CN104715157A - Cognition impairment evaluating system and method based on clock drawing test - Google Patents

Abstract

The invention provides a cognition impairment evaluating system and method based on a clock drawing test. A detection device comprises an image acquisition device for acquiring images of a drawn clock and an information processing device connected with the image acquisition device through signals. The information processing device compares clock images acquired by the image acquisition device with the set standard and compares the compared result with the set stand for evaluation so as to obtain the total score. The cognition impairment evaluating system and method have the advantages that the errors and careless mistakes caused by manual factors are avoided, the test time is shortened, the test reliability is improved, workload is lowered for busy workers, a great amount of paper is saved, and a great contribution is made to the environmental causes.

Description

A kind of cognition dysfunction evaluating system and method based on drawing clock experiment

Technical field

The present invention relates to the technical field of Medical Devices, referring more particularly to a kind of based on drawing kind of the cognition dysfunction evaluating system of experiment and a method.

Background technology

Drawing clock experiment (Clock drawing test, CDT) is conventional, widely used cognition dysfunction screening instruments, fruitful in examination PD.Its method of testing allows tester on paper, draw clock and watch, time figure is filled out to clock and watch, then hour hands, minute hand point to official hour respectively, are used for checking the comprehension of old man, action executing ability, digital knowledge, planned, abstract thinking ability, image reconstruction ability, notice, visual memory, visual space ability etc.

The assessment of traditional picture clock experiment allows tester on a blank sheet of paper, draw a clock, and mark the time of specifying, and requires to complete in 10min by inspection old man.Doctor's subjectivity provide scoring.This assessment mode lacks objective quantitative.

The shortcoming of prior art one

Appraisal procedure on present clinical medicine compares loses time, and this block paper using of survey is also measured and also compared waste, also likely there will be the situation of mistake of statistics when staff calculates, and assessment result information is unfavorable for storing and management.And also learning for result wants doctor hand-written, aobvious is more bothersome.

Summary of the invention

The object of the invention is to overcome the deficiencies in the prior art, providing a kind of based on drawing kind of the cognition dysfunction evaluating system of experiment and a method.

The present invention is realized by above technical scheme:

The invention provides a kind of based on drawing kind of a cognition dysfunction evaluating system for experiment, this pick-up unit comprises: the image collecting device gathering drawn clock and watch image;

Signal conditioning package, is connected with described image collecting device signal, and contrasts according to the clock and watch image of described image acquisition device and the standard of setting, and contrasts according to the result of contrast and the standards of grading of setting, obtains overall score.

Preferably, described signal conditioning package comprises image pre-processing module, the image segmentation module be connected with described image pre-processing module signal, the image analysis module be connected with described image segmentation module signal, and the picture recognition module be connected with described image analysis module.

Preferably, described image pre-processing module comprises gradation processing module, the smoothing denoising module be connected with described gradation processing module signal, and image is carried out the processing module of binaryzation.

Preferably, the Slant Rectify module that described image is corrected also is comprised.

The invention provides a kind of based on drawing kind of a detection method for the cognition dysfunction evaluating system of experiment, the method comprises the following steps:

Gather drawn image;

The clock and watch image of collection and the standard of setting are contrasted, contrasts according to the result of contrast and the standards of grading of setting, obtain overall score.

Preferably, described the image of collection and the standard of setting to be contrasted, and contrast according to the result of contrast and the grade of setting, obtain level evaluation and specifically comprise:

Pre-service is carried out to image;

Internal periphery filling is carried out to pretreated image, and then extracts blank map as outline;

Morphologic thinning processing is carried out to the outline extracted;

Utilize the edge algorithms of chain code following to realize the judgement whether closed clock and watch outward flange, contrast standards of grading analyze the scoring of this step;

When clock and watch close, calculate the circularity of image, and compare analyze the scoring of this step with standards of grading;

Identify institute in circle by template matches and mark numeral, and to analyze between institute's timestamp correctly applicable, provide the scoring of this step;

Calculate each step to mark, provide the last gross score drawing clock experiment.

Preferably, describedly pre-service carried out to image specifically comprise:

Gray proces is carried out to image;

To the smoothing denoising of image;

Binary conversion treatment is carried out to image;

Numeral on image is normalized;

Slant Rectify is carried out to image.

Preferably, described Internal periphery filling is carried out to pretreated image, and then extract blank map and be specially as outline: utilize Opencv to carry out holes filling to bianry image, obtain a black disk, then extract contour curve.

Preferably, the described edge algorithms of chain code following that utilizes realizes the judgement whether closed clock and watch outward flange, contrast standards of grading analyze the scoring of this step and are specially: calculate described contour curve girth and area, and whether the peripheral curve going out clock and watch image according to the comparative analysis of contour area and area threshold closes.

Preferably, describedly identify institute in circle by template matches and mark numeral, and to analyze between institute's timestamp correctly applicable, the scoring providing this step is specially:

First image is divided into 8 parts of equal rectangles, adds up the number of black pixel point in each rectangle as feature, totally 8 features; Then draw two lines respectively by the number through the black pixel point on two lines on image trisection statistics horizontal direction and vertical direction at image level direction, vertical direction, it can be used as 4 features; In last statistical figure image, the number of all black pixel points is as the 13rd feature; The standards of grading of all 13 features and setting are carried out contrasting and being marked.

Beneficial effect of the present invention is: avoid error and careless mistake that human factor causes, shorten the test duration, improve the fiduciary level of test, alleviate workload, save a large amount of paper, for environmental protection cause makes tremendous contribution to busy voluntary labor service personnel.

Accompanying drawing explanation

Fig. 1 is a structured flowchart for the cognition dysfunction evaluating system based on picture kind of experiment that the embodiment of the present invention provides;

Fig. 2 is the process flow diagram of the detection method that the embodiment of the present invention provides.

Embodiment

In order to make object of the present invention, technical scheme and advantage clearly understand, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein only in order to explain the present invention, be not intended to limit the present invention.

Refer to Fig. 1, Fig. 1 is provided by the invention based on drawing kind of a structured flowchart for the cognition dysfunction evaluating system of experiment.

Embodiments providing a kind of based on drawing kind of a cognition dysfunction evaluating system for experiment, should comprise based on drawing kind of the cognition dysfunction evaluating system of experiment: the image collecting device gathering drawn clock and watch image;

Signal conditioning package, is connected with described image collecting device signal, and contrasts according to the clock and watch image of described image acquisition device and the standard of setting, and contrasts according to the result of contrast and the standards of grading of setting, obtains overall score.

In the above-described embodiments, by signal conditioning package, gathered clock and watch image is processed, and the standards of grading of result and setting are contrasted, obtain overall score, thus the situation of detected personnel can be known, greatly improve the efficiency of detection, and reduce the paper detecting waste.

Described signal conditioning package wherein comprises image pre-processing module, the image segmentation module be connected with described image pre-processing module signal, the image analysis module be connected with described image segmentation module signal, and the picture recognition module be connected with described image analysis module.Wherein, described image pre-processing module comprises gradation processing module, the smoothing denoising module be connected with described gradation processing module signal, and image is carried out the processing module of binaryzation.And to the Slant Rectify module that described image is corrected.By image pre-processing module, image segmentation module, image analysis module and picture recognition module, clock and watch image is processed, feature effectively can be obtained, thus contrast with the standard of setting, this standard is a standard round, the score of this clock and watch image can be obtained by the result of contrast, thus obtain the cognitive assessment of detected personnel corresponding to this score.

In addition, the invention provides a kind of based on drawing kind of a detection method for the cognition dysfunction evaluating system of experiment, the method comprises the following steps:

Gather drawn image;

The clock and watch image of collection and the standard of setting are contrasted, contrasts according to the result of contrast and the standards of grading of setting, obtain overall score.

Wherein, described the image of collection and the standard of setting to be contrasted, and contrast according to the result of contrast and the grade of setting, obtain level evaluation and specifically comprise:

Pre-service is carried out to image; Describedly pre-service is carried out to image specifically comprise: gray proces is carried out to image; To the smoothing denoising of image; Binary conversion treatment is carried out to image; Numeral on image is normalized; Slant Rectify is carried out to image.

Internal periphery filling is carried out to pretreated image, and then extracts blank map as outline; Concrete, utilize Opencv to carry out holes filling to bianry image, obtain a black disk, then extract contour curve,

Morphologic thinning processing is carried out to the outline extracted;

Utilize the edge algorithms of chain code following to realize the judgement whether closed clock and watch outward flange, contrast standards of grading analyze the scoring of this step; Concrete, calculate described contour curve girth and area, whether the peripheral curve going out clock and watch image according to the comparative analysis of contour area and area threshold closes.

When clock and watch close, calculate the circularity of image, and compare analyze the scoring of this step with standards of grading;

Identify institute in circle by template matches and mark numeral, and to analyze between institute's timestamp correctly applicable, provide the scoring of this step; Concrete, first image is divided into 8 parts of equal rectangles, adds up the number of black pixel point in each rectangle as feature, totally 8 features; Then draw two lines respectively by the number through the black pixel point on two lines on image trisection statistics horizontal direction and vertical direction at image level direction, vertical direction, it can be used as 4 features; In last statistical figure image, the number of all black pixel points is as the 13rd feature; The standards of grading of all 13 features and setting are carried out contrasting and being marked.

Calculate each step to mark, provide the last gross score drawing clock experiment.

Conveniently to the understanding of the method that the present embodiment provides, below in conjunction with Fig. 2, it is described in detail.

This method is divided into four steps such as Image semantic classification, Iamge Segmentation, graphical analysis and image recognition, obtains overall score by inquartation scoring, concrete,

(1) Image semantic classification

Image semantic classification be image procossing the most basic be also a most important link, directly affect the result of graphical analysis and identification.The pre-service of clock and watch identification mainly contains: gray proces, binaryzation, denoising, refinement, normalization.

1. gray proces: due to 24 that collect very color BMP files, it is made up of R/G/B tri-component values, and its each point represents R/G/B respectively by three bytes, directly processes, operand and difficulty larger, so coloured image will be converted into gray level image.Conversion formula:

$p\left(k\right)=\frac{1}{N}\underset{f(i,j)=k}{\mathrm{\Σ}}N(i,j)---(3-3)$

2. smoothing denoising: smoothing denoising is mainly used to eliminate the noise that individual input habit or the unconscious shake of hand cause.Reduce noise and usually adopt linear smoothing operator or nonlinear smoothing operator.

3. binaryzation: the binaryzation of image is exactly the image that image is set to two gray-scale values only having 0 and 255 by the suitable threshold value of selection one, and image is black and white effect.Common Binarization methods mainly contains fixed threshold binarization method, Da-Jin algorithm (Otsu) [44], Dither matrix binarization method, basic adaptive threshold binarization method etc.The binaryzation of this system adopts Otsu partitioning algorithm, and Da-Jin algorithm is proposed by large Tianjin for 1979.Its basic thought utilizes difference between the grey level histogram of image and target and background, dynamically determines the threshold value of Iamge Segmentation.Suppose that piece image has N number of pixel, there is L gray level, pixel (i, j) gray-scale value is N (i, j), then N (i, j) span is [0, L-1], the probability that note p (k) is gray-scale value k, then grey level histogram can represent with formula 3-4:

$p\left(k\right)=\frac{1}{N}\underset{f(i,j)=k}{\mathrm{\Σ}}N(i,j)---(3-4)$

What to suppose with optimal segmenting threshold T be target area and background area by Iamge Segmentation.Wherein, target area and background area, use respectively f (x, y)≤t} and t < f (x, y) < L} represents, then target area proportion:

$w0\left(t\right)=\underset{0\&le;k\&le;t}{\mathrm{\&Sigma;}}p\left(k\right)---(3-5)$

Background area proportion is:

$w1\left(t\right)=\underset{t\&le;k\&le;L-1}{\mathrm{\&Sigma;}}p\left(k\right)$

(3-6)

Target mean is:

$u0\left(t\right)=\underset{0\&le;k\&le;t}{\mathrm{\&Sigma;}}\mathrm{kp}\left(t\right)/w0\left(t\right)$

(3-7)

Background mean value is:

$u1\left(t\right)=\underset{t\&le;k\&le;L-1}{\mathrm{\&Sigma;}}\mathrm{kp}\left(k\right)/w1\left(t\right)$

(3-8)

Grand mean is:

u＝w0(t)u0(t)+w1(t)u1(t)

(3-9)

Utilize the formula of Otsu algorithm computed image optimal threshold:

$g=\underset{0\&le;t\&le;L-1}{\mathrm{Arg}\max }[w0{(u0\left(t\right)-u)}^2+w1\left(t\right){(u1\left(t\right)-u)}^2]$

(3-10)

4. refinement: so-called refinement is exactly reduce some pixels by original figure, until remaining single pixel, but still can keep the skeleton of former figure.The method comparison of refinement is many, and native system selects Pavlidis thinning algorithm.

5. normalization: the Digital size write due to people differs, deals with cumbersome like this, therefore, when carrying out pretreated to hand-written digital picture, is normalized by handwritten numeral image.Large digital picture is carried out the digital picture narrowing down to fixed size, can not only discrimination be improved like this, also can improve the efficiency of process simultaneously.

6. slant correction: in people's writing process, all can have certain inclination (degree of tilt is generally at 0-45), if not to digital correct image process, directly carry out feature extraction, and will the discrimination of influential system.

(2) graphical analysis

Opencv is utilized to carry out holes filling to bianry image, obtain a black disk, then contour curve is extracted, and calculate its girth and area, and whether the peripheral curve going out clock and watch according to the comparative analysis of contour area and area threshold closes. the circular degree of the peripheral curve of clock and watch is reacted according to circularity.

1. edge extracting: edge extracting also i.e. contours extract, the edge of image is the most basic feature of image, in order to better analyze image, needs to extract the image border of concentrated great amount of images information.At present, the method at common extraction edge has Canny algorithm, Sobel algorithm, Prewitt sharpening algorithm etc., and native system selects Canny algorithm to carry out contours extract.

2. round rate analysis: before introduction circle rate, first introduce lower circularity, circularity is also known as complexity, and formula is: C=L2/A, and in formula, L is the girth of circle, and A is the area in circle region.Documents and materials illustrate, when figure is bowlder, C has minimum value 4 π; When being other figures, C>4 π.And picture shape is more complicated, C value is larger.When after the area and perimeter calculating region, its circularity of circularity formulae discovery just can be utilized.Because circularity is still not directly perceived when analyzing, following definition circle rate: P=4 π/C, in formula, C is circularity, circle rate P size is the decimal between 0 ~ 1, when figure is circular, P value is 1 to the maximum, when for other figures, P value be less than 1 decimal, so can the circular degree of more convenient reaction.

(3) image recognition

The identification of the image recognition section of native system mainly numeral, first needs to build sample characteristics storehouse before recognition, and then the feature will extracted from image to be identified, brings with Sample Storehouse comparison, draws and identify digital value.

1. sample characteristics storehouse: the standard that Sample Storehouse is coupling and identifies, utilizes 13 feature extractions, and extract the feature of sample, be saved in Sample Storehouse, that just establishes sample characteristics storehouse like this.

2. feature extraction: feature extraction is vital link in image recognition, directly affects recognition result.The method of feature extraction has a lot, and native system adopts a kind of method of range statistics feature---13 interest point detect methods.Its feature is respectively: first image is divided into 8 parts of equal rectangles, adds up the number of black pixel point in each rectangle as feature, 8 parts of 8 features; Then draw two lines respectively by the number through the black pixel point on two lines on image trisection statistics horizontal direction and vertical direction at image level direction, vertical direction, it can be used as 4 features; In last statistical figure image, the number of all black pixel points is as the 13rd feature [45].

3. stencil matching: fairly simple, effective method in native system selective recognition method---template matching method [46,47].Its ultimate principle: suppose M classification: n1, n2 ... nm, each class, by several vector representations, as ni class, has:

$\mathrm{Xi}=\left[\begin{array}{c}\mathrm{xi}1\\ \mathrm{xi}2\\ ...\\ \mathrm{xin}\end{array}\right]---(3-11)$

Handwritten numeral X to be identified arbitrarily has:

(3) image recognition

The identification of the image recognition section of native system mainly numeral, first needs to build sample characteristics storehouse before recognition, and then the feature will extracted from image to be identified, brings with Sample Storehouse comparison, draws and identify digital value.

1. sample characteristics storehouse: the standard that Sample Storehouse is coupling and identifies, utilizes 13 feature extractions, and extract the feature of sample, be saved in Sample Storehouse, that just establishes sample characteristics storehouse like this.

2. feature extraction: feature extraction is vital link in image recognition, directly affects recognition result.The method of feature extraction has a lot, and native system adopts a kind of method of range statistics feature---13 interest point detect methods.Its feature is respectively: first image is divided into 8 parts of equal rectangles, adds up the number of black pixel point in each rectangle as feature, 8 parts of 8 features; Then draw two lines respectively by the number through the black pixel point on two lines on image trisection statistics horizontal direction and vertical direction at image level direction, vertical direction, it can be used as 4 features; In last statistical figure image, the number of all black pixel points is as the 13rd feature [45].

3. stencil matching: fairly simple, effective method in native system selective recognition method---template matching method [46,47].Its ultimate principle: suppose M classification: n1, n2 ... nm, each class, by several vector representations, as ni class, has: handwritten numeral X to be identified arbitrarily has:

$\mathrm{Xi}=\left[\begin{array}{c}\mathrm{xi}1\\ \mathrm{xi}2\\ ...\\ \mathrm{xin}\end{array}\right]---(3-11)$

Calculate the distance D (xi, x) between xi and x, if there is certain i value, make

D(Xi,X)＜D(Xj,X),j＝1，2，...，M，i≠j

(3-13)

Then X belongs to ni.So just the numeral of segmentation can be carried out having classified.

Drawing clock experiment utilizes Digital Image Processing to carry out quantitative test to its inquartation, utilizes circle rate to carry out grade classification to disk in program.Experimental data is as follows:

Table 3-4 circle rate table of grading

Can be found out by said method, native system achieves the digitizing of picture clock experiment, avoid error and careless mistake that human factor causes, shorten the test duration, improve the fiduciary level of test, alleviate workload to busy voluntary labor service personnel, save a large amount of paper, for environmental protection cause makes tremendous contribution.

These are only preferred embodiment of the present invention, not in order to limit the present invention, all any amendments done within the spirit and principles in the present invention, equivalent replacement and improvement etc., all should be included within protection scope of the present invention.

Claims (10)

1. based on drawing kind of a cognition dysfunction evaluating system for experiment, it is characterized in that, comprise: the image collecting device gathering drawn clock and watch image;

Signal conditioning package, is connected with described image collecting device signal, and contrasts according to the clock and watch image of described image acquisition device and the standard of setting, and contrasts according to the result of contrast and the standards of grading of setting, obtains overall score.

2. according to claim 1 based on drawing kind of a cognition dysfunction evaluating system for experiment, it is characterized in that, described signal conditioning package comprises image pre-processing module, the image segmentation module be connected with described image pre-processing module signal, the image analysis module be connected with described image segmentation module signal, and the picture recognition module be connected with described image analysis module.

3. according to claim 1 based on drawing kind of a cognition dysfunction evaluating system for experiment, it is characterized in that, described image pre-processing module comprises gradation processing module, the smoothing denoising module be connected with described gradation processing module signal, and image is carried out the processing module of binaryzation.

4. according to claim 3 based on drawing kind of a cognition dysfunction evaluating system for experiment, it is characterized in that, also comprise the Slant Rectify module that described image is corrected.

5., based on drawing kind of a detection method for the cognition dysfunction evaluating system of experiment, it is characterized in that, comprise the following steps:

Gather drawn image;

The clock and watch image of collection and the standard of setting are contrasted, contrasts according to the result of contrast and the standards of grading of setting, obtain overall score.

6. detection method as claimed in claim 5, is characterized in that, describedly the image of collection and the standard of setting is contrasted, and contrasts according to the grade of the result contrasted and setting, obtains level evaluation and specifically comprises:

Pre-service is carried out to image;

Internal periphery filling is carried out to pretreated image, and then extracts blank map as outline;

Morphologic thinning processing is carried out to the outline extracted;

Utilize the edge algorithms of chain code following to realize the judgement whether closed clock and watch outward flange, contrast standards of grading analyze the scoring of this step;

When clock and watch close, calculate the circularity of image, and compare analyze the scoring of this step with standards of grading;

Identify institute in circle by template matches and mark numeral, and to analyze between institute's timestamp correctly applicable, provide the scoring of this step;

Calculate each step to mark, provide the last gross score drawing clock experiment.

7. detection method as claimed in claim 6, is characterized in that, describedly carries out pre-service to image and specifically comprises:

Gray proces is carried out to image;

To the smoothing denoising of image;

Binary conversion treatment is carried out to image;

Numeral on image is normalized;

Slant Rectify is carried out to image.

8. detection method as claimed in claim 7, it is characterized in that, described Internal periphery filling is carried out to pretreated image, and then extract blank map and be specially as outline: utilize Opencv to carry out holes filling to bianry image, obtain a black disk, then extract contour curve.

9. detection method as claimed in claim 8, it is characterized in that, the described edge algorithms of chain code following that utilizes realizes the judgement whether closed clock and watch outward flange, contrast standards of grading analyze the scoring of this step and are specially: calculate described contour curve girth and area, and whether the peripheral curve going out clock and watch image according to the comparative analysis of contour area and area threshold closes.

10. method as claimed in claim 9, is characterized in that, describedly identifies institute in circle by template matches and mark numeral, and analyzes between institute's timestamp correctly applicable, and the scoring providing this step is specially:

First image is divided into 8 parts of equal rectangles, adds up the number of black pixel point in each rectangle as feature, totally 8 features; Then draw two lines respectively by the number through the black pixel point on two lines on image trisection statistics horizontal direction and vertical direction at image level direction, vertical direction, it can be used as 4 features; In last statistical figure image, the number of all black pixel points is as the 13rd feature; The standards of grading of all 13 features and setting are carried out contrasting and being marked.