CN111091910A - Intelligent evaluation system based on drawing clock test - Google Patents
Intelligent evaluation system based on drawing clock test Download PDFInfo
- Publication number
- CN111091910A CN111091910A CN201911300342.4A CN201911300342A CN111091910A CN 111091910 A CN111091910 A CN 111091910A CN 201911300342 A CN201911300342 A CN 201911300342A CN 111091910 A CN111091910 A CN 111091910A
- Authority
- CN
- China
- Prior art keywords
- clock
- detection module
- area
- included angle
- contour
- 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
- 238000012360 testing method Methods 0.000 title claims abstract description 38
- 238000011156 evaluation Methods 0.000 title claims abstract description 31
- 238000001514 detection method Methods 0.000 claims abstract description 105
- 239000013598 vector Substances 0.000 claims description 53
- 238000004364 calculation method Methods 0.000 claims description 28
- 238000007781 pre-processing Methods 0.000 claims description 8
- 238000013527 convolutional neural network Methods 0.000 claims description 5
- 230000035945 sensitivity Effects 0.000 claims description 5
- 238000009877 rendering Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 3
- 238000010422 painting Methods 0.000 claims 7
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 238000000034 method Methods 0.000 description 26
- 230000006870 function Effects 0.000 description 7
- 208000024827 Alzheimer disease Diseases 0.000 description 5
- 238000012549 training Methods 0.000 description 5
- 206010012289 Dementia Diseases 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 206010039966 Senile dementia Diseases 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000011218 segmentation Effects 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 208000010877 cognitive disease Diseases 0.000 description 2
- 230000003920 cognitive function Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 208000028698 Cognitive impairment Diseases 0.000 description 1
- 206010020772 Hypertension Diseases 0.000 description 1
- 201000006347 Intellectual Disability Diseases 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000010391 action planning Effects 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000001631 hypertensive effect Effects 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 208000024891 symptom Diseases 0.000 description 1
- 230000001225 therapeutic effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Public Health (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- General Health & Medical Sciences (AREA)
- Geometry (AREA)
- Pathology (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the field of artificial intelligence, in particular to an intelligent evaluation system based on a drawing clock test, and aims to solve the problem that scoring results given by different doctors are different. The system comprises a roundness detection module, an integrity detection module, a correctness detection module and a grading module. The roundness detection module is configured to detect a clock contour in the image to be evaluated, draw a minimum circumcircle of the clock contour, and calculate the ratio of the area surrounded by the contour to the area of the minimum circumcircle; the integrity detection module is configured to detect whether the number, the hour hand and the minute hand exist in the image to be evaluated and acquire corresponding coordinate values; the correctness detection module is configured to respectively detect whether the directions of the digital position, the hour hand and the minute hand are correct or not according to the coordinate values; the scoring module is configured to score according to the detection result. The invention solves the problem that the patient is inconvenient to go to the hospital to participate in the test, and effectively avoids the problem that the scoring results are different due to different doctors.
Description
Technical Field
The invention relates to the field of artificial intelligence, in particular to an intelligent evaluation system based on a drawing clock test.
Background
With the advent of aging society, diseases such as dementia and intellectual disability related to cognitive impairment of the elderly have attracted extensive attention in various social fields. However, since early symptoms of Alzheimer's disease are not obvious and can be easily overlooked, most Alzheimer's disease patients reach the middle and late stages of the disease and miss the time for optimal therapeutic intervention. Thus, studies have indicated that populations at risk, including hypertensive patients, should be subjected to periodic clock tests for cognitive dysfunction.
The Clock Drawing Test (CDT) is simple and easy to implement, high in accuracy and small in cultural relevance, can comprehensively reflect the dementia screening Test of cognitive function, and can be used as an early screening tool for detecting senile dementia. The method can detect the reconstruction capability of the visual memory graph of the old, the action planning performance, the execution function, the anti-interference capability and other performances. The related data show that the accuracy of the clock drawing for testing the senile dementia is as high as 80-90%.
The drawing test has various scoring forms, but its nature has no more than the following aspects: 1. drawing a locked circle (dial); 2. drawing scales uniformly distributed in a circle; 3. marking out numbers corresponding to the scales; 4. the hour and minute hands are drawn at the time specified by the tester. The resulting images of the bell test are ultimately scored by the clinician. However, the above approaches are limited by insufficient medical resources and inconvenience in the patients to visit the hospital for evaluation, and there are fluctuations and differences in the scoring results given by different doctors.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides an intelligent evaluation system based on a clock drawing test, which solves the problem that a patient is inconvenient to go to a hospital to participate in a test, and effectively avoids the problem that scoring results are different due to different doctors.
The invention provides an intelligent evaluation system based on a drawing clock test, which comprises: the device comprises a roundness detection module, an integrity detection module, a correctness detection module and a grading module;
the roundness detection module is configured to: detecting a clock contour in an image to be evaluated, drawing a minimum circumcircle of the clock contour, and calculating the ratio of the area enclosed by the clock contour to the area of the minimum circumcircle;
the integrity detection module is configured to: detecting whether the number, the hour hand and the minute hand exist in the image to be evaluated, and acquiring corresponding coordinate values;
the correctness detection module is configured to: respectively detecting whether the digital positions are correct and whether the directions of the hour hand and the minute hand are correct according to the coordinate values;
the scoring module is configured to: and scoring the clock drawn in the image to be evaluated according to the detection results of the roundness detection module, the integrity detection module and the correctness detection module.
Preferably, the roundness detection module includes: the device comprises a preprocessing unit, a contour detection unit, an external circle drawing unit and an area rate calculation unit;
the pre-processing unit is configured to: converting the image to be evaluated into a gray scale image;
the contour detection unit is configured to: detecting the clock contour in the grey-scale map;
the circumscribed circle drawing unit is configured to: drawing a minimum circumscribed circle of the clock contour;
the area ratio calculation unit is configured to: and respectively calculating the area enclosed by the clock contour and the area of the minimum circumcircle, and solving the ratio of the area enclosed by the clock contour and the area of the minimum circumcircle.
Preferably, the area ratio calculation unit includes: a first area calculation subunit, a second area calculation subunit, and a ratio calculation subunit;
the first area calculation subunit is configured to: calculating the number of pixels with the internal pixel value of 255 of the clock contour, and further calculating the area enclosed by the clock contour;
the second area calculation subunit is configured to: calculating the area of the minimum circumcircle;
the ratio calculation subunit is configured to: calculating the ratio of the area enclosed by the clock contour and the minimum circumcircle area.
Preferably, the integrity detection module comprises: a convolutional neural network based on masked regions.
Preferably, the correctness detecting module includes: the digital position detection unit and the hour-minute hand pointing detection unit;
the digital position detection unit is configured to: respectively detecting whether the positions of the 12 numbers in the image to be evaluated are accurate or not;
the hour and minute hand pointing detection unit is configured to: and respectively detecting whether the directions of the hour hand and the minute hand in the image to be evaluated are accurate.
Preferably, the digital position detecting unit includes: the device comprises a vector drawing subunit, an included angle calculating subunit and a digital judgment subunit;
the vector rendering subunit is configured to: respectively drawing vectors x pointing to 12 numbers by taking the circle center of the minimum circumcircle as a starting point0、x1、x2、…、x11；
The included angle calculating subunit is configured to: separately calculate the vector x0、x1、x2、…、x11And vector x0To obtain an included angle value A0、A1、A2、…、A11；
The digital judgment subunit is configured to: according to the value of included angle A0、A1、A2、…、A6Sequentially judging A1、A2、…、A6Whether each included angle is larger than a previous included angle and the absolute value of the difference value between each included angle and the previous included angle meets a preset included angle range; according to the value of included angle A7、A8、…、A11、A0Sequentially judging A7、A8、…、A11Whether each included angle is larger than the next included angle or not and the absolute value of the difference value between each included angle and the next included angle meets the preset included angle range.
Preferably, the hour and minute hand pointing direction detecting unit includes: the pointer area subunit, the pointer vector drawing subunit and the pointer judgment subunit are connected;
the pointer region molecular unit is configured to: according to the length relation of the two pointers in the image to be evaluated, an hour pointer and a minute pointer are distinguished;
the pointer vector drawing subunit is configured to: respectively drawing the vectors of an hour hand and a minute hand in the image to be evaluated and the vector of a standard hour hand and the vector of a standard minute hand pointing to a target time point;
the pointer judgment subunit is configured to: judging whether an included angle between the drawn hour hand vector and the standard hour hand vector meets a preset first error range or not; and judging whether an included angle between the drawn minute hand vector and the standard minute hand vector meets a preset second error range or not.
Preferably, the scoring module is specifically configured to: scoring the clock drawn in the image to be evaluated according to a preset scoring standard and detection results of the roundness detection module, the integrity detection module and the correctness detection module;
wherein the preset scoring standard is as follows:
if the ratio of the area surrounded by the clock contour to the minimum circumcircle area meets a preset area ratio range, recording 1 point;
if 12 numbers exist, marking 1 point;
if the positions of the 12 numbers are accurate, 1 point is marked;
if the pointer points to the correct point, the point is marked 1.
Preferably, the roundness detection module further includes: a sensitivity-improving unit;
the sensitivity-improving unit is configured to: and after the processing of improving the sensitivity is carried out on the gray level map, the gray level map is input to the outline detection unit.
Preferably, the system further comprises: an image acquisition module;
the image acquisition module is configured to: and collecting the image to be evaluated.
Preferably, the system further comprises: a display module;
the display module is configured to: and displaying the scoring result.
Compared with the closest prior art, the invention has the following beneficial effects:
the invention provides an intelligent evaluation system based on a drawing clock test, which comprises: the device comprises a roundness detection module, an integrity detection module, a correctness detection module and a grading module.
The roundness detection module measures the roundness by calculating the ratio of the area surrounded by the clock contour to the minimum circumcircle area. The least square method is used to avoid the disadvantage of the least square method to realize circle fitting: the least squares method is very severely affected by outliers, even if there are only a few outliers, the results will vary very drastically and sensitively. In addition, the module calculates the area surrounded by the clock contour by using a pixel point counting method, and is the most effective and simplest means for calculating the area of the irregular graph.
The integrity detection module adopts a fast-RCNN network structure to detect whether numbers, hour hands and minute hands exist in an image to be evaluated or not and acquire corresponding coordinate values, the performance and efficiency of the network far exceed those of the former fast-RCNN, the training speed is abnormal and rapid, and the ROIALIgh perfectly solves the problems of ROIPooling precision and pixel loss in the fast-RCNN.
And the correctness detection module respectively detects whether the digital position is correct and the directions of the hour hand and the minute hand are correct according to the coordinate value returned by the completeness detection module. The coordinate information obtained in the integrity detection module is ingeniously combined, and the accuracy of digital position, hour hand pointing accuracy and minute hand pointing accuracy is detected by utilizing the knowledge of a trigonometric function and a vector; a reliable error range is set, so that the confidence of an experimental result is higher; after a large amount of data training, the effect is obvious.
The evaluation system of the invention utilizes the computer vision principle to automatically and intelligently evaluate the bell-drawn test image, solves the problem that the patient is inconvenient to go to the hospital to participate in the test, and effectively avoids the problem that the scoring results are different due to different doctors.
Drawings
FIG. 1 is a schematic diagram of a first embodiment of an intelligent evaluation system based on a clock test according to the present invention;
fig. 2 is a schematic diagram of a second embodiment of the intelligent evaluation system based on a clock test according to the present invention.
Detailed Description
Preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only for explaining the technical principle of the present invention, and are not intended to limit the scope of the present invention.
It should be noted that the terms "first" and "second" in the description of the present invention are used for convenience of description only and do not indicate or imply relative importance of the devices, elements or parameters, and therefore should not be construed as limiting the present invention.
Fig. 1 is a schematic diagram of a main configuration of a first embodiment of the intelligent evaluation system based on a clock-drawing test according to the present invention. As shown in fig. 1, the intelligent evaluation system of the present embodiment includes: a roundness detection module 10, an integrity detection module 20, a correctness detection module 30, and a scoring module 40.
Wherein, the roundness detection module 10 is configured to: detecting the clock contour in the image to be evaluated, drawing a minimum circumcircle of the clock contour, and calculating the ratio of the area enclosed by the clock contour to the area of the minimum circumcircle; the integrity detection module 20 is configured to: detecting whether numbers, hour hands and minute hands exist in an image to be evaluated, and acquiring corresponding coordinate values; the correctness detection module 30 is configured to: respectively detecting the digital position and whether the directions of the hour hand and the minute hand are accurate or not according to the obtained coordinate values; the scoring module 40 is configured to: and scoring the clock drawn in the image to be evaluated according to the detection results of the roundness detection module, the integrity detection module and the correctness detection module. The 4 modules may all be disposed on the same CPU (e.g., x86 chip), or the roundness detection module 10 may be disposed on the DSP, and the remaining 3 modules may be disposed on the CPU.
Roundness measurement is a measure of how close an actual circle (i.e., a circle drawn by a person who participates in drawing a clock) is to an ideal circle. The circle is a two-dimensional graph which can be presented in a computer in the form of pixel points and can be used for deducing a geometric problem by a programming method, so that the geometric problem can be regarded as a computer vision problem.
In roundness measurement, the prior art generally employs one of three methods:
(1) least squares circle method: the circle center of the circle with the smallest square sum of the distances from each corresponding point on the measured circle outline to the circumference is taken as the circle center, and the radius difference of two concentric circles containing the measured circle outline is taken as the roundness error.
(2) Minimum circumcircle method: and (3) calculating the minimum circumcircle of the actual contour, and calculating the area ratio which is the area included by the actual contour/the area of the minimum circumcircle, wherein the area ratio is closer to 1, and the roundness is higher.
(3) Maximum inscribed circle method: the maximum inscribed circle of the actual contour is obtained, and the area ratio is obtained as the maximum inscribed circle area/actual contour containing area, and the roundness is higher as the area ratio approaches 1.
In the invention, a method combining a minimum circumcircle method and an area rate is adopted, namely, after the minimum circumcircle is obtained, the area rate is calculated, and then the roundness of the actual circle is judged.
Specifically, the circularity detection module 10 in this embodiment may include: the device comprises a preprocessing unit, a contour detection unit, an outer circle drawing unit and an area ratio calculation unit.
The pre-processing unit is configured to convert an image to be evaluated into a gray-scale image; the contour detection unit is configured to detect a timepiece contour in the grayscale map (e.g., the contour may be detected using the RETR _ exterior method in cv 2); the circumscribed circle drawing unit is configured to draw a minimum circumscribed circle of the clock contour (for example, a dot polygon inside the contour can be adjusted by using a pointpolygon test (cnt, (j, i), True) method in cv2 to realize image segmentation, and finally, a minEnclosingCircle method and a circle method in cv2 are used to draw the minimum circumscribed circle); the area ratio calculating unit is configured to calculate an area surrounded by the clock contour and an area of the minimum circumscribed circle, respectively, and find a ratio of the two.
Specifically, the area ratio calculating unit in the present embodiment may include: a first area calculation subunit, a second area calculation subunit, and a ratio calculation subunit.
Wherein the first area calculation subunit is configured to: calculating the number of pixels with the internal pixel value of 255 of the clock contour, and further calculating the area enclosed by the clock contour; the second area calculation subunit is configured to: calculating the area of the minimum circumscribed circle (because the minimum circumscribed circle is fitted, the circle center coordinate and the radius information are obtained at the same time); the ratio calculation subunit is configured to: the ratio of the area enclosed by the clock contour and the minimum circumscribed circle area, i.e. the area ratio, is calculated. The area ratio approaching 1 indicates that the timepiece contour is closer to an ideal circle.
The roundness measuring module in the embodiment has the following advantages:
(1) the disadvantage that the least square method is used for realizing circle fitting is avoided: the least square method is very severely influenced by the outlier, even if only a few outliers exist, the result is very violent and sensitive in change;
(2) the method of counting the pixels is used, which is the most effective and simple means for calculating the area of the unconventional graph.
Specifically, the integrity detection module 20 in this embodiment may include: mask region based convolutional neural network (Mask-RCNN). The module judges whether 12 numbers from 1 to 12 exist or not without missing, and returns the position coordinates of the 12 targets; in addition, whether the hour hand and the minute hand exist or not is judged, and the position coordinates of the two targets are returned. The returned coordinate values include: the coordinates of the center point of the bounding box enclosing each number, hour and minute hands.
If the numbers, hour and minute hands are taken as the target individuals, the best method for detecting their existence and position coordinates is fast-RCNN and Mask-RCNN, which are widely used in the industry.
In the system of this embodiment, Mask-RCNN is selected. The Mask-RCNN comprises the following components in sequence from bottom to top: backbone, FPN, RPN, anchors, RoIAlign, classification, box regression, and mask.
Among them, the backhaul is a convolutional neural network (generally using ResNet101) for extracting image features, ResNet is to train a deeper neural network (because the performance of the neural network degrades when the number of layers is higher), and ResNet uses cross-layer connection, so that training is easier; the backbone of the FPN is a standard convolutional neural network (typically ResNet50 and ResNet101) that acts as a feature extractor. The lower layers detect low-level features (edges and corners, etc.), and the higher layers detect higher-level features (cars, people, sky, etc.). The FPN improves the performance of the standard feature extraction pyramid by adding a second pyramid that can select high-level features from the first pyramid and pass them on to the bottom layer. By this process it allows the features of each level to be combined with the features of both the high and low levels. The final effect of FPN is: the shallow features can detect simple targets, while the deep features can detect more complex targets; the RPN functions to recommend object prosalass based on the region filtered out by the FPN. And extracting the features of the feature graph output by the last layer of the FPN by using a n-x-n sliding window through a convolution network. Respectively judging the target class and finely adjusting the size of a bounding box through the CLS and the REG; the Anchors is configured to generate a series of frames on the pixels of feature maps, and the size of each frame is determined by two parameters, i.e., scale ═ 128, and ratio ═ 0.5,1,1.5, so that each pixel can generate 3 frames with different sizes. The three boxes are formed by keeping the area of the boxes unchanged to change the length-width ratio of the boxes through the value of ratio, so that boxes with different sizes are generated; RoIAlign is used to extract features at a deeper level in the feature map provided by the RPN. Compared with Fast-RCNN, the Mask-RCNN has a more novel idea, and solves the problem of precision loss in Fast-RCNN; classification is used to judge what category the object payload belongs to; the Regression is used for fine adjustment of the bounding box; the prediction of the mask is carried out through a fully connected neural network and belongs to semantic segmentation. Mask-RCNN is different from other segmentation frameworks in that segmentation is performed after classification.
The integrity detection module of the embodiment has the following advantages:
(1) the performance and the efficiency far exceed those of the former Faster-RCNN;
(2) the training speed is abnormally rapid;
(3) ROIAligh perfectly solves the problems of ROIPooling precision and pixel loss in fast-RCNN.
Specifically, the correctness detection module 30 in the present embodiment may include: digital position detecting unit, hour minute hand point to detecting unit.
Wherein the digital position detection unit is configured to: respectively detecting whether the positions of 12 numbers in the image to be evaluated are accurate; the hour and minute hand pointing direction detection unit is configured to: and respectively detecting whether the directions of the hour hand and the minute hand in the image to be evaluated are accurate.
The principle of detecting the accuracy of the digital position in this embodiment is as follows:
because the bounding box (bounding box) and the target number are nearly tangent, the center of the bounding box can be considered as the specific coordinates of the number. Taking the center position of the bounding box corresponding to the 12 and the center of the circumscribed circle to form a vector x0(pointing from the center of the circumscribed circle to "12"). Similarly, the centers of the bounding boxes corresponding to the other 11 number categories can form a vector x with the center of the circumscribed circle1、x2、…、x11(ii) a . Sorting the centers of the bounding boxes of the 12 number categories, and then ideally dividing the x number into the x numberiAnd x0The sequence of angles found using the inverse trigonometric function should be 0, 30, 60, 90, 120, 150, 180, 150, 120, 90, 60, 30 degrees. Therefore, if we want to know whether the sequence of the 12 numbers bounding box is correct, we can first determine the included angle row of the seven numbers 12, 1Whether a column satisfies A [ i ]]＜A[i+1]And | A [ i +1]-A[i]I < 40.0 (wherein 40 is 30+10, 10 degrees is an error range), and then whether the angle arrangement of the following numbers (7, 8, …, 11) satisfies A [ i [ ] is judged]＞A[i+1]And | A [ i +1]-A[i]|＜＝40.0。
Specifically, the digital position detecting unit in this embodiment may include: the device comprises a vector drawing subunit, an included angle calculating subunit and a digital judgment subunit.
Wherein the vector rendering subunit is configured to: respectively drawing vectors x pointing to 12 numbers by taking the center of the minimum circumcircle as a starting point0、x1、x2、…、x11(ii) a The included angle calculation subunit is configured to: separately calculate the vector x0、x1、x2、…、x11And vector x0To obtain an included angle value A0、A1、A2、…、A11(ii) a The digital judgment subunit is configured to: according to the value of included angle A0、A1、A2、…、A6Sequentially judging A1、A2、…、A6Whether each included angle is larger than a previous included angle and the absolute value of the difference value between the included angle and the previous included angle meets a preset included angle range (such as 40 degrees); according to the value of included angle A7、A8、…、A11、A0Sequentially judging A7、A8、…、A11Whether each angle is greater than the next angle and the absolute value of the difference from the next angle satisfies a predetermined angle range (e.g., 40 degrees).
The principle of judging the pointing accuracy of the hour hand and the minute hand in the embodiment is as follows:
under the condition that the hour hand and the minute hand are both tangent by a bounding box, the minute hand and the hour hand can be easily obtained through coordinate information (a geometric length method is adopted: a simple length relation judgment is made). And respectively taking the central points of the two Bounding boxes to obtain a new coordinate of the circle center, and obtaining two vectors corresponding to the hour hand and the minute hand at the moment.
Example (c): the target time point of "10: 45" needs to be judged. The "target time point" referred to herein is a time designated in advance and pointed to in the drawn timekeeping chart by the test subject who is required to participate in the clock drawing test.
The vector (i.e. standard hour hand vector) formed by the circle center and the target "10" is approximately coincided with the hour hand vector within a certain error range, which indicates that the drawn hour hand points accurately. The '45 minutes' can be judged to correspond to the number '9' through simple calculation, and the minute hand vector (namely the standard minute hand vector) formed by the circle center and the '9' is approximately coincident with the minute hand vector within a certain error range, namely the drawn minute hand is indicated to be accurate in direction.
Specifically, the hour and minute hand pointing direction detecting unit in this embodiment may include: the pointer area subunit, the pointer vector drawing subunit and the pointer judgment subunit.
Wherein the pointer distinguishing subunit is configured to: distinguishing an hour hand from a minute hand according to the length relation of the two pointers in the image to be evaluated; the pointer vector rendering subunit is configured to: respectively drawing the vectors of an hour hand and a minute hand in the image to be evaluated and the vector of a standard hour hand and the vector of a standard minute hand pointing to a target time point; the pointer judging subunit is configured to: judging whether an included angle between the drawn hour hand vector and the standard hour hand vector meets a preset first error range or not; and judging whether an included angle between the drawn minute hand vector and the standard minute hand vector meets a preset second error range or not.
The correctness detection module of the embodiment has the following advantages:
(1) the coordinate information obtained in the integrity detection module is skillfully combined;
(2) the detection of the accuracy of the digital position, the pointing direction of the hour hand and the pointing direction of the minute hand is completed by utilizing the knowledge of the trigonometric function and the vector;
(3) a reliable error range is set, so that the confidence of an experimental result is higher;
(4) after a large amount of data training, the effect is obvious.
Further, the scoring module 40 in this embodiment may be specifically configured to: and scoring the clock drawn in the image to be evaluated according to a preset scoring standard and detection results of the roundness detection module, the integrity detection module and the correctness detection module.
The clock drawing test can be used for judging the senile dementia, and the freehand clock drawing is a complex behavior activity, needs a plurality of knowledge functions to participate besides the space construction skill, and relates to a plurality of cognitive functions such as memory, attention, abstract thinking, design, layout arrangement, application, numbers, calculation, time and space directional concepts, operation sequence and the like. The operation is simpler and time-saving, and the patient can accept the method more easily. There are various evaluation methods for the bell test, including a 3-point evaluation method, a 4-point evaluation method, a 5-point evaluation method, a 7-point evaluation method, a 10-point evaluation method, and a 30-point evaluation method. However, the 4-Point method is simple, sensitive and feasible, the diagnosis rate of dementia can reach 75 percent, and patients with dementia often cannot draw a dial of a watch completely.
In this embodiment, a 4-point rating method is adopted, and the preset rating standard is as follows:
(1) if the ratio of the area surrounded by the clock contour and the minimum circumcircle area meets the preset area ratio range, recording 1 point;
(2) if 12 numbers exist, marking 1 point;
(3) if the positions of the 12 numbers are accurate, 1 point is marked;
(4) if the pointer points to the correct point, the point is marked 1.
In an alternative embodiment, the roundness detection module 10 may further include: a sensitivity enhancing unit.
Wherein, the sensitivity improving unit is arranged between the preprocessing unit and the outline detecting unit and is configured to: the gray level image output by the preprocessing unit is subjected to sensitivity improving processing (such as Canny algorithm) and then input to the contour detection unit.
Fig. 2 is a schematic diagram of a second embodiment of the intelligent evaluation system based on a clock test according to the present invention. As shown in fig. 2, the intelligent evaluation system of the present embodiment includes an image acquisition module 50 and a display module 60, in addition to the roundness detection module 10, the integrity detection module 20, the correctness detection module 30 and the scoring module 40.
The configuration of the roundness detection module 10, the integrity detection module 20, the correctness detection module 30, and the scoring module 40 is the same as that in the first embodiment, and will not be described herein again; the image acquisition module 50 is configured to: collecting an image to be evaluated; the display module 60 is configured to: and displaying the scoring result.
For example, the image to be evaluated is captured by the image capturing module 50 (e.g., a digital camera) for detection by the roundness detection module 10, the integrity detection module 20, and the correctness detection module 30, and then scored by the scoring module 40 according to the detection result, the scored result may be displayed by the display module 60 (e.g., a liquid crystal display), and of course, the captured image to be evaluated may also be displayed as needed.
In the present application, the modules, units and sub-units of the intelligent evaluation system based on the drawing clock test are divided only for better understanding of the functions related to the technical solution of the present invention, and in practice, the functions corresponding to these modules may be loaded and executed by a single or multiple hardware.
Those of skill in the art will appreciate that the method steps of the examples described in connection with the embodiments disclosed herein may be embodied in electronic hardware, computer software, or combinations of both, and that the components and steps of the examples have been described above generally in terms of their functionality in order to clearly illustrate the interchangeability of electronic hardware and software. Whether such functionality is implemented as electronic hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
So far, the technical solutions of the present invention have been described in connection with the preferred embodiments shown in the drawings, but it is easily understood by those skilled in the art that the scope of the present invention is obviously not limited to these specific embodiments. Equivalent changes or substitutions of related technical features can be made by those skilled in the art without departing from the principle of the invention, and the technical scheme after the changes or substitutions can fall into the protection scope of the invention.
Claims (11)
1. An intelligent evaluation system based on a picture clock test, the system comprising: the device comprises a roundness detection module, an integrity detection module, a correctness detection module and a grading module;
the roundness detection module is configured to: detecting a clock contour in an image to be evaluated, drawing a minimum circumcircle of the clock contour, and calculating the ratio of the area enclosed by the clock contour to the area of the minimum circumcircle;
the integrity detection module is configured to: detecting whether the number, the hour hand and the minute hand exist in the image to be evaluated, and acquiring corresponding coordinate values;
the correctness detection module is configured to: respectively detecting whether the digital positions are correct and whether the directions of the hour hand and the minute hand are correct according to the coordinate values;
the scoring module is configured to: and scoring the clock drawn in the image to be evaluated according to the detection results of the roundness detection module, the integrity detection module and the correctness detection module.
2. The intelligent evaluation system based on a painting clock test of claim 1, wherein the roundness detection module comprises: the device comprises a preprocessing unit, a contour detection unit, an external circle drawing unit and an area rate calculation unit;
the pre-processing unit is configured to: converting the image to be evaluated into a gray scale image;
the contour detection unit is configured to: detecting the clock contour in the grey-scale map;
the circumscribed circle drawing unit is configured to: drawing a minimum circumscribed circle of the clock contour;
the area ratio calculation unit is configured to: and respectively calculating the area enclosed by the clock contour and the area of the minimum circumcircle, and solving the ratio of the area enclosed by the clock contour and the area of the minimum circumcircle.
3. The intelligent evaluation system based on a painting clock test of claim 2, wherein the area ratio calculation unit comprises: a first area calculation subunit, a second area calculation subunit, and a ratio calculation subunit;
the first area calculation subunit is configured to: calculating the number of pixels with the internal pixel value of 255 of the clock contour, and further calculating the area enclosed by the clock contour;
the second area calculation subunit is configured to: calculating the area of the minimum circumcircle;
the ratio calculation subunit is configured to: calculating the ratio of the area enclosed by the clock contour and the minimum circumcircle area.
4. The intelligent evaluation system based on painting clock test of claim 1, wherein the integrity detection module comprises: a convolutional neural network based on masked regions.
5. The intelligent evaluation system based on the drawing clock test of claim 1, wherein the correctness detection module comprises: the digital position detection unit and the hour-minute hand pointing detection unit;
the digital position detection unit is configured to: respectively detecting whether the positions of the 12 numbers in the image to be evaluated are accurate or not;
the hour and minute hand pointing detection unit is configured to: and respectively detecting whether the directions of the hour hand and the minute hand in the image to be evaluated are accurate.
6. The intelligent evaluation system based on a painting clock test of claim 5, wherein the digital position detection unit comprises: the device comprises a vector drawing subunit, an included angle calculating subunit and a digital judgment subunit;
the vector rendering subunit is configured to: respectively drawing vectors x pointing to 12 numbers by taking the circle center of the minimum circumcircle as a starting point0、x1、x2、…、x11；
The included angle calculating subunit is configured to: are respectively provided withCalculating the vector x0、x1、x2、…、x11And vector x0To obtain an included angle value A0、A1、A2、…、A11；
The digital judgment subunit is configured to: according to the value of included angle A0、A1、A2、…、A6Sequentially judging A1、A2、…、A6Whether each included angle is larger than a previous included angle and the absolute value of the difference value between each included angle and the previous included angle meets a preset included angle range; according to the value of included angle A7、A8、…、A11、A0Sequentially judging A7、A8、…、A11Whether each included angle is larger than the next included angle or not and the absolute value of the difference value between each included angle and the next included angle meets the preset included angle range.
7. The intelligent evaluation system based on a painting clock test of claim 5, wherein the hour and minute hand pointing direction detection unit comprises: the pointer area subunit, the pointer vector drawing subunit and the pointer judgment subunit are connected;
the pointer region molecular unit is configured to: according to the length relation of the two pointers in the image to be evaluated, an hour pointer and a minute pointer are distinguished;
the pointer vector drawing subunit is configured to: respectively drawing the vectors of an hour hand and a minute hand in the image to be evaluated and the vector of a standard hour hand and the vector of a standard minute hand pointing to a target time point;
the pointer judgment subunit is configured to: judging whether an included angle between the drawn hour hand vector and the standard hour hand vector meets a preset first error range or not; and judging whether an included angle between the drawn minute hand vector and the standard minute hand vector meets a preset second error range or not.
8. The intelligent rating system based on a painting clock test of claim 1, wherein the scoring module is specifically configured to: scoring the clock drawn in the image to be evaluated according to a preset scoring standard and detection results of the roundness detection module, the integrity detection module and the correctness detection module;
wherein the preset scoring standard is as follows:
if the ratio of the area surrounded by the clock contour to the minimum circumcircle area meets a preset area ratio range, recording 1 point;
if 12 numbers exist, marking 1 point;
if the positions of the 12 numbers are accurate, 1 point is marked;
if the pointer points to the correct point, the point is marked 1.
9. The intelligent evaluation system based on the draw-clock test according to claim 2 or 3, wherein the roundness detection module further comprises: a sensitivity-improving unit;
the sensitivity-improving unit is configured to: and after the processing of improving the sensitivity is carried out on the gray level map, the gray level map is input to the outline detection unit.
10. The intelligent rating system based on a painting clock test of any of claims 1-8, further comprising: an image acquisition module;
the image acquisition module is configured to: and collecting the image to be evaluated.
11. The intelligent rating system based on picture clock test of claim 10, further comprising: a display module;
the display module is configured to: and displaying the scoring result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911300342.4A CN111091910B (en) | 2019-12-17 | 2019-12-17 | Intelligent evaluation system based on painting clock test |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911300342.4A CN111091910B (en) | 2019-12-17 | 2019-12-17 | Intelligent evaluation system based on painting clock test |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111091910A true CN111091910A (en) | 2020-05-01 |
| CN111091910B CN111091910B (en) | 2023-11-03 |
Family
ID=70395641
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911300342.4A Active CN111091910B (en) | 2019-12-17 | 2019-12-17 | Intelligent evaluation system based on painting clock test |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111091910B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111739636A (en) * | 2020-06-19 | 2020-10-02 | 智恩陪心(北京)科技有限公司 | PPAT-based psychological intelligent analysis system |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1755754A (en) * | 2004-02-27 | 2006-04-05 | 佳能株式会社 | Image display apparatus |
| CN104715157A (en) * | 2015-03-25 | 2015-06-17 | 成都信息工程学院 | Cognition impairment evaluating system and method based on clock drawing test |
| 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 |
| 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 |
| CN110859599A (en) * | 2019-11-26 | 2020-03-06 | 复旦大学 | Cerebrovascular disease nerve injury patient cognitive function automatic screening system |
| RU2745282C1 (en) * | 2020-06-02 | 2021-03-23 | Государственное бюджетное учреждение здравоохранения Московской области "Московский областной научно-исследовательский клинический институт им. М.Ф. Владимирского" (ГБУЗ МО МОНИКИ им. М.Ф. Владимирского) | Method for diagnosing severity of vascular cognitive impairment |
| CA3103781A1 (en) * | 2019-12-23 | 2021-06-23 | Peter Anthony Hall | Method and system for assessing cognitive function of an individual |
| CN113795815A (en) * | 2019-05-06 | 2021-12-14 | 苹果公司 | Clock face for electronic equipment |
-
2019
- 2019-12-17 CN CN201911300342.4A patent/CN111091910B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1755754A (en) * | 2004-02-27 | 2006-04-05 | 佳能株式会社 | Image display apparatus |
| CN104715157A (en) * | 2015-03-25 | 2015-06-17 | 成都信息工程学院 | Cognition impairment evaluating system and method based on clock drawing test |
| 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 |
| 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 |
| CN113795815A (en) * | 2019-05-06 | 2021-12-14 | 苹果公司 | Clock face for electronic equipment |
| 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 |
| CN110859599A (en) * | 2019-11-26 | 2020-03-06 | 复旦大学 | Cerebrovascular disease nerve injury patient cognitive function automatic screening system |
| CA3103781A1 (en) * | 2019-12-23 | 2021-06-23 | Peter Anthony Hall | Method and system for assessing cognitive function of an individual |
| RU2745282C1 (en) * | 2020-06-02 | 2021-03-23 | Государственное бюджетное учреждение здравоохранения Московской области "Московский областной научно-исследовательский клинический институт им. М.Ф. Владимирского" (ГБУЗ МО МОНИКИ им. М.Ф. Владимирского) | Method for diagnosing severity of vascular cognitive impairment |
Non-Patent Citations (3)
| Title |
|---|
| 杨倩: "老年2型糖尿病患者血清羧化不全骨钙素水平与认知功能的相关性研究", 《中国优秀硕士学位论文全文数据库医药卫生科技辑》, no. 9, pages 065 - 51 * |
| 袁书伟;钟传杰;朱兆伟;: "多主干鱼骨型时钟树结构的设计方法及优化", 微处理机, no. 04 * |
| 黄若燕;唐牟尼;佘生林;孙彬;林康广;郁俊昌;陈映梅;郭伟坚;肖;王怀坤;: "画钟测验在认知障碍老人中的鉴别作用", 中国神经精神疾病杂志, no. 08 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111739636A (en) * | 2020-06-19 | 2020-10-02 | 智恩陪心(北京)科技有限公司 | PPAT-based psychological intelligent analysis system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111091910B (en) | 2023-11-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105260709B (en) | Water meter calibration method, apparatus based on image procossing and system | |
| CN107229930B (en) | Intelligent identification method for numerical value of pointer instrument | |
| CN105260710B (en) | Water meter calibration method, apparatus based on image procossing and system | |
| CN104363815B (en) | Image processing apparatus and image processing method | |
| CN105300482B (en) | Water meter calibration method, apparatus based on image procossing and system | |
| WO2020103417A1 (en) | Bmi evaluation method and device, and computer readable storage medium | |
| CN109543682A (en) | A kind of readings of pointer type meters method based on deep learning | |
| CN106529559A (en) | Pointer-type circular multi-dashboard real-time reading identification method | |
| CN108509908B (en) | Pupil diameter real-time measurement method based on binocular stereo vision | |
| CN112818988A (en) | Automatic reading identification method and system for pointer instrument | |
| CN107066998A (en) | A kind of pointer-type circular single instrument board real-time identification method of utilization mobile device | |
| CN102499762B (en) | Three-dimensional spatial positioning system for medical ultrasonic probe relative to part to be checked and method | |
| CN108197564B (en) | Evaluation system and method for picture clock test | |
| CN106530271A (en) | Infrared image significance detection method | |
| CN104732520A (en) | Cardio-thoracic ratio measuring algorithm and system for chest digital image | |
| CN112418251B (en) | Infrared body temperature detection method and system | |
| CN108645490A (en) | Cold-water meter vision detection system based on image processing techniques and detection method | |
| CN111091910A (en) | Intelligent evaluation system based on drawing clock test | |
| CN104851102B (en) | A kind of infrared small target detection method based on human visual system | |
| CN107832695A (en) | The optic disk recognition methods based on textural characteristics and device in retinal images | |
| CN112924967A (en) | Remote sensing monitoring method for crop lodging based on radar and optical data combination characteristics and application | |
| JP2023549865A (en) | Method and system for measuring binocular distance for children | |
| CN110702041A (en) | Method and device for measuring area of rural land homestead and storage medium | |
| CN113469942B (en) | CT image lesion detection method | |
| CN111428577B (en) | Face living body judgment method based on deep learning and video amplification technology |
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 |