EP4266981A1 - Procédé standardisé de détermination d'un indice d'apnées+hypopnées ou d'une labélisation fonction de cet indice - Google Patents
Procédé standardisé de détermination d'un indice d'apnées+hypopnées ou d'une labélisation fonction de cet indiceInfo
- Publication number
- EP4266981A1 EP4266981A1 EP21836200.2A EP21836200A EP4266981A1 EP 4266981 A1 EP4266981 A1 EP 4266981A1 EP 21836200 A EP21836200 A EP 21836200A EP 4266981 A1 EP4266981 A1 EP 4266981A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- patient
- data
- apnea
- morphology
- maxillofacial
- 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.)
- Withdrawn
Links
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/48—Other medical applications
- A61B5/4806—Sleep evaluation
- A61B5/4818—Sleep apnoea
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Measuring devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/107—Measuring physical dimensions, e.g. size of the entire body or parts thereof
- A61B5/1079—Measuring physical dimensions, e.g. size of the entire body or parts thereof using optical or photographic means
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
-
- 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
- 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
- G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
Definitions
- the present invention relates to the field of the diagnosis of pathologies from maxillofacial morphological characteristics with the assistance of automatic learning of image analysis. Its application is particularly advantageous in the diagnosis of obstructive sleep apnea syndrome or of pathologies for which a sleep apnea syndrome is very frequently associated.
- a form of sleep apnea syndrome is linked to airway obstruction. When this obstruction is partial, it is called hypopnea. This obstruction can be complete, we then speak of obstructive apnea.
- Obstructive Sleep Apnea-Hypopnea Syndrome is a frequent pathology in the population.
- the literature on this subject reports a prevalence of 4% in men and 2% in women (Young et al. The occurrence of sleep-disordered breathing among middle-aged adults, N Engl J Med, 328, 1230-1235, 1993), more particularly in adults over 50 (Lévy, et al. Obstructive sleep apnoea syndrome, Nature Reviews Disease Primers, 1, 15015, 2015). II corresponds to repeated obstructions of the upper airways during sleep.
- OSAHS is associated with sleep fragmentation likely to promote daytime sleepiness, attention disorders and increase the risk of accidents (Myers et al.
- this syndrome is usually treated using a nighttime device that maintains continuous positive airway pressure that limits obstruction and stabilizes the upper airway.
- OSAHS screening in populations at risk, representing millions of subjects, such as people with hypertension, heart failure, or type 2 diabetes.
- the Berlin questionnaire is generally used by specialist practitioners, in order to establish a rapid diagnosis which, combined with the specialist's experience, will or will not direct the patient to further examinations for the diagnosis of OSAS.
- the Berlin questionnaire makes an inventory of the patient's situation with regard to snoring, drowsiness, and hypertension, then translated into a score allowing the orientation or not of the patient for an additional examination.
- the complementary examination consists of recording respiratory events during the patient's sleep.
- the reference examination is sleep polysomnography. This examination is carried out in a hospital setting. It quantifies abnormal respiratory events (apnea, hypopnea) during the night, to obtain an apnea+hypopnea index, which can also be referred to in the field as the apnea-hypopnea index, or the acronym AHI.
- Polysomnography includes an electroencephalogram, to assess sleep pattern, and a recording of cardiorespiratory events (nasal airflow, blood oxygen saturation, heart rate, chest and abdominal expansion).
- outpatient polygraphy is more accessible because it does not require hospitalization.
- a device records overnight airflow, blood oxygen saturation, respiratory events, snoring and sleep position to obtain the apnea + hypopnea index.
- An object of the present invention is therefore to propose making reliable and/or standardizing, by the analysis of morphological characteristics of an upper part of the human body, the diagnosis of a pathology, and more particularly the diagnosis of the syndrome of obstructive sleep apnea-hypopnea.
- a method for determining an apnea+hypopnea index of a patient or labeling according to an apnea+hypopnea index comprising:
- the set of data relating to the patient comprises, or makes it possible to determine, data characteristic of the at least maxillofacial morphology of the patient with respect to a reference morphology.
- Said characteristic data are a function of the positioning of at least four homologous points, on a 3D scan of a portion of the patient's head representing his at least maxillofacial morphology.
- the determination of said characteristic data depends on the positioning of at least four homologous points, on a 3D scan of a portion of the patient's head representing his at least maxillofacial morphology, - an introduction of the set of data relating to the patient, comprising the characteristic data of the patient's at least maxillofacial morphology, in an automatic learning model trained to predict an apnea + hypopnea index, or a function labeling an apnea+hypopnea index, for said set of data, from a plurality of sets of data from a database relating to a set of distinct patients, each set of data from the database : o comprising at least maxillofacial morphology data relating to a patient of the set, and o being associated with an apnea + hypopnea index, or a labeling function of an apnea + hypopnea index, so that the learning model predicts an apnea+hypo
- the morphological characteristics of the patient are compared in an automated way with the database, to obtain by comparison the apnea + hypopnea index or the labeling function of this index.
- the homologous points make it possible to make the alignment between the characteristic data of the patient's morphology and those of all the patients in the database more reliable, and thus to make the prediction more reliable.
- This method therefore makes it possible to dispense with the observation of these characteristics by a specialist practitioner, and the variability that may arise from this observation, for example from one practitioner to another.
- the characteristic data of the patient's maxillofacial morphology being taken from a 3D scan comprising a large amount of information, positioning disparities can lead to great variability in the results and therefore limit the reliability of the process.
- the reference morphology corresponds to that of a reference individual and makes it possible to constitute a reference grid to be applied to the 3D scan of the patient in order to limit the irrelevant variability of the data.
- the 3D scan is thus processed in such a way as to align the homologous points of the patient and those of the grid.
- the characteristic data of the patient from the positioning of the homologous points in relation to this reference morphology are therefore obtained by minimizing the deviations unrelated to the apnea-hypopnea syndrome. The process is therefore made more reliable.
- Obtaining the apnea+hypopnea index, or the labeling according to this index, does not require, for the patient to be diagnosed, the recording of the respiratory events during sleep.
- the realization of a polysomnography or a polygraphy is avoided for the patient to be diagnosed, with the constraints that these cause, such as difficulties in falling asleep, multiple displacements.
- the interpretation of polysomnography and polygraphy results are complex and must be carried out by a specialist practitioner, to arrive at the apnea+hypopnea index.
- the method therefore makes it possible to standardize and facilitate, from the morphological characteristics of a patient, the obtaining of the apnea + hypopnea index, or the labeling according to this index, and therefore the diagnosis of a pathology. linked to the apnea+hypopnea index, for example the diagnosis of OSAHS. Comfort and ease of access to care for the patient are also improved compared to existing solutions. Obtaining the apnea+hypopnea index, or the labeling based on this index, is simplified and accelerated compared to existing solutions, since this index is obtained from a 3D scan of the patient.
- the method can be implemented by a practitioner who may or may not be a specialist in pathology, or by any operator such as a healthcare professional, for example a pharmacist. Its result can also be returned quickly, for example from the consultation, obtaining the apnea + hypopnea index or the labeling being done by the automatic learning model.
- the labeling is more particularly a value, qualitative or quantitative, of the risk that the patient presents the pathology.
- the labeling can have at least two values, a value associated with a low risk and at least one value associated with a proven risk.
- the labeling presents, among the at least one value associated with a proven risk, at least one value associated with a moderate proven risk of presenting the pathology, and at least one value associated with a severe proven risk of presenting the pathology .
- the labeling is a function of the apnea-hypopnea index, and in particular a function of the standard threshold value of the apnea+hypopnea index for a pathology, for example OSAS.
- the labeling presents:
- the method can comprise the diagnosis of a pathology of the patient, for example OSAHS.
- the method according to the first aspect can be a pathology diagnostic method, and has the advantages described above. The method thus makes it possible to make reliable and to standardize, by the analysis of morphological characteristics of an upper part of the human body, the diagnosis of a pathology, and more particularly the diagnosis of the syndrome of obstructive sleep apnea-hypopnea.
- the method may comprise:
- a diagnosis of a pathology linked to the apnea+hypopnea index for example OSAS.
- the method may comprise:
- a diagnosis of the pathology linked to the apnea+hypopnea index for example OSAHS.
- the method makes it possible to determine a degree of severity of OSAHS.
- a second aspect, combinable or separable from the first aspect relates to a method executed by a computer and comprising at least the steps stated above in relation to the method according to the first aspect.
- the method executed by a computer can comprise any step of the method capable of being executed on a computer, for example by a computer program product and/or a remote server.
- the method for determining the apnea+hypopnea index of the patient may more particularly comprise the method executed by a computer according to this aspect.
- a third aspect relates to a remote server capable of communicating with a computer program product, and comprising means for implementing the method according to the first aspect and/or the method according to the second aspect.
- the remote server includes:
- a database relating to a set of distinct patients comprising a plurality of data sets, each set of data from the database: o comprising at least maxillofacial morphology data relating to a patient of the set , o being associated with an apnea+hypopnea index, or a labeling function of an apnea+hypopnea index, and
- an automatic learning model trained to predict an apnea+hypopnea index, or a labeling function of an apnea+hypopnea index, for said data set, from the database.
- the remote server is configured for:
- the remote server can be configured to update the database following receipt of a set of data relating to the patient, supplemented by the patient's apnea+hypopnea index measured by polysomnography or ambulatory polygraphy .
- the measured apnea+hypopnea index can be included in the set of data relating to the patient. This set can be added to the database and thus update it, to improve the machine learning model.
- a fourth aspect relates to a computer program product comprising instructions, which when they are carried out by at least one processor, execute at least, from the method according to the first aspect and/or from the method according to the second aspect, a sending to the remote server of the set of data relating to the patient, the set of data relating to the patient comprising, or making it possible to determine, data characteristic of the at least maxillofacial morphology of the patient with respect to a reference morphology, said data characteristics being a function of the positioning of at least four homologous points, on a 3D scan of a portion of the patient's head representing his at least maxillofacial morphology, where appropriate the determination of said data being a function of the positioning of at least four homologous points, on a 3D scan of a portion of the patient's head representing at least his maxillofacial morphology.
- the ordination program product comprises the determination of said data according to the positioning of at least four homologous points, on the 3D scan of a portion of the patient's head representing his at least maxillofacial morphology.
- the computer program product can be configured to receive, from the remote server, the apnea+hypopnea index.
- a fifth aspect relates to a computer program product comprising instructions, which when performed by at least one processor, execute at least the method according to the first aspect and/or the method according to the second aspect:
- the reception of the set of data relating to the patient comprising, or making it possible to determine, data characteristic of the at least maxillofacial morphology of the patient with respect to a reference morphology, said characteristic data being a function of the positioning of at least least four homologous points, on a 3D scan of a portion of the patient's head representing his at least maxillofacial morphology, where appropriate the determination of said data being a function of the positioning of at least four homologous points, on a 3D scan of a portion of the patient's head representing his morphology at least maxillofacial,
- a sixth aspect relates to a kit for implementing the method according to the first aspect, comprising at least one computer program product according to any one of the two preceding claims, a scanner capable of acquiring the 3D scan of a portion of the patient's head representing his morphology at least maxillofacial, and preferably maxillofacial and submandibular.
- the kit may additionally include a patient head alignment device.
- FIG. 1 represents steps of the method for determining an apnea+hypopnea index of a patient, according to an exemplary embodiment, where optional steps of the method are indicated in dotted lines and parallel paths indicate variants of the method .
- FIG. 2 represents the kit and the remote server, according to an exemplary embodiment.
- Figure 3 illustrates the verification of the alignment of the patient's head by a alignment device, according to an exemplary embodiment.
- Figure 4 shows an example of data characteristic of the maxillofacial and submandibular morphology of a patient, determined from seven homologous points.
- Figure 5 is a graph of the proportion of variance expressed by the characteristic morphology data represented in Figure 4, as a function of an increasing number of principal components, after a principal component analysis.
- FIGS. 6A and 6B represent an example of projection of the patient (black), measured several times, and of patients from the database (white) in a plane defined respectively by the first and the second principal component in FIG. 6A; the first and third principal components in FIG. 6B.
- the data set relating to the patient can be provided in the form of: o a 3D scan of a portion of the patient's head representing the morphology at least maxillofacial, and preferably maxillofacial and submandibular, of the patient, o a 3D scan of a portion of the patient's head representing the morphology at least maxillofacial, and preferably maxillofacial and submandibular, of the patient, on which are positioned homologous points o characteristic data of the morphology of the patient, determined according to the positioning of homologous points on a 3D scan representing the morphology at least maxillofacial, and preferably maxillofacial and submandibular of the patient,
- the method comprising, prior to the introduction of the set of data relating to the patient in the model automatic learning: o a positioning of at least four homologous points on the 3D scan, o the determination, from the positioned homologous points, of the characteristic data of the patient's at least maxillofacial morphology compared to the reference morphology .
- the method comprising, prior to the introduction of the set of data relating to the patient in the automatic learning model, the determination, from the at least four positioned homologous points, of the data characteristic of the at least maxillofacial morphology of the patient compared to the reference morphology,
- the provided set of data relating to the patient includes the data characteristic of the at least maxillofacial morphology of the patient compared to the reference morphology, as determined according to the positioning of the homologous points on the 3D scan of a portion of the patient's head,
- the set of data relating to the patient additionally comprises, or makes it possible to additionally determine, data characteristic of the patient's submandibular morphology with respect to the reference morphology, said characteristic data being a function of the positioning of at least three points additional homologs, on the 3D scan of a portion of the patient's head representing his at least maxillofacial and submandibular morphology, where appropriate the determination of said data depending on the positioning of at least three additional homologous points, on the scan 3D of a portion of the patient's head representing his maxillofacial and submandibular morphology,
- the three additional homologous points correspond to locations on the patient's chest, for example the three homologous points correspond to the acromioclavicular joints, and/or to the sternal notch.
- the determination of the characteristic data of the morphology at least maxillofacial, preferably maxillofacial and submandibular, of the patient compared to the reference morphology includes a reduction of the dimensionality of the data resulting from the positioning of the at least four homologous points, preferably of the at least seven homologous points, for example by a principal component analysis,
- the patient data set further includes additional patient clinical information data.
- additional patient clinical information data are, for example, chosen from among an index of the patient's diverence, indices of the obstruction of the upper airways of the clinical examination, data from the medical questionnaire.
- the machine learning model may be trained to predict the apnea+hypopnea index, or labeling as a function of an apnea+hypopnea index, for a set of patient data, from the plurality of sets data from the database, each set of data from the database comprising said additional data of clinical information relating to a patient of the set,
- the set of data relating to the patient additionally comprises, or makes it possible to additionally determine, additional data characteristic of the maxillofacial morphology of the patient in at least one position among a prognathic position and a retrognathic position, preferably a position of maximum prognathia and a position of maximum retrognathia,
- the method includes the acquisition of the 3D scan of a portion of the patient's head representing his morphology at least maxillofacial, preferably maxillofacial and submandibular,
- the acquisition of the 3D scan includes a verification of the alignment of the patient's head, and more particularly of the horizontality of the reference plane of the patient's head, by an alignment device,
- the method comprises the return, from the remote server to a computer program product, of the apnea+hypopnea index, or of the labeling, predicted for the patient data set,
- the method includes displaying, by the computer program product, the apnea+hypopnea index, or labeling, predicted for the patient data set,
- the determination of the characteristic data of the morphology at least maxillofacial of the patient compared to the reference morphology includes: o an application of a reference grid resulting from the reference morphology to the 3D scan of the patient, o the positioning of the homologous points on the 3D scan of the patient, o a alignment of homologous points of the patient and the reference grid,
- the reference grid comprises semi-homologous points arranged between the homologous points, preferably equidistant from each other,
- the determination of the patient's characteristic data includes an iterative adjustment of the semi-homologous points on the reference grid so as to minimize the deformation of the grid compared to the 3D scan of the patient,
- the characteristic data of the at least maxillofacial morphology of the patient compared to the reference morphology comprise a set of homologous points and a set of semi-homologous points arranged between the homologous points, preferably the set of semi-homologous points comprises at least 200 and preferably at least 300 points.
- the data includes a large number of points to more accurately represent the patient's morphology. The reliability of the method is thus improved.
- the term 3D scan refers to an image resulting from a procedure providing the three-dimensional image resulting from at least one medical imaging technique.
- the medical imaging technique is non-irradiating for the patient.
- One type of medical imaging being considered is an infrared light scanner. This technique has the advantage of being harmless and non-ionizing for the patient. Note that it is possible to obtain a 3D scan by photogrammetry, which is just as harmless for the patient. It is also possible to provide for obtaining a 3D scan by lasergrammetry.
- the 3D scanner is made in a frontal position to the patient and represents, around an axis of revolution parallel to the vertical or longitudinal axis of the patient, at least one third, and preferably half, its maxillofacial morphology, and preferably its maxillofacial and submandibular morphology.
- the 3D scan represents the frontal morphology of the patient's head:
- a plurality of two-dimensional scans may further be used to form the 3D scan.
- homologous points points representing the same place of the face or of the bust between different 3D scans, for example of distinct patients. These points represent a homology between the 3D scans, i.e. the same morphological characteristic. As a homologous point, we can cite:
- Clinical Patient Information means data about a patient that has been acquired through examination by a healthcare professional. It may include at least one of the following: age, sex, height, weight, BMI, anthropological type, health status, history, answers to a health questionnaire, data clinical examination and patient risk factors.
- the term "automatic learning”, which can be translated into English by machine learning, refers to one or more computer algorithms capable of automatically performing one or more predictions and/or classifications without explicit programming.
- the computer algorithm can construct the learning model from learning data, and in particular from the database.
- the machine learning model can be linked to the database hosted by the remote server.
- the learning model is itself hosted on the remote server.
- the machine learning model may not be linked to the database hosted by the remote server, when implementing the method.
- the machine learning model was trained on the training data, and then implemented in a computer program product that could operate independently of the server, during the implementation of the method. This does not preclude updates to the machine learning model, for example by communication between the remote server and the computer program product.
- a parameter "substantially equal/greater/less than” a given value that this parameter is equal/greater/less than the given value, to within plus or minus 10%, or even to within plus or minus 5% of this value.
- Method 1 includes providing 10 a set of data relating to a patient.
- the set of data relating to the patient comprises, or makes it possible to determine, data characteristic of the morphology at least maxillofacial, and preferably maxillofacial and submandibular, of the patient compared to a reference morphology.
- data characteristic of the morphology at least maxillofacial, and preferably maxillofacial and submandibular, of the patient compared to a reference morphology.
- These data are determined from a 3D scan of the patient's head 6, representing his maxillofacial morphology, and preferably his maxillofacial and submandibular morphology.
- the 3D scan can be processed to determine 11 characteristic data of the patient's morphology.
- these data are characteristic of the maxillofacial and submandibular morphology of the patient.
- the characteristics described below can be applied to the example in which the characteristic data is determined from a 3D scan representing only the maxillofacial morphology of the patient.
- the set 12 of data relating to the patient comprising these characteristic data of the morphology, is introduced 13 into the automatic learning model.
- the automatic learning model is trained to predict, from a database, an apnea+hypopnea index and/or a labeling function of this index, for the set 12 of data relating to the patient.
- the database comprises a plurality of data sets of distinct persons, hereinafter referred to as patients. These patients may be healthy or present with a pathology linked to the apnea+hypopnea index.
- Each set of data includes at least morphology data and the apnea+hypopnea index, and/or a labeling function of this index, of a patient.
- the apnea+hypopnea index may more particularly have been measured, for example by polygraphy or preferably by polysomnography.
- the labeling can result from a transformation of an apnea+hypopnea index into a qualitative or quantitative value of the risk that the patient presents a pathology. This transformation can for example be carried out during the constitution of the database, or during an update of this database, described later.
- the labeling can have a value associated with a low risk and at least one value associated with a proven risk.
- the labeling presents at least one value associated with a proven moderate risk of presenting the pathology, and at least one value associated with a proven severe risk of presenting the pathology.
- the labeling can be derived from the apnea+hypopnea index, compared to one or more standard threshold values of this index.
- These threshold values are preferably standards for the risk of presenting a pathology, established in the medical field for the pathology in question.
- the labeling can present:
- the machine learning model is a supervised learning model trained to automatically predict the apnea+hypopnea index and/or labeling for the patient data set 12.
- the supervised learning model may in particular comprise a mathematical model whose input data are the set 12 of data relating to the patient, and the output data is the apnea+hypopnea index and/or the labeling.
- the database comprising a plurality of sets of patient data, each comprising the apnea-hypopnea index, and/or a labeling function of this index, and the learning model having been trained on this database , it is understood that the learning model is trained to classify the set 12 of data relating to the patient in order to assign the labeling to it, and/or to predict, from the set 12 of data relating to the patient, the index apneas+hypopneas.
- the learning model can carry out a binary, even ternary or more classification according to the number of possible labeling values.
- the apnea+hypopnea index is predicted 14, the learning model can perform a regression.
- method 1 makes it possible in a standardized and automated way to obtain the apnea+hypopnea index of the patient, from his maxillofacial and submandibular morphology.
- the apnea+hypopnea indices are measured by standard clinical methods for the patients in the database. However, it is not necessary for this measurement to be carried out for the patient whose apnea+hypopnea index is being sought. Indeed, the characteristic data of the morphology are determined from a 3D scan of the head 6 of the patient.
- Method 1 therefore makes it possible to dispense with the measurement by polygraphy or by polysomnography, for this patient.
- Method 1 also makes it possible to dispense with observation of the patient's morphology by a specialized practitioner, and/or at the very least to minimize, or even avoid, variability or a lack of appreciation that may result from this observation.
- a diagnosis 15 of a pathology linked to this index can be made, preferably supplemented by an evaluation 150 of the degree of severity of the the pathology.
- the pathology linked to this index may be OSAS, and more particularly Obstructive Sleep Apnea Syndrome, commonly abbreviated OSAS.
- OSAS is a very prevalent pathology and it is preferable to diagnose it to establish the prognosis of this other pathology, and also better control it. By way of non-limiting examples, this is particularly the case for diabetes and hypertension.
- OSAHS is very frequent and method 1 would make it possible to facilitate their characterization and/or their diagnosis.
- this is particularly the case for acromegaly, Crouzon's disease, Apert syndrome, trisomy 21.
- the method can therefore be used to diagnose a pathology for which OSAS is prevalent and/or for which OSAS is frequent.
- pathologies can be referenced by the ICD-10, the international classification of diseases, encoding in particular the diseases, signs, symptoms, social circumstances and external causes of illnesses or injuries, and published by the World Health Organization:
- Non-Alcoholic SteatoHepatitis non-alcoholic steatohepatitis
- - I509 heart failure
- the apnea+hypopnea index predicted for the patient can be compared to one or more standard values differentiating:
- the labeling predicted for the patient can be observed to diagnose the pathology, preferably according to the degree of severity of the pathology.
- the diagnosis is possible to provide for the diagnosis to be made on the basis of the apnea+hypopnea index, or the predicted labeling, for example supplemented with clinical information on the patient, in particular when the index of 'apnea+hypopnea is substantially around the standard values above, and that the diagnostic probability is reinforced by the symptoms and/or the context.
- the apnea+hypopnea index predicted for the patient can be compared to the threshold value of 15, differentiating a healthy person for an index significantly lower than 15, of a person with OSAS for an index significantly higher than 15.
- the apnea + hypopnea index predicted for the patient can be compared to the threshold value of 30, differentiating a person with OSAHS to a moderate degree for an index substantially between 15 and 30, and a person suffering from OSAHS to a severe degree for an index substantially greater than 30.
- apnea+hypopnea index for the patient can be compared to the threshold value of 15, differentiating a healthy person for an index significantly lower than 15, of a person with OSAS for an index significantly higher than 15.
- the apnea + hypopnea index predicted for the patient can be compared to the threshold value of 30, differentiating a person with OSAHS to a moderate degree for an index substantially between 15 and 30, and a person suffering from OSAHS to a
- Method 1 can be implemented by computer program product 2 and/or remote server 3.
- Computer program product 2 and/or remote server 3 can be configured to implement any computer-executed step of method 1, in particular according to their communication scenario. Different communication scenarios can be envisaged, for example illustrated by FIGS. 1 and 2.
- the set of data relating to the patient can be provided to a computer program product 2 comprising the machine learning model.
- the data set is provided by an operator, for example a healthcare professional, on software.
- Computer program product 2 can be configured to process the 3D scan, or data from this scan, to determine 11 the characteristic data of the patient's morphology.
- the computer program product 2 can be configured to introduce 13 the set 12 of data relating to the patient, comprising the data characteristic of the morphology of the patient, into the automatic learning model. Because the machine learning model is trained, the computer program product does not need to understand the database. Once identified 14, the apnea+hypopnea index predicted for the patient can then be displayed 161 by the computer program product 2.
- the data set relating to the patient can be provided 10 to a remote server 3, for example by means of a computer program product 2.
- the data set is provided 10 by an operator on software, which then sends 101 the set of data for its reception 100 by the remote server 3.
- the computer program product 2 can be configured to process the 3D scan, or data resulting from this scan, to determine 11 the characteristic data of the patient's morphology, prior to sending 101 to the remote server 3.
- the remote server 3 can be configured to process the 3D scan, or data from this scan, to determine 11 the characteristic data of the patient's morphology, following receipt of 100 of the data set.
- the remote server 3 can be configured to introduce 13 the set 12 of data relating to the patient, comprising the characteristic data of the morphology of the patient, into the automatic learning model. Since the remote server 3 hosts the database, the machine learning model can compare 140 to the database the input dataset to predict 14 the apnea+hypopnea index. The apnea+hypopnea index can then be sent 160 by the remote server 3 to the computer program product 2, for example for its display 161.
- the patient data set can be provided 10 in the form:
- the computer program product 2 and/or the remote server 3 can be configured to perform the steps leading to the determination 11 of the data characteristic of the patient's morphology.
- the computer program product 2 and/or the remote server 3 can also be configured to constitute the set 12 of data relating to the patient comprising the characteristic data of his morphology.
- method 1 may include are now detailed, with reference to figure 1.
- the method may include acquiring 18 the scan.
- the 3D scan can be carried out by a scanner 40, for example illustrated in FIG. 2.
- This scanner 40 can be part of a kit 4, in communication with the computer program product 2, which itself can be, if necessary, capable of communicating with the remote server 3, as for example illustrated by FIG. 2.
- the scan can typically be carried out approximately 40 to 50 cm from the head 6 of the patient.
- the scanner 40 can project a cloud of infrared light.
- a System Sense ⁇ commercial 3D scanner can be used. Those skilled in the art will be able to consider any technical means of medical imaging making it possible to acquire the 3D scan.
- the posture of the patient is natural, mouth closed, the lips not expressing any particular expression, and a gaze bearing on the horizon.
- the method may comprise a verification 180 of the alignment of the head 6 of the patient, and more particularly a verification of the horizontality of the reference plane of his head 6
- the reference plane is a plane passing through the upper part of the connection between the ears and the skull, and the eyes, as illustrated for example by the plane P in figure 3.
- the patient's posture can be stabilized by a gantry, for example installed along a wall.
- the horizontality of the reference plane can be checked using a device 41 worn by the patient.
- this device 41 can comprise glasses equipped with means for checking the horizontality of the reference plane. These means are for example mechanical means such as spirit levels, or electronic means such as an inertial unit.
- a slot provided on at least one arm of the glasses, in the longitudinal direction of the arm, can make it possible to check the alignment of the patient's eye with the glasses.
- verifying 180 the patient raises or lowers the head until reaching horizontality.
- the horizontality of the reference plane on the roll and pitch axes can be ensured by measuring the acceleration remaining preferably vertical during the acquisition 18. Maintaining the heading or yaw can be ensured by measuring the the azimuth at the start of or before the acquisition, this value preferably remaining substantially constant during the acquisition 18 of the 3D scan.
- the acquisition 18 can comprise an alert step when a modification of the posture of the patient is observed, for example by the measurements of the acceleration and the azimuth.
- Method 1 can also include a verification of the 3D scan after its acquisition, so as to refuse the scan of a poorly positioned patient, for example with his head tilted or whose posture would have moved during the acquisition. The reliability of method 1 is thus improved.
- Homologous points 1100 are positioned 110 on the 3D scan. These homologous points allow a first processing of the scan, so as to eliminate the effects of scale between the patients, to improve the alignment between the morphological data of different patients (processing commonly referred to by the English term "registration", aiming to align the same object on two different images using a transformation). In addition, these points can provide an anchor for the determination of semi-homologous points, described later.
- the homologous points 1100 can be positioned 110 manually by an operator, for example by a healthcare professional. This positioning 110 can be done in an automated way by the computer program product 2 or the remote server 3. After the automated positioning 110, the method 1 can comprise a verification by the operator of the position of these points, the operator being able to move them if their placement is not correct.
- At least three, preferably four, more preferably six, and even more preferably seven homologous points 1100 can be positioned 110 on the 3D scan. These points can be points located on the face, for example the base of the two ears and/or the pupils, the base of the nose, preferably completed by the tip of the chin. The points located on the face are easily spotted on the 3D scan. When the homologous points 1100 are all points located on the patient's face, the 3D scan may only represent the patient's maxillofacial morphology.
- At least one, preferably two, and more preferably three additional homologous points 1100 are positioned 110 on the 3D scan, these points being placed on the patient's chest, for example at the acromio-articular joints. clavicular, and/or sternal notch.
- the 3D scan then represents the maxillofacial and submandibular morphology of the patient.
- Method 1 thus makes it possible to take into account not only the maxillofacial morphology of the patient, but also the morphology of his bust.
- the morphological characteristics analyzed represent a broader morphology of the patient. Method 1 is therefore improved.
- the determination of the characteristic data of the patient's submandibular morphology is facilitated.
- the submandibular morphology is mainly formed of soft tissues, for which the determination 11 of these data could induce a high variability impacting the processing by the machine learning model, without the addition of these additional 1100 homologous points.
- the reliability of the method is therefore further improved.
- targets are preferably placed beforehand on the patient at these points, for example on the two acromioclavicular joints, and/or at the sternal notch, during the acquisition 18 of the 3D scan.
- the placement of these targets facilitates the identification of homologous points 1100 of the bust regardless of the patient's body morphology, and in particular when the reliefs of the bust are not clearly apparent on the 3D scan due to the presence of fatty tissue.
- the characteristic data of the patient's morphology are determined in relation to a reference morphology.
- the reference morphology can be the morphology of a reference individual.
- a reference individual can be a selected patient from the database.
- the reference individual is chosen and/or determined from the patients of the database, for example the average shape of at least a part, and preferably of all, of all the patients of the database, or the individual closest to it.
- the reference morphology can be obtained by an analysis configured to align the 3D scans in a single and same landmark by operations chosen among rotation, translation and scaling operations. According to one example, this analysis is a Procrustes analysis, preferably carried out from the homologous points 1100 positioned on the 3D scans of the database.
- the reference morphology makes it possible to constitute a grid (which can be translated into English by the term “patch”) of reference to be applied to the 3D scan of the patient.
- the reference individual can be chosen during the constitution of the database, and preferably be updated during an update of the database. data, described in detail later.
- the 3D scan can be processed in order to align the homologous points 1100 of the patient with those of the reference individual.
- the reference grid can be mapped onto the 3D scan.
- the grid is then deformed to make the corresponding points 1100 of the grid coincide with those of the 3D scan of the patient.
- Semi-homologous points 1110 can then be determined 111 to constitute the characteristic data of the patient's morphology. Provision can be made for these points to be arranged on the grid, approximately equidistant from each other between the homologous points 1100, so as to map the surface of the 3D scan.
- the technique known as sliding semi-homologous points which can be translated into English by sliding semi-landmarks, is used.
- the positioning of at least 200 and preferably at least 300 semi-homologous points can thus be optimized 111.
- the homologous and/or semi-homological points can then be replaced in a similar way to the patients of the database, so as to place all the individuals in the same space.
- a second Procrustes analysis can be performed to align all of the homologous 1100 and/or semi-homologous 1110 points with all of the individuals in the database.
- the data characteristic of the morphology of the patient 6 can comprise a set of homologous points 1100 and a set of semi-homologous points 1110 comprising at least 200 and preferably at least 300 points, these sets of points being arranged in a three-dimensional space, as for example shown in Figure 4.
- the method can comprise a reduction 112 of the dimensionality of these data.
- This reduction 112 can be performed using any multidimensional analysis technique accessible to those skilled in the art, depending on the nature of the data (for example, qualitative or quantitative).
- the volume of data introduced into the machine learning model can be reduced, while retaining the essential three-dimensional information of the patient's morphology.
- this reduction 112 can be carried out using an analysis into principal components (commonly abbreviated PCA).
- PCA has the effect of projecting the data in a base of orthogonal principal components, in order to limit the number of variables necessary to apprehend the geometric variations between the individuals, and to keep only the data projected on this limited number of variables.
- FIG. 5 represents the proportion of variance expressed 8 by the characteristic data of the morphology represented, as a function of the number of main components retained 7. By keeping the projected data in a base comprising the first four main components, 75% of the variance is expressed, and 95% by keeping the projected data in a base comprising the first twenty principal components.
- the data projected into a base comprising at least the four, preferably at least the ten, and more preferably at least the twenty, first principal components are kept in the set 12 of data which will be introduced 13 into the model of machine learning.
- Figures 6A and 6B represent an example of projected data of a patient (in black) whose characteristic data of his morphologies have been determined several times, from separate scans, and from all the patients in the database. (in white) in two planes defined by the first three principal components designated PC1: 70, PC2: 71 and PC3: 73.
- PC1: 70, PC2: 71 and PC3: 73 The variability of the data for all the patients is much greater than the variability of the data for a single patient whose analysis was replicated.
- the analytical error possibly related to the posture of the patient, to the acquisition by the 3D scanner, or to the placement of the homologous points turns out to be low in the face of the natural variability of the data of all the patients.
- the set 12 of data relating to the patient, introduced 13 into the automatic learning model can comprise the data characteristic of the morphology of the patient, resulting from the processing operations detailed above. Furthermore, this set 12 can include additional data 120.
- This data 120 can be clinical information data for the patient. The addition of these data makes it possible to complete the morphological characteristics of the patient, and thus to further improve the method 1.
- these data 120 can comprise information obtained by a medical questionnaire, such as the BERLIN and NCSAS questionnaires commonly used for the prediagnosis of OSAS.
- these data 120 may include various morphological or physiological measurements such as hip circumference, cervical and abdominal perimeters, body mass index BMI, calculated on the basis of the patient's mass and height, arterial hypertension.
- these data 120 can comprise a criterion representative of the pharyngeal anatomy, for example qualified according to the Mallampati classification system. Note that dimensionality reduction can be applied for additional data 120.
- this set 12 can comprise additional data 121 characteristic of the morphology of the patient in a prognathic position and/or a retrognathic position, these positions preferably being the maximum positions attainable by the patient.
- a 3D scan can be acquired for one or each of these positions.
- This or these 3D scan(s) can be processed in the manner detailed above for the 3D scan in normal position.
- the addition of these data makes it possible to take into account the data of the cricomental space to obtain the apnea+hypopnea index, and thus further improve method 1.
- taking into account the cricomental space can predicts the success of a mandibular advancement prosthesis for the treatment of OSAHS.
- the patient data set 12 is fed 13 into the machine learning model.
- the machine learning model is trained on the database.
- the training consists of a learning phase, carried out on validated cases.
- the machine learning model may have been trained by several supervised classification procedures.
- On a non-limiting basis mention may be made of different types of discriminant analyses, the support vector machine, the decision tree, neural networks, algorithms based on sets such as for example random forests (also referred to as decision tree), the algorithms commonly referred to in the field as “Extreme gradient boosting”, abbreviated XGBoost, “Adaptive Boosting”, abbreviated ADABoost, “majority voting”, “soft voting”, or their combination.
- the XGBoost algorithm has the particular advantage of accepting missing data in a data set.
- the person skilled in the art is able to choose the appropriate learning procedure, in particular by evaluating the predictive power of the model, as well as the associated error.
- the trained machine learning model is configured to best predict an apnea+hypopnea index from data set 12.
- the error associated with the apnea+hypopnea index predicts 14 for set 12 of data, can be transmitted 160 and/or displayed 161 to the operator with the apnea+hypopnea index.
- the database comprises a plurality patient data sets.
- Each set of data can comprise morphology data of a patient of the set, the apnea+hypopnea index of this patient, supplemented by the additional 120 and additional 121 data detailed previously.
- at least part of the sets, and preferably each set comprises the same data, resulting from the same processing operations, as the set 12 of data introduced 13 into the automatic learning model.
- the database is preferably anonymous, to comply with data protection requirements.
- the database can comprise groups grouping together several sets of data of distinct patients, grouped for example according to their anthropological type. Indeed, the reliability of the prediction by the machine learning model can be improved by comparing the morphological characteristics for a given anthropological type.
- the set 12 of data introduced 13 into the model can comprise a value, for example qualitative, of a class representative of the anthropological type, for example Caucasian, African, Asian.
- the database can be updated 17.
- the index d 'apnea+hypopnea and its data set can be transmitted 171 to the server 3 to be added to the database.
- the machine learning model can be re-evaluated on the new database, in particular to improve its predictive power and minimize the error of the prediction.
- the model can thus itself be updated, to be used for future predictions.
- the updated model can be sent from remote server 3 to computer program product 2.
- the database can preferably be updated with healthy patients, so as not to bias the model with the only introduction of patients suffering from the pathology.
- the invention proposes a solution for making reliable and/or standardizing, by analyzing characteristics morphology of an upper part of the human body, the diagnosis of a pathology, and more particularly the diagnosis of obstructive sleep apnea-hypopnea syndrome.
Landscapes
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Public Health (AREA)
- Medical Informatics (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Pathology (AREA)
- Physics & Mathematics (AREA)
- Veterinary Medicine (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Molecular Biology (AREA)
- Heart & Thoracic Surgery (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Psychiatry (AREA)
- Signal Processing (AREA)
- Dentistry (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| FR2013872A FR3117765B1 (fr) | 2020-12-22 | 2020-12-22 | Procédé standardisé de détermination d’un indice d’apnées+hypopnées ou d’une labélisation fonction de cet indice |
| PCT/EP2021/086651 WO2022136190A1 (fr) | 2020-12-22 | 2021-12-17 | Procédé standardisé de détermination d'un indice d'apnées+hypopnées ou d'une labélisation fonction de cet indice |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| EP4266981A1 true EP4266981A1 (fr) | 2023-11-01 |
Family
ID=77021365
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| EP21836200.2A Withdrawn EP4266981A1 (fr) | 2020-12-22 | 2021-12-17 | Procédé standardisé de détermination d'un indice d'apnées+hypopnées ou d'une labélisation fonction de cet indice |
Country Status (4)
| Country | Link |
|---|---|
| US (1) | US20240050030A1 (fr) |
| EP (1) | EP4266981A1 (fr) |
| FR (1) | FR3117765B1 (fr) |
| WO (1) | WO2022136190A1 (fr) |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009043080A1 (fr) * | 2007-10-04 | 2009-04-09 | Northern Sydney And Central Coast Area Health Service | Appareil et procédé pour évaluer la prédisposition à des états ayant une relation avec la morphologie cranio-faciale |
| ES2386668A1 (es) * | 2011-12-30 | 2012-08-24 | Universidad Politécnica de Madrid | Sistema de análisis de trastornos del sueño a partir de imágenes. |
| ES2984590T3 (es) * | 2019-05-10 | 2024-10-30 | Huijia Health Life Tech Co Ltd | Sistema de detección de cese respiratorio y soporte de almacenamiento |
-
2020
- 2020-12-22 FR FR2013872A patent/FR3117765B1/fr active Active
-
2021
- 2021-12-17 WO PCT/EP2021/086651 patent/WO2022136190A1/fr not_active Ceased
- 2021-12-17 US US18/259,033 patent/US20240050030A1/en active Pending
- 2021-12-17 EP EP21836200.2A patent/EP4266981A1/fr not_active Withdrawn
Also Published As
| Publication number | Publication date |
|---|---|
| WO2022136190A1 (fr) | 2022-06-30 |
| FR3117765B1 (fr) | 2023-02-24 |
| FR3117765A1 (fr) | 2022-06-24 |
| US20240050030A1 (en) | 2024-02-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US12137979B2 (en) | Eye system | |
| US11691073B2 (en) | Extended reality AR/VR system | |
| US11596316B2 (en) | Hearing and monitoring system | |
| Majumder et al. | Smartphone sensors for health monitoring and diagnosis | |
| US10176642B2 (en) | Systems and methods for computer assisted operation | |
| US10492981B1 (en) | Systems and methods for computer assisted operation | |
| US10685488B1 (en) | Systems and methods for computer assisted operation | |
| US20220218198A1 (en) | Method and system for measuring pupillary light reflex with a mobile phone | |
| CN110135528A (zh) | 年龄确定方法、眼部健康程度确定方法和装置 | |
| US12014827B2 (en) | Using artificial intelligence and biometric data for serial screening exams for medical conditions | |
| US20260053433A1 (en) | Multi-modal medical condition identification | |
| Papri et al. | IoT and cloud-based non-invasive diabetes detection system from photoplethysmogram | |
| WO2022136190A1 (fr) | Procédé standardisé de détermination d'un indice d'apnées+hypopnées ou d'une labélisation fonction de cet indice | |
| EP4240230A1 (fr) | Système de détection d'une maladie | |
| Binu | Comprehensive framework for ocular disease detection: utilizing Gegenbauer graph neural networks and fundus image data fusion techniques for enhanced classification of diverse ocular conditions | |
| Kiprijanovska et al. | Smart glasses for gait analysis of Parkinson’s disease patients | |
| Karnowski et al. | Retina image analysis and ocular telehealth: the Oak Ridge National Laboratory-Hamilton Eye Institute case study | |
| WO2022258560A1 (fr) | Système d'aide pour fournir une information de diagnostic | |
| WO2023001549A1 (fr) | Système d'aide pour fournir une information de diagnostic | |
| EP3142561A1 (fr) | Procede de traitement d'images en vue de determiner un degre de mobilite des cordes vocales | |
| CN117218154B (zh) | 一种元诊室场景下基于信息熵的患者关注度判别方法 | |
| US20250218576A1 (en) | Intelligent clinical user interfaces | |
| Zouaghi et al. | AI-Powered Early Detection of Ophthalmic Diseases Through Eye Aspect Ratio Analysis in Smartphone Videos: Dry Eye and Refractive Errors Diseases Study-Case | |
| Hsu et al. | Deep learning for automated diabetic retinopathy screening fused with heterogeneous data from EHRs can lead to earlier referral decisions. Transl Vis Sci Technol. 2021; 10 (9): 18 | |
| CN121586537A (zh) | 肾脏疾病诊断方法及装置 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: UNKNOWN |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE |
|
| PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE |
|
| 17P | Request for examination filed |
Effective date: 20230614 |
|
| AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR |
|
| DAV | Request for validation of the european patent (deleted) | ||
| DAX | Request for extension of the european patent (deleted) | ||
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: EXAMINATION IS IN PROGRESS |
|
| 17Q | First examination report despatched |
Effective date: 20240628 |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE APPLICATION HAS BEEN WITHDRAWN |
|
| 18W | Application withdrawn |
Effective date: 20241217 |