System and method for predicting and diagnosing osteoporosis based on magnetic resonance imaging technology

Abstract

A computer-implemented system and method for predicting and diagnosing osteoporosis based on magnetic resonance imaging technology is disclosed. In particular, the method of the present invention is a reliable, reproducible and sensitive method for providing a reliable, reliable, reproducible and sensitive method for detecting, by simple and fast statistical processing, i.e. Only by statistically correlating recorded MRI of a subject with recorded MRI of a reference group, independent of other standard techniques such as, for example, clinical diagnosis of MOC, QCT, DXA or osteoporosis. Osteoporosis is predicted or diagnosed from the magnetic resonance image. In this aspect, any patient who experiences magnetic resonance imaging for reasons independent of osteoporosis can receive his / her diagnostic information about the medical state of osteoporosis, regardless of how, without additional studies or tests and without the use of ionizing radiation, regardless of the suspicion of the previous osteoporotic pathology. This is very valuable and ideal in that in this way, although the asymptomatic nature of osteoporosis, it is possible to early identify patients at risk of osteoporosis or in which pathology has progressed. In addition, the overall method of the invention is very convenient, so that neither the patient nor the national health system assumes additional expenses.

Description

[0001] Description Field of the invention

[0002] The present invention relates to a computer-implemented method for predicting and diagnosing osteoporosis based on magnetic resonance imaging techniques. In particular, the method of the present invention allows to give a quick and reliable result by simply elaborating the images produced by magnetic resonance (preferably lumbar magnetic resonance) and comparing them with a pre-acquired database. In this respect, the present invention also relates to a computer-implemented system for carrying out the above-mentioned method.

[0003] The present invention also relates to a computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of the above-mentioned method, and to a computer-readable data carrier having said computer program stored thereon. State of the art

[0004] Osteoporosis ("OP") is characterized by a generalized decrease in bone mass, leading to an increased bone fragility and an increased risk of bone fracture. Osteoporosis is an asymptomatic disease, meaning that the patient does not show evident symptoms at the onset or during the progression. Therefore, the diagnosis of osteoporosis is usually made after the patient has suffered a fragility (osteoporotic) bone fracture. Osteoporotic fractures most often occur at the femoral neck, the hip, the spine or the forearm. In these cases, the term "established osteoporosis" is used, i.e. when a broken bone due to osteoporosis has already occurred.

[0005] Osteoporotic fractures are caused by an increased bone fragility, which can occur due to low bone mass and changes in the microarchitecture of the bone tissue. In developed countries, the lifetime risk of being affected by a vertebral or non-vertebral fracture is 30% - 40%, osteoporosis has a similar incidence as coronary heart disease. Moreover, osteoporotic fractures, except for forearm fractures, are associated with an increased mortality. Most fractures cause acute pain and lead to hospitalization, immobilization and usually slow recovery of the patient.

[0006] Furthermore, osteoporotic symptoms are often observed in patients with type 2 diabetes, who are generally at an increased risk of fragility fractures. Diabetes refers to a group of metabolic diseases in which a subject has high blood sugar. Type 2 diabetes is caused by insulin resistance, a condition in which cells fail to use insulin properly, sometimes also with an absolute deficiency of insulin. This form was formerly called non-insulin-dependent diabetes mellitus (NIDDM) or "adult-onset diabetes".

[0007] Therefore, accurate tools for early prediction, diagnosis and fracture risk assessment are essential for directing the right therapeutic approach to OP patients in need thereof.

[0008] Current methods for assessing fracture risk and treatment response include non-invasive imaging techniques and analysis of clinical parameters and biochemical markers of bone turnover.

[0009] The most currently used techniques for diagnosing osteoporosis are CBM (i.e. computerized bone mineral determination), QCT (i.e. quantitative computed tomography) and DXA (i.e. dual-energy X-ray absorptiometry).

[0010] In particular, CBM is a diagnostic technique for assessing bone mineralization. This technique expresses the bone mass density (BMD) as a T-score or Z-score using the penetration of X-rays or ultrasound. In more detail, the T-score expresses the difference in the number of standard deviations between the individual value of the patient and the mean value of a healthy reference population, while the Z-score expresses the difference between the value of the patient and the value of a healthy reference population consisting of subjects of the same gender and age. The BMD value reveals the state of bone degeneration and gives an indication of prognosis and diagnosis of osteoporosis.

[0011] QCT is a medical technique for measuring BMD by converting the Hounsfield Units (HU) of CT images into BMD values using a standard X-ray computed tomography (CT) scanner with calibration standards. Quantitative CT scans are mainly used to assess bone mineral density at the lumbar spine and the hip.

[0012] DXA is a well-standardized and easy-to-use quantitative imaging modality with high precision (maximum acceptable error of precision 2-2.5%) and uses an almost negligible radiation dose. In 1994, BMD testing using DXA has been adopted as a reference standard for managing osteoporosis, using the T-score to classify BMD.

[0013] In the prevention, diagnosis and management of osteoporosis, the assessment of fracture risk and the monitoring of treatment response are the two most important aspects. Therefore, the analysis of bone mass by measuring bone mineral density (BMD) is the only skeletal clinical parameter routinely analyzed in current clinical practice and is part of the WHO FRAX questionnaire (Kanis et al., 2013).

[0014] However, since osteoporosis is an "asymptomatic disease", the subject usually has no reason to undergo the examination techniques in the absence of any symptoms that would prompt him / her to check his / her own medical condition. This exposes the subject to the risk of fracture without the opportunity to prevent it through early investigation.

[0015] Therefore, there is felt the need for a reliable method for the early prediction and diagnosis of osteoporosis that overcomes the drawbacks of the prior art and gives significant results that the patient can demonstrate and understand.SUMMARY SUMMARY

[0016] This and other objects are achieved by a computer-implemented system and method for predicting and diagnosing osteoporosis based on lumbar magnetic resonance imaging techniques as defined in the claims.

[0017] As will be apparent from the following detailed description, the method of the present application allows to reliably predict the risk of suffering from osteoporosis or to diagnose osteoporosis starting from a simple recorded magnetic resonance imaging (in short "MRI") of the lumbar spine, regardless of the reason for which the MRI was prescribed to the subject.

[0018] The method is based on the acquisition of the output of a recorded lumbar MRI, which is then analyzed and statistically processed to give a quick and reliable result on the risk of suffering from osteoporosis (i.e. osteopenia) or on the assessment of ongoing osteoporosis and on the assessment of confirmed osteoporosis.

[0019] The method of the present application is therefore advantageously independent of any other known technique used as a reference standard, such as, for example, CBM, DXA, QCT, or clinical diagnosis of osteoporosis.

[0020] Moreover, the method of the present application has resulted in a very quick and effective service for the medical operator of only a few steps, thus making the whole prediction and diagnosis process very convenient both in terms of economy and operation.

[0021] The present application also relates to a computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of the method of the present application.

[0022] Furthermore, the present application relates to a computer-readable data carrier having stored thereon the computer program. BRIEF DESCRIPTION OF DRAWINGS

[0023] The characteristics and advantages of the present application will be clear from the following detailed description of the present application, from the working examples provided for illustrative and non-limiting purposes, and from the attached drawings, in which: - Figure 1 An exemplary sagittal Tl-weighted magnetic resonance image of the lumbar spine of a reference subject is depicted, which is processed according to the procedure set out in Example 1, - Figure 2 An exemplary visual depiction of the calculation of the M-score of the subjects of the reference group 2, obtained according to the procedure set out in Example 1, is depicted, - Figure 3 A depiction of the condition detected for the subject of interest of Examples 1-3 according to step V-i) of the method of the present application, - Figure 4An overview of the elements of the computer-implemented system of the invention in an exemplified version according to example 4 is depicted, and - Figure 5 An exemplary organization of data within the database according to example 4 is shown. Detailed description of the invention

[0024] The invention thus relates to a computer-implemented method for predicting and diagnosing osteoporosis based on lumbar magnetic resonance imaging technology, the method comprising the steps of: I) collecting recorded lumbar MRI outputs obtained from a magnetic resonance device by means of a medical workflow management system selected from the group consisting of RIS, PACS, DICOM and CIS, and archiving to create a reference database, wherein the lumbar MRI outputs are classified into three different reference groups according to the following criteria: Reference Group 1 (RG1) - MRI outputs of subjects aged 20-30 years Reference Group 2 (RG2) - MRI outputs of subjects aged 75 years and older Reference Group 3 (RG3) - MRI outputs of subjects who have received a confirmed diagnosis of osteoporosis after detection of a fragility fracture within the lumbar region, II) calculating the M-score for each MRI output of step I) according to the following formula: M-score = (SNR L1-L4 - SNR REF ) / SD REF (A) or M-score = (SQR L1-L4 - SQR REF ) / SD REF (B) wherein: SNR L1-L4 denotes the signal-to-noise ratio within the sagittal T1-weighted spin echo sequence of the MRI output, obtained by dividing the intra-vertebral signal intensity in the L1 to L4 vertebrae by the standard deviation of the noise, SNR REF denotes the average SNR of the reference group to which the MRI output belongs, SQR L1-L4 denotes the signal-to-liquor ratio within the sagittal T1-weighted spin echo sequence of the MRI output, obtained by dividing the intra-vertebral signal intensity in the L1 to L4 vertebrae by the standard deviation of the liquor, SQR REF denotes the average SQR of the reference group to which the MRI output belongs, and SD REFthe standard deviation of the reference group to which the MRI output belongs, and then calculating the M-score average of each reference group of step I), named M-score (RG1), M-score (RG2) and M-score (RG3), respectively, III) examining the recorded lumbar MRI output of the subject of interest ("SOI") by analyzing the sagittal T1-weighted spin echo sequence in the vertebrae from L1 to L4, thereby obtaining the relevant SNR L1-L4 ("SOI-SNR L1-L4 ") or SQR L1-L4 ("SOI-SQR L1-L4 "), IV) calculating the M-score of the subject of interest with respect to each reference group of step I) according to the following formula: M-score (1) = [(SOI-SNR L1-L4 ) - SNR REF (RG1)] / SD REF (RG1) M-score (2) = [(SOI-SNR L1-L4 ) - SNR REF (RG2)] / SD REF (RG2) M-score (3) = [(SOI-SNR L1-L4 ) - SNR REF (RG3)] / SD REF (RG3) or M-score (1) = [(SOI-SQR L1-L4 ) - SQR REF (RG1)] / SD REF (RG1) M-score (2) = [(SOI-SQR L1-L4 ) - SQR REF (RG2)] / SD REF (RG2) M-score (3) = [(SOI-SQR L1-L4 ) - SQR REF (RG3)] / SD REF (RG3) and V) comparing the three M-scores obtained in step IV) with the M-score average of the corresponding reference group obtained from step II), thereby assessing the possible risk of the subject of interest suffering from osteoporosis or being diagnosed with osteoporosis.

[0025] In practice, steps I) - V) are implemented by means of a computer program comprising instructions which cause the computer to perform said steps.

[0026] In step I) of the method, lumbar MRI outputs obtained from MR devices are collected and archived. The term "lumbar MRI outputs" refers to any lumbar MRI output obtained from a subject medical study, regardless of the underlying reason for which the MR was initially prescribed, and recorded in a digital folder together with other medical study outputs. The medical workflow management system supports and facilitates the storage, sharing and provision of said recorded MRI outputs for the purpose of creating a reference database. Advantageously, this means that the recorded lumbar MRI outputs of step I) are documents already available and retrieved from the radiology archive of the clinic or hospital.

[0027] In this regard, it should be noted that in the healthcare environment, there are several medical workflow management systems employed on a daily basis. Many, if not all, are frequently found in radiology practices and departments. Four of the most commonly utilized radiology systems include RIS (Radiology Information System), PACS (Picture Archiving and Communication System), DICOM (Digital Imaging and Communications in Medicine), and CIS (Clinical Information System).

[0028] In more detail, RIS is a root system used by imaging departments for electronic management. RIS is a radiology-specific software solution adapted to obtain, store and share medical imaging data and is widely used throughout the healthcare department. RIS is designed to optimize the efficiency of the radiology workflow, which can be used in conjunction with a hospital information system (HIS) and / or a picture archiving and communication system (PACS).

[0029] PACS is a medical imaging technology that provides data storage and convenient image access from multiple modalities. PACS is a common software solution used primarily by hospitals and healthcare organizations. PACS systems are built for secure storage and transmission of electronic patient images and data, eliminating the need for traditional methods involving manual file management (e.g. film jackets) and delivery. Although radiologists have been the primary users of PACS systems, as well as the primary generators of x-ray images, PACS has also been implemented in other health-related fields, including cardiology, oncology, dermatology, pathology, and nuclear medicine imaging. Given its rich use, PACS is designed to handle image formats generated by various modalities, such as mammography (MG), magnetic resonance (MR), ultrasound (US), computed tomography (CT), and digital radiography.

[0030] By utilizing PACS, clinicians can easily digitally access their patients' information. In addition to circumventing unnecessary testing, digital access means accelerated and improved care, minimizing the opportunity for treatment and prescription errors.

[0031] DICOM is the globally accepted standard protocol for managing and transmitting medical images and data, responsible for today's applications of PACS. In fact, DICOM is the globally accepted standard for communication and management of medical images and other patient data. DICOM is frequently utilized throughout the medical field for storing and transmitting medical images, facilitating integration with medical equipment and PACS systems.

[0032] The difference between PACS and DICOM is that PACS is a medical image storage and archive center, fed by medical modalities such as X-ray scanners and MRI machines, while DICOM represents an international communication standard used by healthcare professionals when storing, handling and transmitting medical images and data, rather than a device or structure.

[0033] CIS involves networked software solutions that work together in radiology practices, such as RIS and electronic health record systems. CIS specifically faces clinical care use (e.g., intensive care units), is a network of information systems whose computer systems are utilized in various departments of today's hospitals. These departments include cardiology departments, radiology departments and pathology departments. CIS collects patient data and transmits it into electronic records that attending clinicians can access at the bedside of a visiting patient.

[0034] For the purposes of the present invention, the creation of the reference database of MRI outputs can be assisted by any of the systems indicated above, even if PACS is the most commonly used system for this purpose.

[0035] The reference database of step I) comprises lumbar MRI outputs obtained from MR devices. Preferably, said MRI outputs are obtained from the same MR device model (i.e., different devices but with the same model). This provision allows to see the collection of MRI outputs obtained uniformly in the perspective of the instrument. In fact, MRI outputs can be obtained from different MR device models, however, it is understood that the statistical accuracy of the subsequent calculation of the M-score increases accordingly if the same model is used.

[0036] More preferably, the lumbar MRI outputs of step I) are obtained from the same MR device. This means that all lumbar MRI outputs are obtained uniformly by using the same instrument, thus maximizing the overall accuracy of the method.

[0037] In principle, from a statistical point of view, the higher the number of MRI outputs available when the MRI outputs originate from different MR device models, the lower the impact on the overall accuracy of the method. Therefore, the more homogeneous the source of MRI outputs (e.g. same MS device model, or better still, exactly the same MR device), the lower the number of MRI outputs necessary to give satisfactory accuracy of the method.

[0038] Step I) further comprises classifying the lumbar spine MRI outputs into three different reference groups according to the following criteria: Reference Group 1 (RG1) - MRI outputs of subjects aged 20-30 years, Reference Group 2 (RG2) - MRI outputs of subjects aged over 75 years, Reference Group 3 (RG3) - MRI outputs of subjects who have received a confirmed diagnosis of osteoporosis after detection of a fragility fracture within the lumbar spine region.

[0039] RG1 collects MRI outputs of subjects aged 20-30 years, who are generally and statistically healthy in terms of osteoporosis. RG1 is therefore referred to as the "healthy population" group.

[0040] RG2 collects MRI outputs of subjects aged over 75 years, who are generally and statistically at high risk of suffering from osteoporosis or currently affected by osteoporosis. RG2 is therefore referred to as the "age-related risk population" group.

[0041] RG3 collects MRI outputs of subjects who have received a confirmed diagnosis of osteoporosis after detection of a fragility fracture within the lumbar spine region. RG3 is therefore referred to as the "OP patient population" group.

[0042] In step II) of the method, the M-score of each MRI output of step I) is calculated according to the following formula: M-score = (SNR L1-L4 - SNR REF ) / SD REF (A) or M-score = (SQR L1-L4 - SQR REF ) / SD REF (B) wherein: SNR L1-L4 represents the signal-to-noise ratio obtained by dividing the intra-vertebral signal intensity in the L1 to L4 vertebrae by the standard deviation of the noise within the sagittal T1 -weighted spin echo sequence of the MRI output, SNR REFthe average of the SNR of the reference group to which the MRI output belongs, SQR L1-L4 the signal-to-cerebrospinal fluid ratio, SQR, obtained by dividing the intra-vertebral signal intensity in the L1 to L4 vertebrae by the standard deviation of the cerebrospinal fluid within the sagittal T1 -weighted spin echo sequence of the MRI output, SQR REF the average of the SQR of the reference group to which the MRI output belongs, and SD REF the standard deviation of the reference group to which the MRI output belongs, The average of the M-score of each reference group of step I), named M-score(RG1), M-score(RG2) and M-score(RG3), respectively, is then calculated.

[0043] In particular, for all the MRI outputs of the method, the imaging analysis protocol comprises a sagittal T1 -weighted spin echo sequence, which is optimal for the evaluation of vertebral fatty bone marrow, while segmenting the vertebral body from L1 to L4. Regions of interest (ROIs) are placed in the vertebral body, excluding cortical bone, subchondral abnormalities, local lesions (e.g. angioma) and posterior venous plexus. Three ROIs are used for each vertebra, each acquired on a different slice, whose mean is used for the analysis. ROIs are also positioned in non- artifacted sites outside the patient to measure the noise. The signal-to-noise ratio (SNR) is obtained by dividing the intra-vertebral signal intensity by the standard deviation of the noise. The positioning of the ROIs can be done manually by the operator using the functionalities of any processing system, such as PACS, RIS or DICOM, or, preferably, through a semi-automatic or automatic system, possibly able to use artificial intelligence with machine learning systems.

[0044] The SNR is measured in all the MRI outputs of each reference group L1-L4 and the average (SNR REF ) and the standard deviation (SD REF ) are then calculated; subsequently, the average of the M-score of each reference group, i.e. M-score(RG1), M-score(RG2) and M-score(RG3), respectively, is calculated according to formula A.

[0045] Alternatively and similarly, the average of the M-score of each reference group, i.e. M-score(RG1), M-score(RG2) and M-score(RG3), respectively, can be calculated according to formula B, with the difference that the ROIs are positioned in the cerebrospinal fluid, instead of in non-artifacted sites.

[0046] Preferably, Formula A is used when the MRI output is obtained from an MR device set on a magnetic field strength of 0.035 to 0.5 Tesla (i.e. low field strength), while Formula B is preferably used when the MRI output is obtained from an MR device set on a magnetic field strength of 1.0 to 1.5 Tesla (i.e. high field strength).

[0047] In the case of medium field (i.e. 0.5 Tesla - 1.0 Tesla), either Formula A or Formula B can be used.

[0048] Preferably, in order to increase the statistical significance of the mean value of the M-score of the reference group, a minimum of 20 MRI outputs per reference group are collected and analyzed.

[0049] On average, 30 to 100 MRI outputs per reference group give a good statistical significance.

[0050] In step III) of the method, the recorded lumbar MRI output of a subject of interest ("SOI") is examined, the examination being carried out in a similar manner to the analysis of the MRI outputs of the reference groups, i.e. by analyzing the sagittal T1 weighted spin echo sequence in the vertebrae from LI to L4, thereby obtaining the relative SNR L1-L4 ("SOI-SNR L1-L4 ") or SQR L1-L4 ("SOI-SQR L1-L4 ").

[0051] Likewise, the lumbar MRI output of the SOI is also a lumbar MRI output obtained from a subject medical study, regardless of the underlying reason for which the MR was initially prescribed, and is recorded in a digital folder together with other medical study outputs of the same SOI. The medical workflow management system supports and facilitates the storage, sharing and provision of said recorded MRI outputs for the purpose of examining said MRI outputs. Advantageously, this means that the recorded lumbar MRI output of step III) is also a document that is already available and retrieved from the radiology archive of the clinic or hospital.

[0052] Preferably, the MRI of the SOI is obtained from the same MR device model as the MRI outputs used in step I). Similarly to what discussed above in relation to step I), this provision allows for the analysis of MRI outputs that are uniformly obtained from the point of view of the instrument. In practice, the MRI of the SOI can be obtained from different MR device models, however, it is understood that the statistical precision of the subsequent comparison with the reference groups increases accordingly if the same model is used.

[0053] More preferably, the lumbar spine MRI output of the SOI and the MRI output of step I) are obtained from exactly the same MR device. This means that all lumbar spine MRI outputs are homogeneously obtained by using the same instrument, thus maximizing the overall accuracy of the method.

[0054] In step IV) of the method, the M-score of the subject of interest is calculated with respect to each reference group of step I) according to the following formula: M-score (1) = [(SOI-SNR L1-L4 ) - SNR REF (RG1)] / SD REF (RG1) M-score (2) = [(SOI-SNR L1-L4 ) - SNR REF (RG2)] / SD REF (RG2) M-score (3) = [(SOI-SNR L1-L4 ) - SNR REF (RG3)] / SD REF (RG3) or M-score (1) = [(SOI-SQR L1-L4 ) - SQR REF (RG1)] / SD REF (RG1) M-score (2) = [(SOI-SQR L1-L4 ) - SQR REF (RG2)] / SD REF (RG2) M-score (3) = [(SOI-SQR L1-L4 ) - SQR REF (RG3)] / SD REF (RG3).

[0055] This means that each subject examined receives three different M-scores, which will be used in the subsequent steps for the osteoporosis risk assessment.

[0056] In step V) of the method, the resulting three M-scores of step IV) are compared with the average of the M-scores of the corresponding reference groups resulting from step II), thus assessing the possible risk of the subject of interest to suffer from osteoporosis or directly diagnosing an ongoing osteoporosis based on the recorded lumbar spine MRI output of the SOI.

[0057] In fact, the main advantage of the method of the present application is that the SOI is not required to undergo any additional medical investigations or analyses, since the method operates on the documentation of the records already present in the clinic or hospital where the SOI has previously been for any reason.

[0058] More in detail, the difference between M-score (1) and M-score (RG1) is a calculated indicator of the status of the SOI with respect to the healthy population. For this purpose, it is assumed that a difference lower than "+1" indicates a "healthy subject", while a difference equal to or higher than "+1" indicates a "non-healthy subject".

[0059] The difference between M-score (2) and M-score (RG2) is a calculated indicator of the status of the SOI with respect to the age-related risk population. For this purpose, it is assumed that a difference higher than "-1" indicates an "OP patient subject" (i.e. a subject at risk of suffering from osteoporosis or affected by osteoporosis), while a difference equal to or lower than "-1" indicates a "non-OP patient subject".

[0060] The difference between M-score (3) and M-score (RG3) is a calculated indicator of the status of the SOI with respect to the OP patient population. For this purpose, it is assumed that a difference higher than "-1" indicates an "eOP patient subject" (i.e. a subject affected by established osteoporosis (in short "eOP"), therefore characterized by fragility fractures), while a difference equal to or lower than "-1" indicates a "non-eOP patient subject" (i.e. a subject characterized by the absence of fragility fractures).

[0061] The combination of these three calculated indicators allows to unravel the overall medical status of the osteoporosis of the SOI.

[0062] Preferably, step V) of the method comprises a sub-step V-i) in which the difference between M-score (1) and M-score (RG1) and the difference between M-score (2) and M-score (RG2) are considered and compared, thus generating a first indication of the SOI. In this regard, three combinations of the calculated indicators are possible, as follows: 1) "healthy subject" and "non-OP patient subject", i.e. the SOI is characterized by satisfactory bone mineralization; 2) "non-healthy subject" and "non-OP patient subject", i.e. the SOI is characterized by osteopenia; 3) "non-healthy subject" and "OP patient subject", i.e. the SOI is characterized by osteoporosis.

[0063] [The fourth combination ["healthy subject" and "OP patient subject"] is not included, since it is in fact not feasible, as the SOI cannot be both healthy and show signs of osteoporosis.] Preferably, step V) of the method comprises a sub-step V-ii) in which the difference between M-score (1) and M-score (RG1) and the difference between M-score (3) and M-score (RG3) are considered and compared, thus yielding a second calculated indication of the SOI.

[0064] In this respect, two combinations of the calculated indicators are possible, as follows: 1) "healthy subject" and "non-eOP patient subject", i.e. SOI features satisfactory bone mineralization and does not show signs of osteoporosis => negative; 2) "non-healthy subject" and "eOP patient subject", i.e. SOI is definitely affected by diagnosed osteoporosis => positive; [other combinations of indicators are not included as they are practically unfeasible, since opposite indicators cannot be assessed on the same subject (e.g. a subject cannot be both healthy and characterized by diagnosed osteoporosis)].

[0065] In other words, the method of the present application is a method of calculating indicators of the SOI, starting from the sagittal T1-weighted spin echo sequence in the vertebrae from LI to L4 shown by the recorded lumbar MRI output of the SOI, while performing the above steps I) - IV), thus obtaining calculated indicators to be compared in step V) with reference indicators of osteopenia, osteoporosis risk or diagnosed osteoporosis. This means that the above method comprising steps I) to V) represents a data processing method in a subsequent method that can be used for the prognosis or diagnosis of osteoporosis.

[0066] If the method yields an indication of osteopenia or directly yields a diagnosis of osteoporosis or diagnosed osteoporosis, a corresponding alert can be sent to the SOI, so that the latter can immediately be aware of his / her own condition and contact an expert for further investigations and assistance.

[0067] In preferred embodiments, step V) of the method comprises both sub-step V-i) and sub-step V-ii), preferably in this order. These embodiments are particularly advantageous, as the combination of steps V-i) and V-ii) goes as far as allowing not only comparing the MRI signal of the SOI with the signal observed for the healthy population, but also comparing the MRI signal of the SOI with the signal observed for the age-related risk population and for the eOP patient population, thus yielding a complete and consistent overview of the SOI condition. In fact, comparing the MRI of the SOI with the MRI of the age-related risk population allows assessing whether there is ongoing osteoporosis, and comparing the MRI of the SOI with the MRI of the eOP patient population allows assessing whether there is a fragility fracture, i.e. diagnosed osteoporosis.

[0068] The method of the present application is therefore a reliable, reproducible and sensitive method for predicting or diagnosing osteoporosis from magnetic resonance images, independently from other standard techniques, such as, for example, MOC, QCT, DXA or clinical diagnosis of osteoporosis, by simple and fast statistical processing, i.e. by statistically correlating the MRI of the subject with the MRIs of the reference group.

[0069] In view of the above, the method of the present application is evidently characterized by several advantages and benefits under many points of view.

[0070] First of all, any patient who undergoes lumbar magnetic resonance imaging for reasons unrelated to osteoporosis can anyway receive his / her diagnostic information on the medical status of osteoporosis without further investigations or tests and without the use of ionizing radiation, regardless of the suspicion of previous osteoporotic pathology.

[0071] This is very valuable and desirable because in this way, despite the asymptomatic nature of osteoporosis, it is possible to identify early patients at risk or with ongoing osteoporosis pathology.

[0072] Moreover, the overall method of the present application is very convenient so that neither the patient nor the national health system incurs additional costs.

[0073] Therefore, those who directly benefit from the method of the present application are: - patients at risk of suffering from osteoporosis or with ongoing osteoporosis, who are completely unaware of their own condition. These patients are timely alerted by the method described above and can start treatment in time with their doctor and the therapeutic- preventive indication of the case; - the national health system: prediction and / or early diagnosis based on the magnetic resonance test already on record, without additional costs. Moreover, early diagnosis allows to significantly reduce the incidence of fractures attributable to osteoporosis and therefore the relatively high costs of their management, which equals considerable overall health care costs saved for the national health system and for the patient himself; - public, private or accredited private radiology services with magnetic resonance equipment: without additional costs or increased machine time compared to standard examinations, they can provide information on osteoporotic pathology to patients, regardless of the potential reason for which the MR was prescribed.

[0074] In view of the large number of magnetic resonance investigations performed every day for various clinical indications independent of osteoporosis, the method of the present application advantageously allows to identify a much greater number of patients at risk or with ongoing osteoporosis pathology than the number of classic CBM investigations, who would be able to obtain adequate preventive or therapeutic measures.

[0075] Another advantage is that the method of the application can be applied to any MRI equipment, regardless of the supplier or the magnetic field strength, and to any data storage and processing system, such as those mentioned above, for example PACS, RIS or DICOM. Moreover, there are no exclusion criteria for patients, since the method is universally applicable to any MRI output.

[0076] It is also clear that the reference database and the corresponding three reference groups enjoy a continuous upload of MRI outputs during the daily clinical practice, so that the accuracy of the resulting method results, i.e. the prediction and / or diagnosis of osteoporosis, is therefore and advantageously increased over time. In other words, the more the method is implemented, the more potential osteoporosis patients are alerted.

[0077] The application also relates to a computer program comprising instructions which, when the program is executed by a computer, cause the computer to perform the steps of the method of the application.

[0078] Furthermore, the application relates to a computer-readable data carrier having stored thereon the computer program.

[0079] Additionally, the application relates to a computer-implemented system for performing the above method.

[0080] In particular, the computer-implemented system comprises: a) at least one digital repository in communication with at least one medical workflow management system of a clinic or hospital, the repository being for archiving recorded MRI outputs received from the at least one medical workflow management system, wherein, according to step I) of the above method, the MRI outputs are anonymized via software before being archived in the reference database, b) a software platform operating on the reference database, the software platform comprising a computer program designed to perform the calculations, checks and comparisons of steps II) to V) of the above method, and c) computing devices of a plurality of users for accessing the software platform in order to send queries and receive reports, each user being associated with a digital identity comprising an authentication device, wherein the authentication device comprises a processor capable of automatically generating an authentication code in response to an interrogation for verifying the digital identity of the user.

[0081] A "digital repository" is any medium or digital system capable of storing digital information. Examples of suitable digital repositories are hard disks, cloud servers, fogs, U-USB keys, etc. Preferably, the digital repository of the application is a cloud server.

[0082] Reference is made to Figure 4 , a preferred configuration of the system implementing the application. The system is cloud-based and comprises the following technical features.

[0083] M Score PACS “M-score PACS” is the name of the software instance. In cloud computing, an instance is a server resource provided by a third-party cloud service. This software has been customized to be able to integrate with the M-score viewer and provide its users with information related to the M-score method (SNR values, SNRL, SD, etc.).

[0084] The main application is developed mainly in PHP, with system services that manage DICOM, installed on an AWS EC2 Windows server machine. Alternatively, image storage is managed by AWS FsX as a Windows service.

[0085] The reference database is PostgresSQL on AWS RDS.

[0086] M Score Viewer “M-score viewer” is the name of the DICOM viewer, customized to be able to extract signal intensity values on MR images in a guided way.

[0087] It is possible to select an ROI on the region of interest, extract the data and calculate the SNR or SNRL values, as shown in Figure 1 .

[0088] M Score Platform “M-score platform” is the name of the application backend implemented through the use of the Laravel framework supported by JetStream, responsible for generating an assessment of bone health based on the graphical analysis of MR images (M-score viewer) and for statistical analysis (M-score platform) through the use of aggregated anonymous data of a group of patients called the reference population. The M-score platform operates only with anonymous data and manages users, reference populations, requests and results for each clinic or hospital. The clinic or hospital has the opportunity to access a reserved area where they can obtain an API key, configure their company data and check the number of credits available to make requests and possibly obtain detailed information for the purchase of new credits. [An API key is an alphanumeric string used by API developers to control access to their API. An API is a communication mechanism that allows data exchange between two software modules. After creating an API for your module, other application developers can call your API to integrate your functionality into their code.] Optionally, Single Sign-On (SSO) refers to the fact that a user gets access to different applications with a single authentication. As for Multi-Factor Authentication (MFA), it adds a layer of security to verify the identity of the user at the time of authentication.

[0089] The application comprises only one or more administrator accessible management section, which contains management panels for users (and thus for registered clinics / hospitals), reference populations, requests and bone health reports.

[0090] The management panel of requests allows the administrator to view the measurements via the M-score viewer and then determine the values for generating the reference populations and bone health reports and save them in the PACS.

[0091] The data is organized in a database, as exemplarily illustrated in Figure 5 .

[0092] A "user" is a clinic / hospital that has registered for the service. A user can be created via a registration form. Then, the administrator can associate a specific user with a specific unique identifier of a clinic / hospital in order to be able to associate the data present in the PACS with the data present in the application database.

[0093] A "population" is the aggregated statistical data of a population used as a reference, as referred to above as "reference group". The reference group or population is generated by the administrator using data from patients of a clinic / hospital that have had their bone density assessed, to which the data of the patients of the request report can be compared.

[0094] A "request" is a request submitted by a clinic / hospital about their patients.

[0095] A "report" is a bone health report prepared for delivery. These reports have a graphical appearance similar to the one depicted in Figure 3 .

[0096] The administrator has the following sections available: 1. User management 2. Management of reference populations 3. Management of requests The user has the following sections available: 1. User management (personal information, security options, password change) 2. API key management 3. Credit management M Score Bridge The "M-score bridge" is the name of a set of dedicated software that allows the Q / R on the clinic / hospital PACS to identify and receive the relevant tests and anonymize the patient data received from the M-score clinic portal of the M-score platform and de-anonymize the report data.

[0097] M Score Clinic Portal The "M-score Clinic Portal" is the name of the native client application designed to reside on one or more clinic / hospital computers and be used by workers to request bone health reports and download the results in PDF format when available. The application communicates with the M-score platform via an API (the API key obtained in the user section of the M-score platform must be entered into the M-score Clinic Portal application).

[0098] The application has at least 3 navigable sections: 1. API Key Configuration 2. Search Exams 3. Request List Request List The user can view a list of existing requests and their status. The data populating this section comes from the M-score platform application database, specifically from the requests and results tables, and is in anonymized form. In addition, the first and last name pair obtained during the search phase and saved when the request was created is added.

[0099] If there is a result associated with the request, the "Generate" button is enabled to generate and save the bone health report in PDF format. If there is no result associated with a specific request, the result for that specific request is "in progress".

[0100] Results If the request has been processed, the associated bone health report can be downloaded.

[0101] The report is transmitted from the M-score platform in JSON format, containing the anonymized data of the record in the results table. The M-score Clinic Portal uses the M-score Bridge to be responsible for de-anonymizing the report and finally generating a PDF file containing the report that the user can download through any electronic device.

[0102] It should be understood that the various preferred aspects described above for the preferred embodiment of the system should equally be considered as being various preferred for the method and system of the invention as exemplarily depicted in Figure 4

[0103] It should also be understood that the method and system as described above are not limited to the lumbar region, but are in fact similarly applicable to other bone segments, simply by selecting the ROI on a different anatomical region to identify the signal intensity and use the noise to obtain the signal-to-noise ratio.

[0104] The following is a non-limiting list of selectable regions segmented on a T1 MRI sequence with the relevant ROI for each bone segment.

[0105] Hip and Femur ​Femoral neck: The femoral neck is a critical region for assessing osteoporosis, and can be analyzed by segmenting the region starting at the base of the femoral neck and extending to the proximal metaphysis of the femur. It is important to include both the cortical bone and the internal trabecular bone while avoiding areas where soft tissue is significantly present.

[0106] Greater trochanter: Another relevant region is the greater trochanter, which shows significant changes in bone density in cases of osteoporosis. The ROI here includes the internal trabecular portion while avoiding the insertion of muscles and tendons to minimize interference.

[0107] Data obtained - SNR GT-CF Wrist and Hand Distal radius: In the distal part of the radius, it is useful to segment the region between the styloid process and the metaphysis. This region indicates changes in bone density, particularly in cases of osteopenia and osteoporosis.

[0108] Carpal bones: Although less common, specific bone density loss can be examined in the carpal bones. The ROI can be defined by including the scaphoid and lunate bones, which are representative of the bone health of the wrist.

[0109] Data obtained - SNR OS-RD Cervical and Thoracic Spine Cervical and dorsal vertebral bodies: Similar to the lumbar spine, the cervical and dorsal vertebrae can be examined by segmenting the vertebral bodies. The ROI is delineated to include the central portion of the vertebral bodies from C3 to D12, focusing on the internal trabecular bone to assess the presence of osteoporotic changes.

[0110] Data obtained - SNR C3-C7 and SNR D6-D12 Knee Tibial plateau: A critical region for osteoporotic analysis is the tibial plateau, which can be divided into two main ROIs: the medial portion and the lateral portion. Segmentation of these regions should include the subchondral trabecular bone, focusing on any changes in bone density and trabecular structure.

[0111] Proximal tibial metaphysis: This region is located just below the tibial plateau and is important for assessing the transition between cortical bone and trabecular bone, which can be affected by osteoporosis.

[0112] Femoral condyles: The medial femoral condyle and the lateral femoral condyle are crucial for assessing osteoporosis in the knee. The ROI in these regions includes the subchondral trabecular bone and adjacent areas to monitor bone density and identify any structural changes indicative of osteoporosis.

[0113] Distal femoral metaphysis: Similar to the tibia, this region is also important for the assessment of osteoporosis. Segmentation should cover a significant portion of the trabecular bone in the metaphysis, examining changes in density and structure.

[0114] Data obtained - SNR CF-PT Shoulder Central region of the humeral head: This region is fundamental for assessing changes in bone density and microarchitecture related to osteoporosis. The ROI should focus on the trabecular bone, paying special attention to changes in trabecular density and configuration under the cortical bone.

[0115] Anatomical neck of the humerus: The region around the anatomical neck (where the humeral head connects to the shaft) is crucial for assessing osteoporosis. Segmentation of this region should include trabecular bone to monitor potential osteoporotic sparsity.

[0116] Data obtained - SNR CO-TO Elbow Lateral epicondyle of the humerus: A region often involved in overuse injuries, such as lateral epicondylitis ("tennis elbow"). Analysis of this region can help detect early changes in bone density or signs of repeated microtrauma.

[0117] Medial epicondyle of the humerus: Important for identifying medial epicondylitis ("golfer's elbow") and other conditions that affect the tendons and bone attachments in this region.

[0118] Ulnar olecranon: The olecranon is the lever point for the forearm extensors and can be subject to fractures or other traumatic lesions. It is a key region for assessing bone integrity after trauma.

[0119] Ulnar head: The ulnar head forms part of the distal radioulnar joint, which can be affected by osteoarthritis or other degenerative conditions.

[0120] Radial head: The radial head plays a crucial role in the rotational movements of the elbow joint and forearm. Fractures and degenerative changes in this region are common and can be diagnosed through analysis of MRI images.

[0121] Data obtained - SNR EO-RU Ankle and Foot Metaphysis and tibial epiphysis: These regions are particularly prone to stress fractures, compression fractures, and degenerative changes. Analyzing the density and structural integrity of these regions can reveal the presence of osteoporosis, osteoarthritis, or other bone diseases.

[0122] Tibial plafond: The key area involved in the formation of the ankle joint. Damage to this area usually occurs after a twisting or direct impact trauma, such as an ankle sprain.

[0123] Lateral malleolus: It is less involved in load transfer than the medial malleolus, but it is very prone to fracture due to trauma or foot twisting. This area should be examined for assessing post-traumatic conditions and fractures.

[0124] Calcaneal body: As the bone that forms the heel, the calcaneus is essential for absorbing the impact during walking. Calcaneal fractures can be complex and require careful analysis to identify the fracture pattern and plan surgery or conservative treatment.

[0125] Data obtained - SNR MT-CC Depending on the chosen skeletal segment, the method and system of the present application can be successfully applied, allowing to achieve reliable and useful results in terms of assessing the risk of suffering from osteoporosis (i.e. osteopenia) and in terms of assessing ongoing osteoporosis as well as assessing established osteoporosis.

[0126] The following are working examples of the present application provided for illustrative purposes. Example

[0127] Example 1. The method of the present application has been implemented at the "Centro di Medicina Preventiva" of Rozzano (IT), a private medical center specialized in diagnostic imaging research.

[0128] In the system PACS with the interface viewer OsiriX, 2,000 MRI outputs have been collected since July 2014 to date, all obtained from the same MR device, i.e. S-SCAN (0.25 Tesla field strength) provided by ESAOTE SpA (IT).

[0129] Out of the 2,000 MRI outputs, those that meet the three reference group criteria have been selected and classified accordingly, resulting in the following: RG1 - Healthy population - 70 MRI outputs, RG2 - Age-related risk population - 90 MRI outputs, RG3 - OP patient population - 30 MRI outputs (whose patients are over 50 years old).

[0130] For all MRI outputs, the sagittal T1-weighted spin echo sequence was used, with segmentation of the vertebral bodies from L1 to L4, as Figure 1The region of interest (ROI) is placed in the vertebral body excluding cortical bone, subchondral abnormalities, focal lesions (e.g. hemangioma) and posterior venous plexus as shown in Fig. 1. Three ROIs are used for each vertebra, each acquired on a different slice, their mean value being used for analysis. ROIs are also positioned in non- artifacted areas outside the patient to measure noise. The signal-to-noise ratio (SNR) is obtained by dividing the signal intensity within the vertebra by the standard deviation of the noise. The positioning of the ROIs can be done manually by an operator.

[0131] The SNR is measured in all MRI outputs of each reference group L1-L4 and the standard deviation (SD REF ), the results are as follows: RG1 => SNR L1-L4 = 39.88 - SD = 6.13, RG2 => SNR L1-L4 = 69.10 - SD = 10.04, RG3 => SNR L1-L4 = 76.61 - SD = 9.77.

[0132] The M-score of each MRI output is calculated according to step II) of the method. An exemplary visual description of the calculation of the M-score of the RG2 subject is provided in Fig. 2. Figure 2

[0133] Subsequently, for step II) of the method, the M-score mean value of each reference group is calculated according to the formula provided above: After timely classification and analysis of the reference groups, the MRI outputs of the subject of interest of the anonymous examination are considered.

[0134] The SOI-SNR L1-L4 = 75.00 is used to calculate the M-score (1), M-score (2) and M-score (3) according to the formula described in step IV) of the method above.

[0135] According to step V) of the method, these three SOI M-scores are compared to the M-score mean value of each reference group.

[0136] M-score (1) = (75.00-39.88) / 6.13 = 5.73, M-score (2) = (75.00-69.10) / 10.04 = 0.59, M-score (3) = (75.00-76.61) / 9.77 = -0.16.

[0137] ​The results of this method on the subject of interest give the following information: Result of step V-i): combination 3) "non-healthy subjects" and "OP patient subjects", i.e. SOI characterized by osteoporosis; Result of step V-ii): positive, i.e. SOI definitely affected by confirmed osteoporosis.

[0138] In this case, SOI is diagnosed as having ongoing confirmed osteoporosis and is therefore immediately reminded to contact an expert for further help.

[0139] Example 2. The same reference groups of example 1 are maintained, while considering the MRI output of another subject of interest examined anonymously.

[0140] The resulting SOI-SNR L1-L4 = 42.00 for the calculation of M-score (1), M-score (2) and M-score (3) according to the formula described in step IV) of the method above.

[0141] According to step V) of the method, these three SOI M-scores are compared with the average of the M-scores of each reference group.

[0142] M-score (1) = (42.00-39.88) / 6.13 = 0.34 M-score (2) = (42.00-69.10) / 10.04 = -2.70 M-score (3) = (42.00-76.61) / 9.77 = -3.54 The results of this method on the subject of interest give the following information: Result of step V-i): combination 1) "healthy subjects" and "non-OP patient subjects", i.e. SOI characterized by satisfactory bone mineralization and showing no risk of suffering from osteoporosis; Result of step V-ii): negative.

[0143] In this case, SOI is a healthy subject.

[0144] Example 3. The same reference groups of example 1 are maintained, while considering the MRI output of another subject of interest examined anonymously.

[0145] The resulting SOI-SNR L1-L4 = 58.50 for the calculation of M-score (1), M-score (2) and M-score (3) according to the formula described in step IV) of the method above.

[0146] According to step V) of the method, the three SOI M-scores are compared with the M-score average value of each reference group.

[0147] M-score (1) = (58.50-39.88) / 6.13 = 3.00 M-score (2) = (58.50-69.10) / 10.04 = -1.06 M-score (3) = (58.50-76.61) / 9.77 = -1.85 The results of the method on the subject of interest give the following information: Result of step V-i): combination 2) of "non-healthy subjects" and "non-OP patient subjects", i.e. SOI characterized by osteopenia, which potentially predicts the development of osteoporosis; Result of step V-ii): negative.

[0148] In this case, the SOI is characterized by osteopenia, so the corresponding alert is sent.

[0149] In Figure 3 , the depiction of the above conditions detected after step V-i) for the subjects of interest of examples 1-3 is shown. In particular, the curves for RG1 and RG2 are drawn, with respect to which the results of the SOI are positioned. It is worth noting that even this single sub-step is sufficient for evaluating the SOI osteoporosis condition from a simple MRI analysis. In this regard, it should be reminded that the method of the present application allows for the first time to diagnose osteoporosis by relying on MRI outputs already available, and to confirm osteoporosis, i.e. without the need for additional and expensive investigations or tests and without the use of ionizing radiation, regardless of the suspicion of previous osteoporotic pathologies.

[0150] Example 4. With reference to Figure 4 , the preferred configuration of the system of the present application is implemented.

[0151] The system is cloud-based and comprises the following technical features.

[0152] M Score PACS The M-score PACS is a software instance. In cloud computing, an instance is a server resource provided by a third-party cloud service. The software has been customized to be able to integrate with the M-score viewer and provide its users with information related to the M-score method (SNR values, SNRL, SD, etc.).

[0153] The main application is developed mainly in PHP with a system service that manages DICOM, installed on an AWS EC2 Windows server machine. Alternatively, image storage is managed by AWS FsX for Windows service.

[0154] The reference database is PostgresSQL on AWS RDS.

[0155] M Score Viewer The M-score viewer is a DICOM viewer customized to be able to extract signal intensity values on MR images in a guided way.

[0156] It is possible to select a ROI on the region of interest, extract data and calculate SNR or SNRL values, as shown in Figure 1 .

[0157] M Score Platform The M-score platform (application backend) is implemented by using the Laravel framework supported by JetStream, responsible for generating an assessment of bone health based on the graphical analysis of MR images (M-score viewer) and statistical analysis (M-score platform) by using aggregated anonymous data of a group of patients called the reference population. The M-score platform operates only with anonymous data and manages users, reference populations, requests and results for each clinic or hospital. The clinics or hospitals have the opportunity to access a reserved area where they can obtain an API key, configure their company data and check the number of credits available to make requests and possibly obtain detailed information for the purchase of new credits.

[0158] The application includes an administrative section accessible only to one or more administrators, which contains management panels for users (and therefore for registered clinics / hospitals), reference populations, requests and bone health reports.

[0159] The management panel of the requests allows the administrator to make measurements via the M-score viewer and then determine the values for generating the reference population and bone health reports and save them in the PACS.

[0160] Data is organized in the database as exemplarily illustrated in Figure 5 .

[0161] A "user" is a clinic / hospital that has registered for the service. A user can be created via a registration form, which includes the following data, such as: -- 1. id_user -- 2. name -- 3. email -- 4. email_verified_at -- 5. Password -- 6. ext (PACS id) -- 7. Is_Admin (0 user, 1 admin) -- 8. Phone number.

[0162] The administrator can then associate a specific user with a specific unique identifier of the clinic / hospital in order to be able to correlate data present in the PACS with data present in the application database.

[0163] A "population" is the aggregated statistical data of the population used as a reference, as referred to above as "reference group". The reference group or population is generated by the administrator using data from patients of the clinic / hospital who have had their bone density assessed, against which the data of the patient for which a report is requested can be compared.

[0164] A "query" is a request submitted by a clinic / hospital regarding their patients.

[0165] A "report" is a bone health report prepared for delivery. These reports have a graphical appearance similar to the one depicted in Figure 3 .

[0166] The administrator has the following sections available: -- 1. User management, -- 2. Management of the reference population, -- 3. Management of requests.

[0167] The user has the following sections available: -- 1. User management (personal information, security options, password change), -- 2. API key management, -- 3. Credit management.

[0168] M Score Bridge The M-score bridge is a set of dedicated software that allows Q / R on the clinic / hospital PACS to identify and receive relevant tests, and anonymize patient data received from the M-score clinic portal of the M-score platform and de-anonymize report data.

[0169] M Score Clinic Portal The M-score clinic portal (local client application) is designed to reside on one or more clinic / hospital computers and is used by the workers to request bone health reports and download the results in PDF format when available. The application communicates with the M-score platform via an API (the API key obtained in the user section of the M-score platform must be entered into the M-score clinic portal application).

[0170] The application has 3 navigable sections: -- 1. API key configuration, -- 2. Search check, -- 3. Request list.

[0171] Request List The user can view a list of existing requests and their status. The data populating this section comes from the M-score platform application database, specifically from the requests and results tables, and is in an anonymized form. In addition, the first and last name pair obtained during the search phase and saved when the request was created is added.

[0172] If there are results related to the request, the “Generate” button is enabled to generate and save a bone health report in PDF format. If there are no results related to a specific request, the results for that specific request are “in progress”.

[0173] Results If the request has already been processed, the related bone health report can be downloaded. The report is transmitted from the M-score platform in JSON format, containing the anonymized data of the records in the results table. The M-score clinic portal is responsible for de-anonymizing the report using the M-score bridge and finally generating a PDF file containing the report, which the user can download through any electronic device.

Claims

1. A computer-implemented method for predicting and diagnosing osteoporosis based on lumbar magnetic resonance imaging technology, said method comprising the following steps: I) collecting recorded lumbar MRI outputs obtained from a magnetic resonance device and archiving to create a reference database by means of a medical workflow management system selected from RIS, PACS, DICOM and CIS, wherein said lumbar MRI outputs are classified into three different reference groups according to the following criteria: - Reference Group 1 (RG1) - MRI outputs of subjects aged 20-30 years, - Reference Group 2 (RG2) - MRI outputs of subjects aged over 75 years, - Reference Group 3 (RG3) - MRI outputs of subjects who have received a confirmed diagnosis of osteoporosis after detecting a fragility fracture within the lumbar region, II) calculating the M-score of each MRI output in step I) according to the following formula: M score = (SNR L1-L4 - SNR REF ) / SD REF (A) or M score = (SQR L1-L4 – SQR REF ) / SD REF (B) wherein: SNR L1-L4 represents a signal-to-noise ratio obtained by dividing the signal intensity within the vertebrae among the L1 to L4 vertebrae by the standard deviation of the noise within the sagittal Tl-weighted spin echo sequence of the MRI output, SNR REF represents the average SNR of the reference group to which the MRI output belongs, SQR L1-L4 a signal-cerebrospinal fluid ratio, representing, within the sagittal Tl-weighted spin echo sequence of the MRI output, a signal-cerebrospinal fluid ratio obtained by dividing the intra-vertebral signal intensity in the Ll to L4 vertebrae by the standard deviation of the cerebrospinal fluid, SQR REF SQR average value representing a reference group to which the MRI output belongs, and SD REF standard deviation representing a reference group to which the MRI output belongs, then calculating the M-score mean value of each reference group in step I), named M-score (RG1), M-score (RG2) and M-score (RG3) respectively, III) examining the recorded lumbar MRI output of the subject of interest SOI by analyzing the sagittal Tl-weighted spin echo sequence in the vertebrae from LI to L4, thereby obtaining a relevant SNR L1-L4 ("SOI-SNR L1-L4 ”) or SQR L1-L4 ("SOI-SQR L1-L4 ”), IV) calculating the M-score of the subject of interest with respect to each reference group in step I) according to the following formula: M score (1) = [(SOI - SNR L1-L4 ) - SNR REF (RG1)] / SD REF (RG1) M score (2) = [(SOI - SNR L1-L4 ) - SNR REF (RG2)] / SD REF (RG2) M score (3) = [(SOI - SNR L1-L4 ) - SNR REF (RG3)] / SD REF (RG3) or M score(1) = [(SOI-SQR) L1-L4 ) – SQR REF (RG1)] / SD REF (RG1) M score (2) = [(SOI - SQR L1-L4 ) - SQR REF (RG2)] / SD REF (RG2) M score (3) = [(SOI - SQR L1-L4 (RG3)) / SD REF (RG3) REF (RG3) and V) comparing the three M-scores obtained in step IV) with the M-score mean value of the corresponding reference group obtained from step II), thereby assessing the possible risk of suffering from osteoporosis or diagnosing osteoporosis in the subject of interest.

2. The method according to claim 1, wherein said steps I)-V) are implemented by means of a computer program comprising instructions which cause the computer to perform said steps.

3. The method according to claim 1 or 2, wherein said lumbar MRI outputs of step I) are obtained from the same MR device model.

4. The method according to any one of claims 1-3, wherein a minimum of 20 MRI outputs are collected and analyzed for each reference group.

5. The method according to claim 4, wherein 30 to 100 MRI outputs are collected and analyzed for each reference group.

6. The method of any one of claims 3-5, wherein, In step III), the MRI of the SOI is obtained from the same MR device model as the MRI outputs used in step I).

7. The method of claim 6, wherein, The MRI outputs of the SOI of step III) and the MRI outputs of step I) are obtained from exactly the same MR device.

8. The method according to any one of claims 1-7, wherein in step V): - the difference between M-score (1) and M-score (RG1) is an indicator of the status of the SOI with respect to the healthy population, and a difference lower than "+1" indicates a "healthy subject", while a difference equal to or higher than "+1" indicates a "non-healthy subject"; - the difference between M-score (2) and M-score (RG2) is an indicator of the status of the SOI with respect to the age-related risk population, and a difference higher than "-1" indicates an "OP patient subject", while a difference equal to or lower than "-1" indicates a "non-OP patient subject"; and - the difference between M-score (3) and M-score (RG3) is an indicator of the status of the SOI with respect to the osteoporosis population, and a difference lower than "-1" indicates an "OP patient subject", while a difference equal to or higher than "-1" indicates a "non-OP patient subject". - the difference between M-score (3) and M-score (RG3) is an indicator of the status of the SOI with respect to the population of OP patients, and a difference higher than "-1" indicates an "eOP patient subject", while a difference equal to or lower than "-1" indicates a "non-eOP patient subject".

9. The method according to claim 8, wherein step V) comprises a sub-step V-i) in which the difference between M-score (1) and M-score (RG1) and the difference between M-score (2) and M-score (RG2) are considered and compared, resulting in the following combination of three calculated indicators: 1) "healthy subject" and "non-OP patient subject", i.e. the SOI is characterized by satisfactory bone mineralization and shows no risk of suffering from osteoporosis; 2) "non-healthy subject" and "non-OP patient subject", i.e. the SOI is characterized by osteopenia, which can potentially predict suffering from osteoporosis; 3) "non-healthy subject" and "OP patient subject", i.e. the SOI is characterized by osteoporosis, which is also considered a predictor of fragility fractures.

10. The method according to claim 8 or 9, wherein step V) comprises a sub-step V-ii) in which the difference between M-score (1) and M-score (RG1) and the difference between M-score (3) and M-score (RG3) are considered and compared, resulting in the following combination of two calculated indicators: 1) "healthy subject" and "non-eOP patient subject", i.e. the SOI is characterized by satisfactory bone mineralization and shows no signs of osteoporosis => negative; 2) "non-healthy subject" and "eOP patient subject", i.e. the SOI is definitely affected by confirmed osteoporosis => positive.

11. The method according to claim 10, wherein step V) comprises both sub-step V-i) and sub-step V-ii), and in this order.

12. A computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of the method according to any one of claims 1-11.

13. A computer-readable data carrier having stored thereon the computer program according to claim 12.

14. A computer-implemented system for performing the method according to any one of claims 1-11, the computer-implemented system comprising: a) at least one digital repository in communication with at least one medical workflow management system of a clinic or hospital, the repository being for archiving recorded MRI outputs received from the at least one medical workflow management system, wherein, according to step I) of the method, the MRI outputs are anonymized via software before being archived in a reference database, b) a software platform operating on the reference database, the software platform comprising a computer program designed to perform the calculations, checks and comparisons of steps II) to V) of the method, and 15. A method for diagnosing osteoporosis in a subject, the method comprising the steps of: a) obtaining a set of MRI outputs of the subject, wherein the MRI outputs are obtained according to the method of any one of claims 1-11, b) determining the SOI of the subject according to the method of any one of claims 1-11, and c) diagnosing osteoporosis in the subject based on the SOI of the subject. c) a plurality of computing devices of users for accessing the software platform for sending queries and receiving reports, each user being associated with a digital identity comprising an authentication device, wherein the authentication device comprises a processor capable of automatically generating an authentication code in response to an interrogation for verifying the digital identity of the user.

15. The computer-implemented system of claim 14, the system being cloud-based.

Images