image field: the upper left corner of the picture is defined as the origin (0,0), the abscissa of the picture is the x-axis, and the ordinate is the y-axis, as shown in fig. 4. The coordinate transformation formula is as follows:
α and β are adjustment coefficients, and the transformed coordinates of the upper left corner (x1 ', y 1') and the coordinates of the lower right corner (x2 ', y 2') are used as the basis for clipping the picture.
And cutting the picture according to the coordinates given in the step.
A second object detection network 2 for detecting the positions of the hour numbers 1-12 and the hour and minute hands in the picture of the drawing clock. The input of the image is the image processed by the last step of cutting module, and the output is the label, the corresponding clock point number and the pointer coordinate; the detection process comprises the following steps: if the numbers or the pointers are repeatedly identified, respectively identifying and outputting respective coordinates; if the number or the pointer does not appear, the coordinates of the corresponding output upper left corner and lower right corner are both (0, 0); thus the result is a minimum of 14: the numbers 1-12 correspond to the labels 1-12, and the pointer located in the left half area of the clock corresponds to the label 13, which is considered as a minute hand; the hands located in the area of the right half of the timepiece correspond to the tags 14, considered as hour hands.
A scoring module for scoring the cognitive function of the subject; the input of the system is a picture clock image processed by a cutting module, and a label and a corresponding coordinate output by the target detection network 2, and a score is obtained according to a clock september rule, and finally, the score values of 0-7 are obtained.
And (4) carrying out coordinate preprocessing, and calculating the coordinates cx and cy of the central point according to the coordinates of the upper left corner and the coordinates of the lower right corner of all the labels according to the following central point calculation formula (2). Each label will correspond to three coordinates, the top left (x1, y1), bottom right (x2, y2), and center point (cx, cy).
cx＝(x1+x2)/2
cy＝(y1+y2)/2 (2)。
The scoring rules are as follows:
(one) require that all numbers from 1-12 be present, otherwise the term is not scored. Checking whether the coordinates of the upper left corner and the lower right corner corresponding to the labels of the numbers 1-12 appear (0,0), if so, indicating that the numbers corresponding to the coordinates are not detected in the current image, and the score of the item is 0. If (0,0) is not present in any of the 12 sets of coordinates, indicating that the numbers 1-12 were all detected in the picture, the term scores 1.
And (II) respectively calculating the coordinates cx and cy of the center point of each digit according to the coordinates of the upper left corner and the lower right corner of each digit. The following 12 judgments are sequentially made according to the arrangement sequence of the clock figures:
(1) the central point coordinates cx and cy of the number 1 are both larger than the central point coordinates cx and cy of the number 12;
(2) the central point coordinates cx and cy of the number 2 are both larger than the central point coordinates cx and cy of the number 1;
(3) the central point coordinates cx and cy of the number 3 are both larger than the central point coordinates cx and cy of the number 2;
(4) the central point coordinate cx of the number 4 is smaller than the central point coordinate cx of the number 3, and the central point coordinate cy of the number 4 is larger than the central point coordinate of the number 3;
(5) the central point coordinate cx of the number 5 is smaller than the central point coordinate cx of the number 4, and the central point coordinate cy of the number 5 is larger than the central point coordinate of the number 4;
(6) the central point coordinate cx of the number 6 is smaller than the central point coordinate cx of the number 5, and the central point coordinate cy of the number 6 is larger than the central point coordinate of the number 5;
(7) the central point coordinate cx of the number 7 is smaller than the central point coordinate cx of the number 6, and the central point coordinate cy of the number 7 is smaller than the central point coordinate of the number 6;
(8) the central point coordinate cx of the number 8 is smaller than the central point coordinate cx of the number 7, and the central point coordinate cy of the number 8 is smaller than the central point coordinate of the number 7;
(9) the central point coordinate cx of the number 9 is smaller than the central point coordinate cx of the number 8, and the central point coordinate cy of the number 9 is smaller than the central point coordinate of the number 8;
(10) the central point coordinate cx of the number 10 is larger than the central point coordinate cx of the number 9, and the central point coordinate cy of the number 10 is smaller than the central point coordinate of the number 9;
(11) the central point coordinate cx of the number 11 is larger than the central point coordinate cx of the number 10, and the central point coordinate cy of the number 11 is smaller than the central point coordinate of the number 10;
(12) the central point coordinate cx of the number 12 is larger than the central point coordinate cx of the number 11, and the central point coordinate cy of the number 12 is smaller than the central point coordinate of the number 11;
if all of the 12 judgments are satisfied, the score is 1, otherwise, the score is 0.
And (III) the coordinate system adopted before is a rectangular coordinate system, and all the digital distribution is converted into a polar coordinate system when judged. The center of the drawing clock is taken as the origin of the polar coordinate system, the angles of the numbers 1 and 2 are 0-90 degrees, the angles of the numbers 4 and 5 are 270-360 degrees, the angles of the numbers 7 and 8 are 270-180 degrees, and the angles of the numbers 10 and 11 are 90-180 degrees.
A straight line a is determined from the coordinates of the center points of the numerals 3 and 9, a straight line B is determined from the coordinates of the center points of the numerals 6 and 12, and then the coordinates of the center points of the two straight lines are calculated A, B and are denoted as (clock _ x, clock _ y). The midpoint is taken as the midpoint of the entire picture clock image. First, coordinate transformation is performed, and the coordinates of the former rectangular coordinate system use the upper left corner of the picture as the origin, and now use (clock _ x, clock _ y) as the origin instead. The transformation formula is as follows:
x′＝x-clock_x (3)
y′＝y-clock_y
and then, converting the transformed rectangular coordinates into a polar coordinate format, wherein the transformation formula is as follows:
θ＝arctan(y/x) (4)
finally, since the picture taking device is taking a picture of the drawing clock, it may cause the drawing clock to rotate due to the problem of the shooting angle, i.e. the number 12 is not located right above. To correct for this rotation, the polar angle of the number 1-12 is subtracted by the polar angle of the number 3.
Subsequently, a determination is made of the polar angle (hereinafter referred to as angle) of 8 digits:
(1)0< number 1 angle < 90;
(2)0< number 2 angle < 90;
(3)90< number 11 angle < 180;
(4)90< number 10 angle < 180;
(5)180< number 8 angle < 270;
(6)180< number 7 angle < 270;
(7)270< number 5 angle < 360;
(8)270< number 4 angle < 360;
if all the six judgments are satisfied, the score is 1, otherwise, the score is 0.
(IV) this item requires two pointers to appear, otherwise the item is not scored. And respectively checking whether the coordinates of the upper left corner and the lower right corner corresponding to the hour hand label and the minute hand label appear (0,0), if so, indicating that the pointer corresponding to the coordinates is not detected in the current image, and the score of the item is 0. If (0,0) is not present in any of the 2 sets of coordinates, indicating that both pointers were detected in the picture, the term scores 1.
And (V) the pointer label determined by the part is 14, the coordinates of the pointer 14 and the upper left corner and the lower right corner of the number 4 are checked firstly, if (0,0) is found, the pointer 14 or the number 4 is not detected, the score of the item is zero, otherwise, the determination is continued.
The coordinates of the upper left corner and the lower right corner of the pointer 14 form straight lines, the distances d3 and d4 from the center points of the numbers 3 and 4 to the straight lines and the distance d34 between the center points of the numbers 3 and 4 are respectively calculated, and the judgment is carried out according to the following formula:
((d3-d4))/d34≥δ
where δ is defined as the threshold. If this condition is met, the term score is 1, otherwise it is 0.
And (VI) the pointer label determined by the part is 13, the coordinates of the pointer 13 and the upper left corner and the lower right corner of the number 8 are checked firstly, if (0,0) is found, the pointer 13 or the number 4 is not detected, the score of the item is zero, otherwise, the determination is continued. The coordinates of the upper left corner and the lower right corner of the pointer 13 are (x1, y1), (x2, y2), respectively, the coordinates of the upper right corner are (x2, y1), and the coordinates of the lower left corner are (x1, y 2). The coordinates of the upper right corner and the lower left corner form a straight line D, and the distances D7, D8, D9 from the numbers 7,8,9 to the straight line D are calculated respectively. If d8 is minimal, the term scores 1, otherwise it is 0.
(seventh) the pointer labels judged by the part are 13 and 14, the coordinates of the upper left corner and the lower right corner of the pointers 13 and 14 are checked firstly, if (0,0) is found, the pointer 13 or 14 is not detected, the score of the item is zero, otherwise, the judgment is continued; the lengths d13, d14 of the pointer 13 and pointer 14 are calculated, respectively. Then, the judgment is carried out according to the following formula:
(d13-d14)/d13≥ω
where ω is a threshold. If the above determination is satisfied, the term is scored as 1, otherwise it is 0.
The judging module adds the 7 scores obtained by the scoring module, and according to the academic record information of the subject, the academic record is in primary school and below, the cognitive normal is in scores of 5 and above, and the cognitive abnormal is in the rest; the study history is in the middle school and above, with scores of 6 and above being normal cognition and the rest being abnormal cognition.
In the scoring process, the values of the adjusting coefficients α and β are α -0.7-1.0 and β -0.9-1.2, generally, α takes an empirical value of 0.8, and β takes an empirical value of 1.2.
The threshold value delta is-0.1-0.3; the threshold value omega is 0.10-0.25. Usually, δ is taken to be-0.2 and ω is taken to be 0.15.
The automatic analysis of the system combines the knowledge in the field of medical experts and the algorithm of machine learning, and trains the computer algorithm to learn and judge the normal abnormity of behaviors and cognition by extracting the main index characteristics of the medical experts in the screening process. The screening by using the computer algorithm can ensure a certain accuracy by using expert knowledge, and can achieve the screening consistency of region-crossing, time-crossing and personnel-crossing.
Compared with the prior art, the invention has the following beneficial effects:
the invention provides an automatic screening system for cognitive functions of patients with cerebrovascular disease nerve injury, which is established on the basis of expert knowledge in the medical field and performs automatic feature extraction and discrimination through a machine learning algorithm.
Drawings
FIG. 1 is a simplified schematic of the information collection and automated analysis of the automated method of the present invention.
FIG. 2 is a schematic diagram of an original picture of a clock captured by the present invention.
Fig. 3 is a clock outline identified by the first object detecting network 1 of the present invention.
Fig. 4 is a conventional rectangular coordinate system illustration of a digital image field picture.
FIG. 5 is an enlarged, cut-out body of the present invention based on the outline of the painted clock identified in FIG. 3.
Fig. 6 shows the numbers, pointers and their corresponding locations on the body of the drawing clock identified by the second object detection network 2 according to the invention.
Fig. 7-10 are test clock images drawn by the subject.
Detailed Description
The present invention will be described more fully hereinafter in the reference to the accompanying drawings, which provide preferred embodiments of the invention, and which are not to be considered as limited to the embodiments set forth herein.
Firstly, an image acquisition device is used for acquiring and storing a drawn picture of a picture clock, and the acquisition device can be a mobile phone camera/a computer camera/an ipad or other types of photographing devices.
The picture of the drawing clock test is shown in fig. 2, the picture of the drawing clock is input into the first target detecting network 1, the network outputs the detected coordinates of the upper left corner (x1, y1) and the lower right corner (x2, y2) of the outline of the drawing clock, and the outline of the drawing clock framed by the target detecting network 1 may be too small or too large compared with the actual one considering that the source of the original picture may be complex and the background may be cluttered. The too large outline of the drawing clock has little influence on the next identification; but too small or even not fully framing the number can have a large impact on the subsequent recognition. Therefore, the coordinate values according to the cutting step are given according to the following coordinate transformation formula, and the coordinate system of the coordinate transformation formula is defined according to the coordinate system commonly used in the field of digital images: the upper left corner of the picture is defined as the origin (0,0), the abscissa of the picture is the x-axis, and the ordinate is the y-axis, as shown in fig. 4.
x1′＝α*x1
y1′＝α*y1
x2′＝β*x2
y2′＝β*y2
α takes an empirical value of 0.8, β takes an empirical value of 1.2, and the transformed coordinates of the upper left corner (x1 ', y 1') and the lower right corner (x2 ', y 2') are used as the basis for clipping the picture.
The picture is cropped according to the coordinates given in the above step, as shown in fig. 5.
The second object detection network 2 is responsible for detecting the numbers 1-12 and the positions of the hour and minute hands. The network input is the picture 2 cut in the last step, and the output is the label and the corresponding clock point number and the pointer coordinate. The detection process comprises the following steps: if the numbers or the pointers are repeatedly identified, respectively identifying and outputting respective coordinates; if the number or pointer is not present, the corresponding output upper left and lower right corner coordinates are both (0, 0). Therefore, the result has at least 14 items. Wherein, the numbers 1-12 correspond to the labels 1-12, and the pointer positioned in the left half part area of the clock corresponds to the label 13 and is regarded as a minute hand; the hands located in the area of the right half of the timepiece correspond to the tags 14, considered as hour hands.
And the scoring module receives the labels and the corresponding coordinates output by the picture 5 and the target detection network 2 and outputs scores of 0-7.
And (4) carrying out coordinate preprocessing, and calculating the coordinates cx and cy of the central point according to the coordinates of the upper left corner and the coordinates of the lower right corner of all the labels according to the following central point calculation formula. Each label will correspond to three coordinates, the top left (x1, y1), bottom right (x2, y2), and center point (cx, cy).
cx＝(x1+x2)/2
cy＝(y1+y2)/2
This term requires that all numbers from 1-12 appear, otherwise the term does not score. Checking whether the coordinates of the upper left corner and the lower right corner corresponding to the labels of the numbers 1-12 appear (0,0), if so, indicating that the numbers corresponding to the coordinates are not detected in the current image, and the score of the item is 0. If (0,0) is not present in any of the 12 sets of coordinates, indicating that the numbers 1-12 were all detected in the picture, the term scores 1.
And respectively calculating the coordinates cx and cy of the center point of each digit according to the coordinates of the upper left corner and the lower right corner of each digit. The following 12 judgments are sequentially made according to the arrangement sequence of the clock figures:
1. the central point coordinates cx and cy of the number 1 are both larger than the central point coordinates cx and cy of the number 12;
2. the central point coordinates cx and cy of the number 2 are both larger than the central point coordinates cx and cy of the number 1;
3. the central point coordinates cx and cy of the number 3 are both larger than the central point coordinates cx and cy of the number 2;
4. the central point coordinate cx of the number 4 is smaller than the central point coordinate cx of the number 3, and the central point coordinate cy of the number 4 is larger than the central point coordinate of the number 3;
5. the central point coordinate cx of the number 5 is smaller than the central point coordinate cx of the number 4, and the central point coordinate cy of the number 5 is larger than the central point coordinate of the number 4;
6. the central point coordinate cx of the number 6 is smaller than the central point coordinate cx of the number 5, and the central point coordinate cy of the number 6 is larger than the central point coordinate of the number 5;
7. the central point coordinate cx of the number 7 is smaller than the central point coordinate cx of the number 6, and the central point coordinate cy of the number 7 is smaller than the central point coordinate of the number 6;
8. the central point coordinate cx of the number 8 is smaller than the central point coordinate cx of the number 7, and the central point coordinate cy of the number 8 is smaller than the central point coordinate of the number 7;
9. the central point coordinate cx of the number 9 is smaller than the central point coordinate cx of the number 8, and the central point coordinate cy of the number 9 is smaller than the central point coordinate of the number 8;
10. the central point coordinate cx of the number 10 is larger than the central point coordinate cx of the number 9, and the central point coordinate cy of the number 10 is smaller than the central point coordinate of the number 9;
11. the central point coordinate cx of the number 11 is larger than the central point coordinate cx of the number 10, and the central point coordinate cy of the number 11 is smaller than the central point coordinate of the number 10;
12. the central point coordinate cx of the number 12 is larger than the central point coordinate cx of the number 11, and the central point coordinate cy of the number 12 is smaller than the central point coordinate of the number 11;
if all of the 12 judgments are satisfied, the score is 1, otherwise, the score is 0.
The coordinate system adopted before is a rectangular coordinate system, and all the digital distribution is converted into a polar coordinate system when the digital distribution is judged. The center of the drawing clock is used as the origin of the polar coordinate system, the angle of the numeral 12 is 0-90 degrees, the angle of the numeral 45 is 270-360 degrees, the angle of the numeral 78 is 270-180 degrees, and the angle of the numeral 1011 is 90-180 degrees.
First, a straight line A is determined according to the coordinates of the central points of the numbers 3 and 9, a straight line B is determined according to the coordinates of the central points of the numbers 6 and 12, and then the coordinates of the middle points of the two straight lines AB are calculated and recorded as (clock _ x, clock _ y). The midpoint is taken as the midpoint of the entire picture clock image. First, coordinate transformation is performed, and the coordinates of the former rectangular coordinate system use the upper left corner of the picture as the origin, and now use (clock _ x, clock _ y) as the origin instead. The transformation formula is as follows:
x′＝x-clock_x
y′＝y-clock_y
and then, converting the transformed rectangular coordinates into a polar coordinate format, wherein the transformation formula is as follows:
θ＝arctan(y/x)
finally, since the picture taking device is taking a picture of the drawing clock, it may cause the drawing clock to rotate due to the problem of the shooting angle, i.e. the number 12 is not located right above. To correct for this rotation, the polar angle of the number 1-12 is subtracted by the polar angle of the number 3.
Subsequently, a determination is made of the polar angle (hereinafter referred to as angle) of 8 digits:
1.0< number 1 Angle <90
2.0< number 2 angle <90
3.90< number 11 angle <180
4.90< number 10 angle <180
5.180< number 8 angle <270
6.180< number 7 angle <270
7.270< number 5 Angle <360
8.270< number 4 angle <360
If all the six judgments are satisfied, the score is 1, otherwise, the score is 0.
This entry requires the presence of two pointers, otherwise the entry does not score. And respectively checking whether the coordinates of the upper left corner and the lower right corner corresponding to the hour hand label and the minute hand label appear (0,0), if so, indicating that the pointer corresponding to the coordinates is not detected in the current image, and the score of the item is 0. If (0,0) is not present in any of the 2 sets of coordinates, indicating that both pointers were detected in the picture, the term scores 1.
The pointer label determined in this section is 14, the coordinates of the pointer 14 and the upper left corner and the lower right corner of the numeral 4 are checked first, if (0,0) is found, it indicates that the pointer 14 or the numeral 4 is not detected, the score of this item is zero, otherwise, the determination is continued.
The coordinates of the upper left corner and the lower right corner of the pointer 14 form straight lines, the distances d3 and d4 from the center points of the numbers 3 and 4 to the straight lines and the distance d34 between the center points of the numbers 3 and 4 are respectively calculated, and the judgment is carried out according to the following formula:
((d3-d4))/d34≥δ
where δ is taken as an empirical value of-0.2. If this condition is met, the term score is 1, otherwise it is 0.
The pointer label determined in this section is 13, the coordinates of the pointer 13 and the upper left corner and the lower right corner of the numeral 8 are checked first, if (0,0) is found, it indicates that the pointer 13 or the numeral 4 is not detected, the score of this item is zero, otherwise, the determination is continued. The coordinates of the upper left corner and the lower right corner of the pointer 13 are (x1, y1), (x2, y2), respectively, the coordinates of the upper right corner are (x2, y1), and the coordinates of the lower left corner are (x1, y 2). The coordinates of the upper right corner and the lower left corner form a straight line D, and the distances D7, D8, D9 from the numbers 7,8,9 to the straight line D are calculated respectively. If d8 is minimal, the term scores 1, otherwise it is 0.
The pointer labels determined by the part are 13 and 14, the coordinates of the upper left corner and the lower right corner of the pointers 13 and 14 are checked firstly, if (0,0) is found, the pointer 13 or 14 is not detected, the score of the item is zero, otherwise, the determination is continued. The lengths of the pointers are given by the following formulas, and the lengths d13, d14 of the pointer 13 and the pointer 14 are calculated, respectively. Then, the judgment is carried out according to the following formula:
(d13-d14)/d13≥ω
where ω is the empirical value of 0.15. If the above determination is satisfied, the term is scored as 1, otherwise it is 0.
Adding the scores, wherein the scholastic calendar of the subject is in the elementary school and below according to the scholastic calendar information of the subject, the cognitive normal is divided by 5 or above, and the cognitive abnormal is divided by the rest; the study history is in the middle school and above, with scores of 6 and above being normal cognition and the rest being abnormal cognition.
The scoring result format given by the doctor and the machine is "number complete + number order + number distribution + two pointers + one pointing to 4+ one pointing to eight + one long and one short", so if all seven items are correct, then a score of "1, 1".
Test example 1, a clock image drawn by a subject is shown in fig. 7.
The doctor is classified as 1, 0, according to the method of the invention, the machine is classified as 1, 0, 1, and the pointer pointing to 8 is longer than the pointer pointing to 4, so that the doctor has a mistake in the classification and the machine makes a correct judgment.
Test example 2 a clock image drawn by the subject is shown in fig. 8.
The doctor is classified as 1, 0, 1, according to the method of the invention, the machine is classified as 1, 0, and the length of the two pointers is approximately consistent, so that the doctor has a mistake in the classification, and the machine makes a correct judgment.
Test example 3, a clock image drawn by the subject is shown in fig. 9.
The doctor is classified as "1, 0, 1", according to the method of the present invention, the machine is classified as "1, 0, 1", it can be seen from the figure that the left hand pointer points to 3, and the right hand pointer points to 8, so that the correct classification is "1, 0, 1", the doctor has a mistake in the classification, and the machine makes a correct judgment.
Test example 4 a clock image drawn by the subject is shown in fig. 10.
The doctor is classified as "1, 0, 1", according to the method of the present invention, the machine is classified as "1, 0, 1, 0", it can be seen from the figure that the left hand pointer points to 3, the right hand pointer points to 8, and the lengths of the pointers are approximately the same, so that the correct classification is "1, 0, 1, 0", the doctor has a mistake in the classification, and the machine makes a correct judgment.
In the scoring standard, the length of each pointer can be divided into one long pointer and one short pointer, but in actual operation, the length difference between the two pointers can be considered as one long pointer and one short pointer, and different doctors give different answers. In the invention, omega is used as a measuring standard of the length difference of the pointer and is subjected to normalization processing, so that the scoring of the item has consistency.
Similarly, the pointer pointing to 4 can be one minute, but in practice, since the subject is required to draw three forty or four points, the pointer should be in the middle of the numbers 3 and 4 and biased to 4, but in practice, the degree of bias to 4 is not well defined during scoring. In the present invention, δ is used to measure the degree of pointer deviation 4, so the score of the item is consistent.
Claims (5)
1. An automatic cognitive function screening system for patients with cerebrovascular disease and nerve injury is characterized in that the system inputs two parts of information: basic information of a subject, clock image information for testing; the basic information comprises age, gender, school calendar and past medical history and is used for assisting automatic screening analysis; the clock image information is a circular clock drawn on paper by a subject, and the clock has an hour hand and a minute hand, 1-12 hours of Araba numbers are marked on the clock, and the hands point to 3 hours and 40 minutes; after the information is input into the system, the information is automatically analyzed by an intelligent system, and finally a screening result report for cognition of the testee is output;
the system comprises: the system comprises an image acquisition device and an intelligent analysis system; the intelligent analysis system comprises an image preprocessing module, a first target detection network, a cutting module, a second target detection network, a grading module and a judging module; wherein:
the image acquisition device is used for acquiring clock image information drawn by the subject; inputting the image into an intelligent analysis system;
the image preprocessing module is used for preprocessing the original image acquired by the image acquisition device, and roughly cutting off background patterns irrelevant to the surface; performing local light supplement and contrast adjustment on the picture;
the first target detection network is used for detecting the coordinates (x1, y1) of the upper left corner and the coordinates (x2, y2) of the lower right corner of the clock contour, so that the clock contour is placed in a box formed by horizontal lines and vertical lines with the coordinates (x1, y1) of the upper left corner and the coordinates (x2, y2) of the lower right corner as starting points;
the cutting module is used for cutting the clock image obtained by the first target detection network, the cutting is actually coordinate transformation, the upper left corner of the clock image is defined as an origin (0,0), the abscissa is an x-axis, the ordinate is a y-axis, and the coordinate transformation formula is as follows:
α is an adjusting coefficient, and the transformed coordinates of the upper left corner (x1 ', y 1') and the lower right corner (x2 ', y 2') are used as the basis for clipping the picture;
the second target detection network is used for detecting the positions of the clock numbers 1-12 and the hour and minute hands in the picture clock image; the input of the image is the image processed by the last step of cutting module, and the output is the label, the corresponding clock point number and the pointer coordinate; the detection process comprises the following steps: if the numbers or the pointers are repeatedly identified, respectively identifying and outputting respective coordinates; if the number or the pointer does not appear, the coordinates of the corresponding output upper left corner and lower right corner are both (0, 0); the results are a minimum of 14: the numbers 1-12 correspond to the labels 1-12, and the pointer located in the left half area of the clock corresponds to the label 13, which is considered as a minute hand; the hands located in the right half of the timepiece correspond to the tags 14, considered as hour hands;
the scoring module is used for scoring the cognitive function of the subject; the input of the system is a picture clock image processed by a cutting module, a label and a corresponding coordinate output by a target detection network 2, a score is obtained according to a clock seventh method rule, and finally, unequal score values of 0-7 are obtained;
the judging module adds the 7 scores obtained by the scoring module, and according to the academic record information of the subject, the academic record is in primary school and below, the cognitive normal is in scores of 5 and above, and the cognitive abnormal is in the rest; the study history is in the middle school and above, with scores of 6 and above being normal cognition and the rest being abnormal cognition.
2. The automated cerebrovascular disease neurological impairment patient cognitive function screening system according to claim 1, wherein the scoring module scores the cognitive function of the subject as follows;
coordinate preprocessing is carried out, and according to the following central point calculation formula (2), central point coordinates cx and cy are calculated according to the upper left corner coordinates and the lower right corner coordinates of all the labels; each tag will correspond to three coordinates: respectively, the upper left corner coordinates (x1, y1), the lower right corner coordinates (x2, y2), and the center point coordinates (cx, cy);
cx＝(x1+x2)/2 (2)
cy＝(y1+y2)/2
the scoring rules are as follows:
checking whether coordinates of the upper left corner and the lower right corner corresponding to the labels of the numbers 1-12 appear (0,0) respectively, if so, indicating that the numbers corresponding to the coordinates are not detected in the current image, and scoring to be 0; if (0,0) does not appear in any of the 12 sets of coordinates, it indicates that the numbers 1-12 are all detected in the picture, and the score is 1;
and (II) respectively calculating the coordinates cx and cy of the center point of each figure according to the coordinates of the upper left corner and the lower right corner of each figure, and sequentially judging for 12 times according to the arrangement sequence of the numbers of the clock:
(1) the central point coordinates cx and cy of the number 1 are both larger than the central point coordinates cx and cy of the number 12;
(2) the central point coordinates cx and cy of the number 2 are both larger than the central point coordinates cx and cy of the number 1;
(3) the central point coordinates cx and cy of the number 3 are both larger than the central point coordinates cx and cy of the number 2;
(4) the central point coordinate cx of the number 4 is smaller than the central point coordinate cx of the number 3, and the central point coordinate cy of the number 4 is larger than the central point coordinate of the number 3;
(5) the central point coordinate cx of the number 5 is smaller than the central point coordinate cx of the number 4, and the central point coordinate cy of the number 5 is larger than the central point coordinate of the number 4;
(6) the central point coordinate cx of the number 6 is smaller than the central point coordinate cx of the number 5, and the central point coordinate cy of the number 6 is larger than the central point coordinate of the number 5;
(7) the central point coordinate cx of the number 7 is smaller than the central point coordinate cx of the number 6, and the central point coordinate cy of the number 7 is smaller than the central point coordinate of the number 6;
(8) the central point coordinate cx of the number 8 is smaller than the central point coordinate cx of the number 7, and the central point coordinate cy of the number 8 is smaller than the central point coordinate of the number 7;
(9) the central point coordinate cx of the number 9 is smaller than the central point coordinate cx of the number 8, and the central point coordinate cy of the number 9 is smaller than the central point coordinate of the number 8;
(10) the central point coordinate cx of the number 10 is larger than the central point coordinate cx of the number 9, and the central point coordinate cy of the number 10 is smaller than the central point coordinate of the number 9;
(11) the central point coordinate cx of the number 11 is larger than the central point coordinate cx of the number 10, and the central point coordinate cy of the number 11 is smaller than the central point coordinate of the number 10;
(12) the central point coordinate cx of the number 12 is larger than the central point coordinate cx of the number 11, and the central point coordinate cy of the number 12 is smaller than the central point coordinate of the number 11;
if all the 12 judgments are met, the score is 1, otherwise, the score is 0;
thirdly, the coordinate system adopted before is a rectangular coordinate system, and all the digital distribution is converted into a polar coordinate system when the digital distribution is judged; taking the clock center as the origin of a polar coordinate system, the angles of the numbers 1 and 2 are 0-90 degrees, the angles of the numbers 4 and 5 are 270-360 degrees, the angles of the numbers 7 and 8 are 180-270 degrees, and the angles of the numbers 10 and 11 are 90-180 degrees;
determining a straight line A according to the coordinates of the central points of the numbers 3 and 9, determining a straight line B according to the coordinates of the central points of the numbers 6 and 12, and then calculating A, B the coordinates of the middle points of the two straight lines to be recorded as (clock _ x, clock _ y); the midpoint is taken as the midpoint of the whole picture clock image; firstly, coordinate transformation is carried out, the coordinates of the former rectangular coordinate system take the upper left corner of a picture as an origin, and now (clock _ x, clock _ y) is changed to be taken as the origin; the transformation formula is as follows:
x′＝x-clock_x (3)
y′＝y-clock_y
then, the transformed rectangular coordinates are converted into a polar coordinate format, and the transformation formula is as follows:
θ＝arctan(y/x) (4)
in order to correct clock rotation deviation possibly caused when the picture acquisition device acquires a clock picture, subtracting a number 3 from the polar coordinate angles of the numbers 1-12;
then the polar angles of 8 numbers are determined:
(1)0< number 1 angle < 90;
(2)0< number 2 angle < 90;
(3)90< number 11 angle < 180;
(4)90< number 10 angle < 180;
(5)180< number 8 angle < 270;
(6)180< number 7 angle < 270;
(7)270< number 5 angle < 360;
(8)270< number 4 angle < 360;
if all the six judgments are met, the score is 1, otherwise the score is 0;
respectively checking whether the coordinates of the upper left corner and the lower right corner corresponding to the hour hand label and the minute hand label appear (0,0), if so, indicating that the pointer corresponding to the coordinates is not detected in the current image, and the score is 0; if (0,0) does not appear in 2 sets of coordinates, the two pointers are detected in the picture, and the score is 1;
(V) checking the coordinates of the pointer 14 and the upper left corner and the lower right corner of the number 4, if (0,0) is found, indicating that the pointer 14 or the number 4 is not detected, and the score is zero, otherwise, continuing to judge;
the coordinates of the upper left corner and the lower right corner of the pointer 14 form straight lines, the distances d3 and d4 from the center points of the numbers 3 and 4 to the straight lines and the distance d34 between the center points of the numbers 3 and 4 are respectively calculated, and the judgment is carried out according to the following formula:
((d3-d4))/d34≥δ
wherein, δ is a threshold, if the condition is satisfied, the score is 1, otherwise, the score is 0;
(VI) checking the coordinates of the pointer 13 and the upper left corner and the lower right corner of the number 8, if (0,0) is found, indicating that the pointer 13 or the number 4 is not detected, and the score is zero, otherwise, continuing to judge;
the coordinates of the upper left corner and the lower right corner of the pointer 13 are (x1, y1), (x2, y2), the coordinates of the upper right corner (x2, y1) and the coordinates of the lower left corner (x1, y2), respectively; forming a straight line D by coordinates of the upper right corner and the lower left corner, and respectively calculating distances D7, D8 and D9 from the numbers 7,8 and 9 to the straight line D; if d8 is minimum, the score is 1, otherwise it is 0;
(VII) checking the coordinates of the upper left corner and the lower right corner of the pointers 13 and 14, if (0,0) is found, indicating that the pointer 13 or 14 is not detected, and the score is zero, otherwise, continuing to judge;
the lengths of the pointers are given by the following formulas, and the lengths d13, d14 of the pointer 13 and the pointer 14 are calculated respectively; then, the judgment is carried out according to the following formula:
(d13-d14)/d13≥ω；
where ω is a threshold, if the above determination is satisfied, the score is 1, otherwise it is 0.
3. The automated cerebrovascular disease neurological impairment patient cognitive function screening system of claim 2, wherein the image capture device is a cell phone camera, a computer camera, an ipad or other types of photographable devices.
4. The automatic screening system for cognitive functions of patients with cerebrovascular disease and nerve injury according to claim 2, wherein the adjustment coefficients α and β take the values of α -0.7-1.0 and β -0.9-1.2.
5. The automated cerebrovascular disease neurological impairment patient cognitive function screening system of claim 2, wherein the threshold δ is taken to be-0.1-0.3; the threshold value omega is 0.10-0.25.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911176388.XA CN110859599B (en) | 2019-11-26 | 2019-11-26 | Automatic cognitive function screening system for cerebrovascular disease nerve injury patients |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911176388.XA CN110859599B (en) | 2019-11-26 | 2019-11-26 | Automatic cognitive function screening system for cerebrovascular disease nerve injury patients |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110859599A true CN110859599A (en) | 2020-03-06 |
| CN110859599B CN110859599B (en) | 2022-05-20 |
Family
ID=69656614
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911176388.XA Active CN110859599B (en) | 2019-11-26 | 2019-11-26 | Automatic cognitive function screening system for cerebrovascular disease nerve injury patients |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110859599B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111091910A (en) * | 2019-12-17 | 2020-05-01 | 中国科学院自动化研究所 | Intelligent evaluation system based on drawing clock test |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017191275A1 (en) * | 2016-05-06 | 2017-11-09 | Ai Mediq S.A. | Device and method for improving cognitive and functional abilities using hypoxic and hyperoxic gas mixtures |
| CN108197564A (en) * | 2017-12-29 | 2018-06-22 | 复旦大学附属中山医院 | A kind of assessment system and method for drawing clock experiment |
| CN108492870A (en) * | 2017-02-23 | 2018-09-04 | 中国科学院软件研究所 | Picture clock test detection method based on digital pen and system |
| WO2018169098A1 (en) * | 2017-03-13 | 2018-09-20 | 중앙대학교 산학협력단 | Cognitive and memory test device and method |
| CN110384484A (en) * | 2019-07-25 | 2019-10-29 | 科大讯飞股份有限公司 | A kind of pattern evaluation method, device, equipment and storage medium |
-
2019
- 2019-11-26 CN CN201911176388.XA patent/CN110859599B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017191275A1 (en) * | 2016-05-06 | 2017-11-09 | Ai Mediq S.A. | Device and method for improving cognitive and functional abilities using hypoxic and hyperoxic gas mixtures |
| CN108492870A (en) * | 2017-02-23 | 2018-09-04 | 中国科学院软件研究所 | Picture clock test detection method based on digital pen and system |
| WO2018169098A1 (en) * | 2017-03-13 | 2018-09-20 | 중앙대학교 산학협력단 | Cognitive and memory test device and method |
| CN108197564A (en) * | 2017-12-29 | 2018-06-22 | 复旦大学附属中山医院 | A kind of assessment system and method for drawing clock experiment |
| CN110384484A (en) * | 2019-07-25 | 2019-10-29 | 科大讯飞股份有限公司 | A kind of pattern evaluation method, device, equipment and storage medium |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111091910A (en) * | 2019-12-17 | 2020-05-01 | 中国科学院自动化研究所 | Intelligent evaluation system based on drawing clock test |
| CN111091910B (en) * | 2019-12-17 | 2023-11-03 | 中国科学院自动化研究所 | Intelligent evaluation system based on painting clock test |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110859599B (en) | 2022-05-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103249358A (en) | Medical image processing device | |
| CN112967285B (en) | Chloasma image recognition method, system and device based on deep learning | |
| CN111341437B (en) | Digestive tract disease judgment auxiliary system based on tongue image | |
| Sivasankar et al. | Retracted: FCM based iris image analysis for tissue imbalance stage identification | |
| CN115862819B (en) | Medical image management method based on image processing | |
| CN116705300A (en) | Medical decision assistance method, system and storage medium based on sign data analysis | |
| CN110859599B (en) | Automatic cognitive function screening system for cerebrovascular disease nerve injury patients | |
| CN116504392A (en) | Intelligent auxiliary diagnosis prompt system based on data analysis | |
| CN115185936A (en) | Medical clinical data quality analysis system based on big data | |
| CN114343577A (en) | Cognitive function evaluation method, terminal device, and computer-readable storage medium | |
| CN113688942A (en) | Method and device for automatically evaluating cephalic and lateral adenoid body images based on deep learning | |
| CN113539486A (en) | Health state identification system based on traditional Chinese medicine facial and tongue manifestation dynamic change | |
| CN112634221A (en) | Image and depth-based cornea level identification and lesion positioning method and system | |
| CN109711306B (en) | Method and equipment for obtaining facial features based on deep convolutional neural network | |
| CN115359548B (en) | Handheld intelligent pupil detection device and detection method | |
| CN111048202A (en) | Intelligent traditional Chinese medicine diagnosis system and method thereof | |
| CN115019972A (en) | Medical image data sharing system based on big data analysis | |
| CN111274953B (en) | Method and system for judging pain according to expression | |
| CN110931129A (en) | Painting and drawing computer analysis method for evaluating schizophrenia mental state | |
| CN113158822A (en) | Eye detection data classification method and device based on cross-modal relationship reasoning | |
| CN113033330A (en) | Tongue posture abnormality distinguishing method based on light convolutional neural network | |
| Haja et al. | Advancing glaucoma detection with convolutional neural networks: a paradigm shift in ophthalmology | |
| CN113034451A (en) | Chest DR image identification method based on deep learning | |
| CN117671774B (en) | Face emotion intelligent recognition analysis equipment | |
| CN111091910A (en) | Intelligent evaluation system based on drawing clock test |
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 |