Quantitative analysis and evaluation method and system for spasmodic torticollis

Abstract

The invention discloses a spasmodic torticollis quantitative analysis and evaluation method and system. The method comprises the following steps: step S1, collecting demographic information of a patient with primary cervical myodynia in an outpatient service and a ward; s2, two clinicians carry out clinical evaluation on the spasmodic torticollis of the patient; s3, recording the video according to the specification; s4, designing features through a face key point recognition algorithm, carrying out feature extraction, recognizing key anatomical points, and constructing auxiliary lines; s5, calculating angles and time through visual identification, and defining various scale indexes and scoring rules; and S6, outputting a score and a quantitative score result, and comparing the score and the score with a clinical expert score for verification. By means of the method, clinical cardiotoxin injection doctors can be assisted in accurately judging motion types, accurately positioning muscles and recommending specific injection doses, and therefore the disease curative effect and the clinical efficiency of the doctors are improved.

Description

Technical Field

[0001] The present invention relates to the medical field, and in particular to a method and system for quantitative analysis and evaluation of spasmodic torticollis. Background Art

[0002] Dystonia is a type of movement disorder characterized by abnormal, repetitive movements or postures caused by sustained or intermittent abnormal muscle contractions. Primary cervical dystonia is the most common focal dystonia, a chronic movement disorder characterized by abnormal postures or involuntary movements of the head, neck, and shoulders. The movement pattern is patterned and may be accompanied by tremors and pain. It is often induced or aggravated by voluntary movements, accompanied by abnormal muscle activation. Treatment methods include drug therapy, botulinum toxin injection therapy, and surgical intervention. Among them, botulinum toxin injection therapy can effectively slow down abnormal muscle contractions and pain, and is recommended as the first-line treatment for cervical dystonia. However, in actual clinical practice, the efficacy of botulinum toxin injection therapy is only 47-83%. The main reasons for treatment failure include inaccurate judgment of the target muscle, incorrect identification of the target muscle, insufficient injection dose, and the production of botulinum toxin antibodies.

[0003] Therefore, how to distinguish the basic types of primary cervical dystonia as accurately as possible, precisely locate the target muscles and give accurate injection doses is of great significance for the treatment of this type of disease.

[0004] The cervical muscles are divided into three layers according to the depth of the layer. Some muscles start and end at the head-neck complex, some muscles start and end at the C2 vertebra and below, and some muscles span the entire skull and head and neck. The anatomical structure and responsible movement type are complex. The anatomical and functional basis of cervical dystonia includes the C2-7 vertebrae and the head-neck complex (skull, C1 and C2 and the two joints formed), and the "head-neck" concept was proposed by the Reichel research team: cervical dystonia is divided into rotation, lateralization, flexion, and extension involving the head and neck, and the target muscles are selected. At present, the identification of abnormally activated muscles in cervical dystonia is mainly based on the anatomical positioning of the operator on the spot, and for deep muscles, it is assisted by portable electromyography (EMG) and musculoskeletal ultrasound technology. However, the current systematic review of botulinum toxin injection for the treatment of cervical dystonia also mentioned that EMG-guided injection has not significantly improved the efficacy due to technological innovation, and there is currently a lack of a unified injection scheme for specific movement subtypes (including target muscles and specific doses). Other imaging techniques include using PET / CT or SPECT to identify muscles with diffusely increased metabolism before injection, but due to the high economic cost of nuclear medicine imaging and patients' concerns about radiological examinations, such methods are not widely used. With the increase of research data and the sharing and opening of relevant databases, and the maturity and improvement of anatomical theories and imaging foundations, prediction and platform construction based on computer language have become possible.

[0005] Movement disorders are a type of disease with characteristic movement patterns, and their movement representation can be identified, classified, and constructed through motion sensors or sites in videos. Machine algorithms were first applied to Parkinson's disease, which has a high incidence rate. They mainly include motion sensor recognition and classification training and video recognition and classification training. In clinical practice, they are used to quantitatively determine the severity of bradykinesia and assist in differential diagnosis with essential tremor. In the field of dystonia, some scholars have initially trained algorithms with the help of musculoskeletal ultrasound images of patients and controls and dynamic image data of three-axis whole-body movement patterns, and can specifically distinguish the neck muscle contours of patients with cervical dystonia from the normal population.

[0006] In summary, although machine algorithms have achieved some results in assisting in disease severity determination and differential diagnosis in the field of movement disorders in recent years, there are no results in building relevant prediction platforms in the field of assisting treatment, especially in cervical dystonia. This field is a blind spot and breakthrough point. Summary of the invention

[0007] In order to overcome the above technical defects, the purpose of the present invention is to construct an online prediction platform to assist clinical botulinum toxin injection doctors to accurately determine movement typing, precisely locate muscles and recommend specific injection doses, thereby improving disease efficacy and clinical efficiency of doctors.

[0008] To this end, one aspect of the present invention provides a method for quantitative analysis and evaluation of spasmodic torticollis, comprising the following steps:

[0009] Step S1, collecting demographic information of patients with primary cervical dystonia in outpatient clinics and wards;

[0010] Step S2, a clinician performs a clinical assessment of the patient's spasmodic torticollis;

[0011] Step S3, recording the video according to regulations;

[0012] Step S4, designing features and performing feature extraction through a facial key point recognition algorithm, identifying key anatomical points and constructing auxiliary lines;

[0013] Step S5, calculating the angle and time through visual recognition, and defining various scale indicators and scoring rules;

[0014] Step S6, output the scoring and quantitative scoring results, and compare them with the clinical expert scores for verification.

[0015] Furthermore, the clinical assessment in step S2 is based on the TWSTRS (Toronto Western Spasmodic Torticollis Rating Scale).

[0016] Furthermore, the shooting content in step S3 includes: basic information of the patient, natural stepping of the patient, correction of torticollis posture, head tilt to the left, head tilt to the right, head tilt forward, and head tilt backward.

[0017] Furthermore, the calculation contents in step S5 include: maximum deviation angle, range of motion, shoulder lift, duration factor, and time.

[0018] Further, the step of calculating the duration factor includes:

[0019] Step S401, selecting the most serious item from the score range of rotation, lateral flexion, cervical flexion, and cervical tilt in the maximum deviation angle as the reference tracking item in the duration factor;

[0020] Step S402, taking the frame number range of the video frame number*5% as the maximum angle calculation range, taking the median as the maximum angle of the current video, and frames greater than or equal to the angle are regarded as the onset of the maximum angle.

[0021] Another aspect of the present invention provides a spasmodic torticollis quantitative analysis and evaluation system, comprising an acquisition module, an evaluation module, a recording module, a recognition module, a calculation module, and an output module, wherein:

[0022] The collection module is used to collect demographic information of patients with primary cervical dystonia in the outpatient clinic and ward;

[0023] Assessment module, 2 clinicians performed clinical assessment of the patient's spasmodic torticollis;

[0024] A recording module, used to record videos according to regulations;

[0025] The recognition module is used to design features and extract features through the facial key point recognition algorithm, identify key anatomical points and construct auxiliary lines;

[0026] The calculation module is used to calculate the angle and time through visual recognition and define various scale indicators and scoring rules;

[0027] The output module is used to output the scoring and quantitative scoring results, which are compared with the scores of clinical experts for verification.

[0028] Furthermore, the clinical assessment in the assessment module is based on the TWSTRS scale.

[0029] Furthermore, the recording module includes basic information of the patient, natural steps of the patient, correction of torticollis posture, head tilt to the left, head tilt to the right, head tilt forward, and head tilt backward.

[0030] Furthermore, the calculation contents in the calculation module include: maximum deviation angle, range of motion, shoulder lifting, duration factor, and time.

[0031] Further, the step of calculating the duration factor includes:

[0032] Step S401, selecting the most serious item from the score range of rotation, lateral flexion, cervical flexion, and cervical tilt in the maximum deviation angle as the reference tracking item in the duration factor;

[0033] Step S402, taking the frame number range of the video frame number*5% as the maximum angle calculation range, taking the median as the maximum angle of the current video, and frames greater than or equal to the angle are regarded as the onset of the maximum angle.

[0034] Compared with the prior art, the above technical solution has the following beneficial effects:

[0035] The present invention constructs an online spasmodic torticollis prediction platform through quantitative analysis methods combined with advanced image recognition technology, which has many technical advantages and practical application value.

[0036] First, the platform uses facial key point recognition algorithms and visual recognition technology to accurately identify the patient's key anatomical points and movement characteristics, and accurately quantify the severity and specific manifestations of spasmodic torticollis. This high-precision data collection and analysis can effectively avoid subjective bias in manual evaluation, thereby improving the reliability and accuracy of diagnosis.

[0037] Secondly, the system provided by the present invention can help clinicians better determine the location and dosage of botulinum toxin injection, thereby optimizing the treatment effect. Through accurate movement typing judgment and muscle positioning, doctors can more effectively select the appropriate injection site and reduce side effects and adverse reactions caused by improper injection. In addition, the system can also recommend specific dosages based on the individual conditions of the patient, making the treatment more personalized and targeted, improving the efficacy while reducing drug waste.

[0038] In addition, the online nature of the platform enables real-time application in various clinical settings, making it easier for doctors to assess and intervene in patients in outpatient clinics and wards. This online assessment system can not only improve doctors' work efficiency and reduce duplication of work, but also provide a standardized assessment process for clinical work, ensuring consistency of assessment results between different doctors.

[0039] Finally, by comparing and verifying with the scoring results of clinical experts, the credibility and scientificity of the system scoring were further proved. This data-driven evaluation and decision support system can greatly improve the treatment effect of spasmodic torticollis, shorten the treatment cycle of patients, and improve the quality of life of patients. At the same time, this system also has strong scalability and can be applied to the evaluation and treatment of other movement disorders, and has broad clinical application prospects.

[0040] In summary, the online prediction platform of the present invention combines multiple technical means such as image recognition, quantitative analysis and personalized treatment recommendations. It is not only significantly superior to existing manual evaluation methods in terms of accuracy and reliability, but also provides strong support for improving the clinical efficiency of doctors and the treatment effects of patients. It has extremely high clinical application value and promotion potential. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 Flow chart of the quantitative analysis and evaluation method for spasmodic torticollis.

[0042] Figure 2 Schematic diagram of the logic of the method for the quantitative analysis and evaluation of spasmodic torticollis.

[0043] Figure 3 Comparison of visual recognition and clinician scoring in a multicenter validation. DETAILED DESCRIPTION

[0044] The advantages of the present invention are further described below in conjunction with the accompanying drawings and specific embodiments. Those skilled in the art should understand that the following specific description is illustrative rather than restrictive, and should not be used to limit the scope of protection of the present invention.

[0045] like Figure 1 and Figure 2 Shown is a flow chart and logic diagram of the quantitative analysis and evaluation method for spasmodic torticollis, which includes the following steps:

[0046] Step S1, demographic information collection, collecting basic information of patients with primary cervical dystonia in outpatient clinics and wards.

[0047] Step S2, clinical assessment, 2 clinicians performed a clinical assessment of the patient's spasmodic torticollis (based on the TWSTRS scale).

[0048] Step S3, video recording.

[0049] The following basic requirements must be met when shooting videos:

[0050] 1. At least the upper body (navel and above) should be in the frame;

[0051] 2. Let your hands hang naturally or complete the prescribed movements and must not be brought close to the head, face, or neck without permission.

[0052] The video content is as follows:

[0053] 1. Name, age, current date, duration of illness;

[0054] 2. Step naturally without correcting the torticollis posture, 60 seconds from the front and 60 seconds from the side;

[0055] 3. Correct torticollis posture, try to maintain a neutral posture, 60 seconds from the front and 60 seconds from the side;

[0056] 4. Tilt your head left or right (keep your ear as close to your shoulder as possible); tilt your head left or right; shoot from the front, and maintain each action for 15 seconds.

[0057] 5. Tilt your head forward and backward. Shoot each movement from the front for 15 seconds and from the side for 15 seconds.

[0058] Step S4, identify key anatomical points and construct auxiliary lines for the actions in the video. Design features and extract features through facial key point recognition algorithm, including the facial features plane, eyes, ears, nose, and mouth.

[0059] The key anatomical points include the highest point of the shoulders, the midpoint of the neck, the center of the eyebrows, the acromion, etc.

[0060] For the following actions, construct auxiliary lines as follows:

[0061] Torsion: The angle between the plane of facial features in the frontal image and the horizontal plane;

[0062] Lateral flexion: the angle between the midpoint of the line between the two ears and the line between the two shoulders from the front view;

[0063] Anterior inclination and posterior flexion: the angle between the ear-shoulder line and the vertical line;

[0064] Lateral deviation: the angle between the head vertical line and the vertical line in the frontal image + the angle between the neck vertical line and the vertical line;

[0065] Sagittal deviation: the angle between the vertical line of the head and the vertical line + the angle between the vertical line of the cervical segment and the vertical line;

[0066] Shoulder lift: From the front, look at the angle between the line connecting the two acromion peaks and the horizontal line.

[0067] Step S5, calculate the angle and time through visual recognition, and define various scale indicators and scoring criteria based on the selected features. Including the following items:

[0068] A. The scoring rules for the maximum deviation angle are as follows (the number before the “:” in the following items is the score, and the number after the “:” is the deviation angle range):

[0069] 1. Rotation: 0:(0,5], 1:(5,22.5], 2:(22.5,45], 3:(45,67.5], 4:(67.5,90]

[0070] 2. Lateral flexion: 0:(0,5], 1:(5,15.5], 2:(15.5,35], 3:(35,90]

[0071] 3. Neck flexion or posterior tilt: 0:(0,10], 1:(10,36], 2:(36,63], 3:(63,90]

[0072] 4. Lateral offset: 0:(0,8), 1:[8,90]

[0073] 5. Sagittal shift: 0:(0,45), 1:[45,90]

[0074] B. To specify the scoring rules for duration factors, you need to follow the following steps:

[0075] Step 1: Select the most severe item from the score range of rotation, lateral flexion, cervical flexion, and cervical tilt in the maximum deviation angle as the benchmark tracking item in the duration factor;

[0076] Step 2: Use the frame number range of the video frame number * 5% as the maximum angle calculation range, and take the median as the maximum angle of this video instead of directly using the maximum angle in Project A. Frames greater than or equal to this angle are considered to be the onset of the maximum angle. The number of frames in a video represents the number of images recorded per second in the video. For example, if a video has 1,000 frames, then 5% of the frames are 50 frames. These 50 frames represent a calculation range. Within this 5% frame range, the system will find the offset angle of the patient's body or head in each frame, and select the median of these angles as the "maximum angle". In order to reduce the impact of extreme values ​​on the average and obtain more stable and representative results, using the median can better represent the actual state of the patient in the video, rather than being disturbed by the extreme angles of individual frames.

[0077] Note: Although only one item is tracked, the angle ranges of the other items must also be listed; if two scores are the same in the first step, calculate them separately and take the most serious result.

[0078] The following are the scoring rules:

[0079] 0: All angles are less than the specified angles (5 degrees for rotation and 10 degrees for cervical flexion / posterior tilt);

[0080] 1: ≤25% of the time is smaller than the maximum angle;

[0081] 2: >25% of the time, the angle is smaller than the maximum angle, and ≤25% of the time, the angle is ≥ the maximum angle;

[0082] 3:>25% and ≤50% of the time is greater than or equal to the maximum angle;

[0083] 4:>50% and ≤75% of the time is greater than or equal to the maximum angle;

[0084] 5:>75% of the time is greater than or equal to the maximum angle.

[0085] C. Integral rules for shoulder lifting (the following items are preceded by the “:” for duration and followed by the “:” for deviation angle);

[0086] None: (0,5], <1 / 4 range: (5,10], 1 / 4-3 / 4 range: (10,20], >3 / 4 range: (20,30];

[0087] The corresponding score is:

[0088] 0: The whole process is in (0,5]

[0089] 1: appeared or has been in (5,10]

[0090] 2: Appeared or always in (10,20], appeared, i.e. ≤ 75% of the time in (20,30]

[0091] 3: >75% of the time in (20,30]

[0092] The scores are evaluated in order from high to low, that is, 3 points, 2 points, 1 point, and 0 points are used for filtering. If both 1 point: ≤ 75% of the time in (10, 20] and 2 points: ≤ 75% of the time in (20, 30] are met, the higher score of 2 points is taken.

[0093] D. Scoring rules for range of motion:

[0094] Condition 1: Able to move to the midline position: ≤5 degrees for both left and right, ≤10 degrees for both front and back;

[0095] Condition 2: Ability to move to opposite positions: The maximum angle difference between left and right lateral flexion is ≤5 degrees, the maximum angle difference between left and right rotation is ≤5 degrees, and the maximum angle difference between forward and backward flexion is ≤10 degrees.

[0096] (1) If conditions 1 and 2 are both met, the score is 0;

[0097] (2) Condition 1 can be met, but condition 2 cannot be met:

[0098] The maximum sum of the angles of rotation on both sides is greater than 95 degrees, the lateral flexion is greater than 50 degrees, and the anterior and posterior flexion is greater than 55 degrees, which is 1 point;

[0099] 2 points if the sum of the maximum angles of rotation on both sides is ≤95° or lateral flexion is ≤50° or anterior and posterior flexion is ≤55°;

[0100] (3) Neither conditions 1 nor 2 can be met:

[0101] 3 points if one side has rotation energy ≥85 (90-5), lateral flexion energy ≥40 (45-5), and anterior and posterior flexion energy ≥35 (45-10);

[0102] If both sides cannot achieve rotation ≥85 (90-5) or lateral flexion ≥40 (45-5) or anterior and posterior flexion ≥35 (45-10), the score is 4 points;

[0103] If there is no angle in the opposite direction within the range of motion, for example, all angles are in the positive direction and there is no negative direction, then it is determined that condition 1 is not met.

[0104] E. Time scoring rules:

[0105] Determine the first second in which the deviation from 10 degrees occurs, and then the corresponding score belongs to that second (the score is before the “:” and the standard is after the “:”):

[0106] 0: ≧60 seconds;

[0107] 1: = 45.5-60 seconds;

[0108] 2: = 30.5-45.5 seconds;

[0109] 3: = 15.5-30.5 seconds;

[0110] 4: ≦15.5 seconds;

[0111] Among them, for lateral flexion, cervical flexion and posterior cervical tilt, the calculation method for rotation ≥ 1 minute (> 5 degrees) is defined and judged frame by frame.

[0112] Step S6: output the scoring and quantitative scoring results, and perform verification.

[0113] First verification: The model was used to diagnose 110 patients with spasmodic torticollis from Ruijin Hospital affiliated to Shanghai Jiao Tong University School of Medicine. The results showed that the accuracy of visual recognition scores for each item of TWSTRS was compared with the scores of clinical experts: A1: 95.3%, A2: 100%, A3a+b: 92.2%, A4: 91%, A5: 96.7%, B: 86.7%, C: 94.3%, D: 72%, E: 83%.

[0114] Multicenter validation: The model was validated using videos collected from 51 patients with spasmodic torticollis from Ruijin Hospital affiliated to Shanghai Jiao Tong University School of Medicine, 50 patients from Tongji Hospital, and 20 patients from Wuhan. Figure 3 The diagnostic accuracy is obtained by visual recognition of videos with different acquisition quality in each center. Except that the accuracy of some scores (such as range of motion, E_Range_of motion) is slightly lower due to video quality, the accuracy of other items is ideal.

[0115] It should be noted that the embodiments of the present invention have better practicability and do not impose any form of limitation on the present invention. Any technician familiar with the field may use the technical content disclosed above to change or modify it into an equivalent effective embodiment. However, any modification or equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A method for quantitative analysis and evaluation of spasmodic torticollis, characterized in that: The following steps are involved: Step S1, collecting demographic information of patients with primary cervical dystonia in outpatient clinics and wards; Step S2, a clinician performs a clinical assessment of the patient's spasmodic torticollis; Step S3, recording the video according to regulations; Step S4, designing features and performing feature extraction through a facial key point recognition algorithm, identifying key anatomical points and constructing auxiliary lines; Step S5, calculating the angle and time through visual recognition, and defining various scale indicators and scoring rules; Step S6, output the scoring and quantitative scoring results, and compare them with the clinical expert scores for verification.

2. The method for quantitative analysis and evaluation of spasmodic torticollis according to claim 1, characterized in that: The clinical assessment in step S2 is based on the TWSTRS scale.

3. The method for quantitative analysis and evaluation of spasmodic torticollis according to claim 1, characterized in that ,The shooting content in step S3 includes: basic information of the patient, the patient’s natural stepping, correcting torticollis posture, head tilting to the left, head tilting to the right, head tilting forward, and head tilting backward.

4. The method for quantitative analysis and evaluation of spasmodic torticollis according to claim 1, characterized in that ,The calculation contents in step S5 include: maximum deviation angle, ,range of motion, shoulder lifting, duration factor, and time.

5. The method for quantitative analysis and evaluation of spasmodic torticollis according to claim 4, characterized in that ,The steps of calculating the duration factor include: Step S501, selecting the most serious item from the score range of rotation, lateral flexion, cervical flexion, and cervical tilt in the maximum deviation angle as the reference tracking item in the duration factor; Step S502, taking the frame number range of the video frame number*5% as the maximum angle calculation range, taking the median as the maximum angle of the current video, and frames greater than or equal to the angle are regarded as the onset of the maximum angle.

6. A quantitative analysis and evaluation system for spasmodic torticollis, characterized in that: It includes acquisition module, evaluation module, recording module, recognition module, calculation module and output module, among which: The collection module is used to collect demographic information of patients with primary cervical dystonia in the outpatient clinic and ward; Assessment module, where clinicians conduct a clinical assessment of the patient’s spasmodic torticollis; A recording module, used to record videos according to regulations; The recognition module is used to design features and extract features through the facial key point recognition algorithm, identify key anatomical points and construct auxiliary lines; The calculation module is used to calculate the angle and time through visual recognition and define various scale indicators and scoring rules; The output module is used to output the scoring and quantitative scoring results, which are compared with the scores of clinical experts for verification.

7. The spasmodic torticollis quantitative analysis and evaluation system according to claim 6, characterized in that: The clinical assessment in the assessment module is based on the TWSTRS scale.

8. The spasmodic torticollis quantitative analysis and evaluation system according to claim 6, characterized in that ,The shooting content in the recording module includes the ,basic information of the patient, the patient stepping naturally, the ,correction of torticollis posture, head tilting to the left, head tilting to the right, head tilting forward, and head tilting backward.

9. The spasmodic torticollis quantitative analysis and evaluation system according to claim 6, characterized in that ,The calculation contents in the calculation module include: maximum deviation angle, ,range of motion, shoulder lift, duration factor, and time.

10. The spasmodic torticollis quantitative analysis and evaluation system according to claim 9, characterized in that ,The steps of calculating the duration factor include: Step S501, selecting the most serious item from the score range of rotation, lateral flexion, cervical flexion, and cervical tilt in the maximum deviation angle as the reference tracking item in the duration factor; Step S502, taking the frame number range of video frame number*5% as the maximum angle calculation range, taking the median as the maximum angle of the current video, and frames greater than or equal to the angle are regarded as the onset of the maximum angle.

Images