Prediction of candidates for spinal neuromodulation

Abstract

Described herein are various embodiments of systems and methods for determining the likelihood that a patient will respond favorably to a neuromodulation procedure based on a quantitative or objective score, or based on a determination based on multiple indicators of pain (e.g., chronic low back pain originating from one or more vertebral bodies or vertebral endplates of a patient). The systems and methods may include the application of artificial intelligence techniques (e.g., trained algorithms, machine learning or deep learning algorithms, and / or trained neural networks).

Description

Related Applications

[0001] This application claims priority to U.S. Provisional Application No. 63 / 129,374, filed December 22, 2020, the entire contents of which are incorporated herein by reference. [Technical Field]

[0002] Spinal neuromodulation procedures (e.g., intravertebral nerves, such as those innervating the vertebral endplates) Basilar nerve Described herein are various embodiments of systems and methods for identifying patients who may respond favorably to spinal neuromodulation treatment (e.g., ablation of the vertebral trunk or other intraosseous nerves) to prevent and / or treat back pain (e.g., chronic low back pain). The systems and methods described herein can incorporate artificial intelligence techniques (e.g., trained algorithms, machine learning or deep learning algorithms, or neural networks). Some embodiments include the use of a combination of indicators or factors (e.g., features identified from ultrashort echo time (“UTE”) magnetic resonance imaging, features identified from conventional T1- or T2-weighted magnetic resonance imaging, features identified from other imaging modalities, multifidus muscle features (e.g., atrophy), disc calcification, bone turnover, and / or other biomarkers) to identify patients who are likely to have back pain that would result in a favorable response to spinal neuromodulation treatment (e.g., back pain originating from one or more vertebral bodies or vertebral endplates). [Background technology]

[0003] Back pain is a very common health problem worldwide and a leading cause of work-related disability benefits and compensation. At any given time, back pain affects nearly 30% of the U.S. population and leads to 62 million annual visits to hospitals, emergency departments, outpatient clinics, and physicians' offices. Back pain can result from tense muscles, ligaments, or tendons in the back, and / or structural problems with the bones or intervertebral discs. Back pain can be acute or chronic. Existing treatments for chronic back pain vary widely and include physical therapy and exercise, chiropractic care, injections, rest, pharmacological therapies such as opioids, analgesics, or anti-inflammatory medications, and surgical interventions such as spinal fusion, discectomy (e.g., total disc replacement), or disc repair. Existing treatments are costly, addictive, temporary, potentially ineffective, and / or may increase pain or require long recovery times. Furthermore, existing treatments do not provide adequate relief for most patients, and only a small proportion are surgically eligible. Summary of the Invention

[0004] Applicant's existing technology (Intracept® System and Procedure, commercially available from Relievant Medsystems, Inc.) provides: Basilar nerve As disclosed herein, some embodiments provide a safe and effective minimally invasive procedure to target (e.g., ablate) nerves (and / or other intraosseous nerves or nerves innervating the vertebral endplates) whether a patient has back pain (e.g., chronic lower back pain) originating from one or more vertebral bodies or vertebral endplates, and therefore (e.g., as provided by Applicant's existing Intracept® treatment). Basilar nerve Systems and methods are provided for determining (eg, via automated computer-implemented methods) whether a subject is likely to respond favorably to ablation or other spinal neuromodulation procedures.

[0005] The determination may be based on a combination (e.g., a weighted combination) of indicators or factors. For example, the method may be based on a particular spinal neuromodulation procedure (e.g., intra-vertebral stimuli in one or more vertebral bodies of a patient). Basilar nerve The method may include generating (e.g., calculating, determining) an objective or quantitative score or other output (e.g., a percentage value, a number on a scale, a binary YES / NO output) based on the identification and analysis of indicators or combinations of factors that indicate the likelihood of a favorable response (e.g., pain prevention or pain relief) to ablation therapy. In other words, the method provides an objective or quantitative prediction or assessment (as opposed to a subjective or qualitative prediction or assessment) of whether a particular spinal neuromodulation procedure is likely to be successful in treating back pain (e.g., chronic low back pain whose source is in one or more vertebral bodies or vertebral endplates). A clinical professional can decide whether to perform a particular spinal neuromodulation procedure based on the objective or quantitative prediction or assessment (e.g., score). For example, if the objective or quantitative prediction or assessment is above a predetermined threshold, a treatment recommendation or treatment protocol can be provided, and the clinician can decide to provide or adjust treatment based on the recommendation or protocol.

[0006] According to some embodiments, the quantitative score, value, or other output may be generated based on the execution of one or more computer-implemented algorithms stored (e.g., on a non-transitory computer-readable storage medium) and executed by one or more processors (e.g., one or more hardware processors of a server). The quantitative score, value, or other output may be based on a combination (e.g., weighting) of multiple indicators. Some indicators may be weighted or deemed more important in quantitative prediction or assessment than other indicators. For example, the algorithm may be based on the effectiveness of a particular spinal neuromodulation treatment (e.g., through clinical studies or past experience). Basilar nerveA quantitative score, value, or other output may be generated based on the identification and analysis of one or more indices (e.g., higher tier or first tier indices) that are believed to have a stronger correlation with and / or be more reliable for predicting the type of back pain that will be successfully treated by ablation (e.g., ablation procedure). In some embodiments, vertebrae with a quantitative score (e.g., quantitative endplate score) above a threshold may be selected for treatment (e.g., ablation procedure). Basilar nerve The quantitative score may include a quantitative endplate score based on the severity, extent, and / or amount of the identified indicators (e.g., indicators of pain originating from one or more vertebral endplates).

[0007] Algorithms (e.g., program instructions stored on a non-transitory computer-readable storage medium and executed by one or more hardware processors) may also be used to implement certain spinal neuromodulation treatments (e.g., Basilar nerve The quantitative score or other output may be validated or provided with additional reliability based on the identification and analysis of one or more additional indicators (e.g., lower or second tier indicators) that may correlate with and / or be reliable for predicting the type of back pain (e.g., chronic low back pain) that will be successfully treated by a back pain ablation procedure. Validation or reliability checks may advantageously help reduce false positives or false negatives. Any one factor or indicator may not be completely reliable or accurate in predicting the likelihood of treatment success. Furthermore, making subjective predictions based on visualization and / or subjective feedback or pain scores from the patient alone may not be reliably accurate. Identifying indicators and / or generating objective scores based on trained algorithms trained on previous patient data or other normative data may generate more reliable objective scores and treatment recommendations or protocols, thereby resulting in greater patient satisfaction and reducing unnecessary treatments that are likely to be ineffective or providing more specifically tailored treatment protocols.

[0008] Certain spinal neuromodulation procedures (e.g., Basilar nerve The types of back pain that are preferably treated by ablation procedures include pain originating from one or more vertebral bodies or vertebral endplates (e.g., Basilar nerve The back pain may be pain originating from other intraosseous nerves within the trunk or vertebral body or nerves innervating the vertebral end plates. In one embodiment, the type of back pain desired to be treated is not discogenic back pain originating from one or more intervertebral discs. However, in some embodiments, discogenic back pain may also be treated even if it is not the focus or target of the treatment or procedure. The one or more indicators may include an indicator of one or more discogenic pain.

[0009] The algorithms may include the application of artificial intelligence techniques or trained algorithms (e.g., machine learning or deep learning models and algorithms implemented by a trained artificial neural network). Portions of the algorithms may be applied to a trained neural network to facilitate the identification of indicators and / or to facilitate the calculation of an objective score. The indicators or factors may be identified, for example, from various images obtained using one or more imaging modalities or techniques (e.g., magnetic resonance imaging ("MRI") images, such as conventional T1- and / or T2-weighted MRI images, fat-suppressed MRI images, ultrashort echo time ("UTE") MRI sequence images, iterative decomposition and least-squares estimation of water and fat with echo asymmetry ("IDEAL") MRI sequence images, fast spin-echo MRI sequence images, computed tomography ("CT") images, including single-photon emission computed tomography ("SPECT") images, positron emission tomography ("PET") bone images, x-ray images, fluoroscopy, and / or other imaging modalities or techniques).

[0010] The images can include images of a particular patient and can also include images of other patients or subjects (e.g., for comparison and / or training of neural networks for artificial intelligence implementations). The indicators or factors can include identifying one or more characteristics based on the images (e.g., the amount of atrophy of paraspinal muscles surrounding a particular vertebral body or the entire spine, vertebral endplate loss or degradation, bone marrow intensity changes such as Modic or pre-Modic change characteristics, vertebral fat fraction, changes in the water-to-fat ratio in the bone marrow, active bone turnover, disc calcification, and other tissue characteristics). Modic changes can include, for example, Modic type 1 changes or Modic type 2 changes. The one or more indicators or factors determined from the images can include edema, inflammation, and / or tissue changes (e.g., tissue lesions, fibrosis, fissures, or other changes in tissue type or characteristics) of the lining, contour, or profile of the bone, bone marrow, and / or endplates. Vertebral endplate defects can include, for example, focal defects, erosive defects, marginal defects, and corner defects of the vertebral endplate of a vertebral body. Indicators or factors can include identification of specific spinal anatomical characteristics or conditions (e.g., scoliosis, spondylolisthesis, herniated disc, joint dysfunction, spondylosis, osteoarthritis, spinal stenosis, kyphosis, spondylolisthesis, etc.).

[0011] The indicators or factors may also include evaluation of one or more biomarkers (e.g., biomarkers related to pain, inflammation, or neurotransmission). Biomarkers may also be used to assess whether a particular subject is likely to be a candidate for nerve ablation therapy for the treatment of back pain. For example, biomarkers may indicate pre-Modic changes or conditions (e.g., inflammation, edema, bone marrow lesions, or fibrosis) that are likely to lead to Modic changes or endplate damage. Evaluation of biomarker levels can indicate which vertebral bodies in a particular subject are candidates for treatment to prevent (or reduce the likelihood of) the onset or worsening of back pain or to treat existing back pain. Pre-treatment biomarker evaluation may also be combined with pre-treatment imaging. Biomarkers may include one or more of inflammatory cytokines (e.g., interleukins, interferons, tumor necrosis factors, prostaglandins, and chemokines), pain indicators (e.g., substance P, calcitonin gene-related peptide (CGRP)), edema factors, and / or other inflammatory factors. Biomarkers can be obtained, for example, from one or more serum samples (e.g., plasma). Biomarkers can be acquired over an extended period of time (e.g., over a period of days, weeks, or months) or at a single time instance. Biomarkers can also be identified in the image itself, and can be tissue characteristics, bone marrow intensity changes, etc., as described above.

[0012] Indicators or factors may also include patient parameters, information, or risk factors, such as age, sex, body mass index, bone mineral density measurements, back pain history, indication of previous spinal treatment (such as spinal fusion or discectomy), patient-reported outcome or quality of life measures, and / or candidate patient and / or treatment (e.g., Basilar nerve The present invention may also include other known risk factors for vertebral endplate degeneration or loss (such as smoking, occupational or recreational physical demands or conditions) in identifying candidate vertebral bodies for ablation.

[0013] According to some embodiments, certain subjects (e.g., humans) undergo a neuroablation procedure to treat back pain (e.g., an INTRACEPT® nerve ablation procedure performed using commercial technology from Relievant Medsystems, Inc.). Basilar nerve A method for quantitatively predicting the likelihood of a particular subject responding favorably to ablation is provided. The method includes identifying a plurality of indicators of back pain (e.g., chronic back pain) based on one or more images of at least a portion of the spine (e.g., the lumbosacral region of the spine) of a particular subject. The method further includes quantifying the identified plurality of indicators and, based on the quantification of the identified plurality of indicators, predicting that the particular subject will respond favorably to ablation. Basilar nerve and calculating an objective score indicative of the likelihood of responding favorably to ablation.The entire method or parts of the method may be fully computer implemented and automated.

[0014] The images may be obtained, for example, from one or more of the following imaging modalities: MRI imaging, T1-weighted MRI imaging, T2-weighted MRI imaging, fat-suppressed MRI imaging, UTE MRI sequence imaging, IDEAL MRI sequence imaging, fast spin-echo MRI sequence imaging, CT imaging, PET bone imaging, X-ray imaging, and fluoroscopy. The images may include images of a particular patient, or may also include images of other patients or subjects (e.g., for comparison and / or training of neural networks for artificial intelligence implementations).

[0015] Identifying the plurality of indicators includes identifying one or more bone marrow intensity changes and / or identifying one or more features of vertebral endplate defects or vertebral endplate degeneration. In some embodiments, each of the plurality of indicators may fall into only one of these two categories. For example, within an overall category or classification of bone marrow intensity changes or an overall category or classification of vertebral endplate degeneration or defects, there may be multiple identified indicators (e.g., different spatial locations, different types, different subgroups or subsets within the same category or classification). The plurality of indicators may be classified as both bone marrow intensity changes and features of vertebral endplate defects or vertebral endplate degeneration.

[0016] Identifying the one or more bone marrow intensity alterations may include identifying the one or more bone marrow intensity alterations as either a Type 1 Modic alteration or a Type 2 Modic alteration (or optionally a Type 3 Modified alteration). Identifying the one or more vertebral endplate defects or characteristics of vertebral endplate degeneration may include identifying irregularities or deviations from the normal continuous inner layer of the vertebral endplate, identifying deviations from the normal contour profile of the vertebral endplate, identifying changes in fat fraction, and / or identifying one or more phenotypic subtypes of vertebral endplate defects.

[0017] Quantifying the identified plurality of indicators may include determining the amount of bone marrow intensity alteration and / or vertebral endplate defect, determining the level of severity (e.g., spatial distribution, prevalence) of the bone marrow intensity alteration and / or vertebral endplate defect, and / or quantifying the identified fat fraction alteration.

[0018] The method may further include determining a confidence level for the objective score (e.g., Oswestry Disability Index score, visual analog pain score) based on one or more additional indicators of back pain, such as changes in multifidus muscle characteristics obtained for a particular subject, bone turnover identified in SPECT images, and / or pain scores. In other embodiments, these additional indicators are used in determining the objective score, but not in determining a separate confidence level for the objective score. The method may further include displaying an output of the objective score on a display. The output may be a numeric score on a scale, a binary YES or NO output, a percentage score, or the like.

[0019] According to some embodiments, certain subjects may be given Basilar nerve A method (e.g., a computer-implemented method executed by one or more hardware processors) for quantitatively predicting the likelihood of responding favorably to ablation includes receiving one or more images (e.g., MRI images) of at least a portion of a particular subject's spine; applying pre-processing imaging techniques to the one or more images; extracting features from the one or more images to identify a plurality of indicators of back pain; and, based on the extraction, determining whether the particular subject is Basilar nerve and determining an objective score indicative of the likelihood of responding favorably to ablation. The plurality of indicators may include (i) bone marrow intensity changes and / or (ii) features of vertebral endplate defects or vertebral endplate degeneration. Extracting features from the one or more images may include applying a trained neural network to the one or more images to automatically identify the plurality of indicators of back pain. Determining the objective score may also include applying the trained neural network to the extracted features. In some embodiments, the plurality of indicators may also include additional indicators in addition to features of bone marrow intensity changes and / or vertebral endplate defects or vertebral endplate degeneration.

[0020] The method may further include applying one or more rules to the extracted features to generate a confidence level. The one or more rules may be based on one or more additional indicators, such as those described herein. The additional indicator may be, for example, an indicator of the other of the two categories (either bone marrow intensity changes or vertebral endplate defects or vertebral endplate degeneration). Determining the objective score may include quantifying the multiple indicators. Quantification may be based on degree (e.g., amount of indicator, severity of indicator (e.g., size or volume) and / or spatial assessment (prevalence in different locations or regions of the vertebral body or endplate or other location). The method may further include displaying the output of the objective score on a display (e.g., a monitor of a desktop or portable computing device).

[0021] According to some embodiments, certain subjects may be given Basilar nerve A method (e.g., a computer-implemented method including stored program instructions executed by one or more hardware processors) for quantitatively predicting the likelihood of responding favorably to ablation of a particular subject includes receiving one or more magnetic resonance images (MRIs) of at least the lumbosacral region of the spine of the particular subject; applying pre-processing imaging techniques to the one or more MRIs to provide one or more uniform MRIs for feature detection; detecting features from the one or more MRIs to identify a plurality of indicators of chronic low back pain; quantifying the identified plurality of indicators based on the degree of the plurality of indicators, where the degree may include volume, severity, and / or spatial assessment; and determining, based on the quantification, that the particular subject will respond favorably to ablation. Basilar nerve and determining an objective score indicative of the likelihood of responding favorably to the ablation procedure. The multiple indices can include both bone marrow intensity changes and features of vertebral endplate defects or vertebral endplate degeneration, or multiple indices (e.g., subgroups or subsets) within only one of these categories or classifications.

[0022] According to some embodiments, a particular subject is successful Basilar nerveA computer-implemented method (e.g., execution of stored program instructions by one or more hardware processors) for training a neural network to determine whether a subject is a likely candidate for an ablation procedure includes collecting a set of digital images from a database. For example, each digital image may include a digital image of at least a portion of a spine of a subject having at least one indicator of back pain (e.g., chronic low back pain resulting from one or more vertebral endplates or vertebral bodies and / or resulting from one or more intervertebral discs). The method further includes applying one or more transforms to each digital image to create a rectified set of digital images. The method also includes creating a first training set including the set of collected digital images, the set of rectified digital images, and a set of digital images of at least a portion of a spine of one or more subjects without indicators of chronic low back pain (e.g., healthy subjects). The method further includes training a neural network in a first stage using the first training set, creating a second training set for a second stage of training that includes the first training set and digital images of at least a portion of the spine of one or more subjects without indicators of chronic low back pain who have been incorrectly determined to have at least one indicator of chronic low back pain, and training the neural network in the second stage using the second training set.

[0023] The digital image may include a magnetic resonance image, a computed tomography image, an X-ray image, or any other type of image described herein, In some embodiments, the digital image may alternatively be an analog image.

[0024] Applying one or more transformations may include preprocessing the set of acquired magnetic resonance images to make them more uniform for training. Preprocessing may include rotation, cropping, scaling, noise removal, segmentation, smoothing, contrast or color enhancement, and / or other image processing techniques. Preprocessing may also include spatial orientation identification, vertebral level identification, general anatomical feature identification, etc. In some embodiments, preprocessing may be performed by running the images through a pre-trained neural network that is trained to clean up, enhance, reconstruct, or improve the quality of images, such as noisy MRI images.

[0025] The method also includes, for at least a portion of the set of acquired magnetic resonance images: Basilar nerve The method may include identifying indicators of back pain that are likely to be successfully treated by an ablation procedure. Basilar nerve The step of identifying images from the set of acquired magnetic resonance images that were successfully treated with the ablation procedure can be included. In some embodiments, the set of acquired magnetic resonance images includes magnetic resonance images of subjects who have previously undergone spinal fusion or discectomy or other procedures that may have resulted in indicators or contributing factors to chronic low back pain (e.g., stimulated vertebral endplates or degeneration or loss of vertebral endplates, or bone marrow intensity changes or multifidus muscle atrophy).

[0026] The systems and methods described herein may also be applied to identifying pain other than back pain. For example, the systems and methods may be applied to peripheral neuralgia. According to some embodiments, a method for quantitatively predicting the likelihood that a particular subject will respond favorably or unfavorably to neuromodulation includes identifying a plurality of indicators of back and / or peripheral neuralgia, quantifying the identified plurality of indicators, and calculating an objective score indicative of the likelihood that the particular subject will respond favorably or unfavorably to the neuromodulation based on the quantification. Identifying a plurality of indicators may include identifying one or more bone marrow intensity changes and / or identifying one or more features of vertebral endplate defects or vertebral endplate degeneration. However, other indicators may alternatively or additionally be identified (e.g., for peripheral neuralgia or for back pain other than chronic low back pain).

[0027] The indices can be identified based on imaging data (e.g., magnetic resonance imaging data). The indices can be identified based on scanned data. The indices can be identified based on acoustic data. The data can be stored and retrieved from memory or received in real time from an imaging device (e.g., an MRI scanner). The indices can be identified automatically by computer processing techniques and algorithms (e.g., a trained algorithm or a trained neural network), or can be identified by a human and entered by a human using a user input device (e.g., a keyboard, a touchscreen graphical user interface, a computer mouse, a trackpad, etc.).

[0028] Neuromodulation can include denervation or neurostimulation. Basilar nerve , other intraosseous nerves, or peripheral nerve denervation or ablation.

[0029] In some embodiments, the adverse response may be due to the subject being unable to continue a particular treatment protocol (e.g., Basilar nerveIn some embodiments, a favorable response may qualify the subject for a treatment protocol.

[0030] The method may further include classifying a plurality of subjects based on the objective scores calculated for the subjects. The classifying step may include identifying subjects who are likely to have a successful outcome from a particular therapeutic treatment and subjects who are not likely to have a successful outcome based on the objective scores. Subject-specific data (e.g., age, lifestyle factors, pain scores, imaging) may be used to facilitate subject classification and to facilitate the recommendation of any treatment protocols. The method may also include recommending a treatment protocol based on the objective scores. The method may further include treating a particular subject (e.g., if the calculated objective score exceeds a predetermined threshold).

[0031] At least a portion of any of the methods described above or elsewhere herein may be performed by application of artificial intelligence techniques and techniques (e.g., trained machine learning or deep learning algorithms).

[0032] According to some embodiments, a particular subject Basilar nerve A method is provided for training a neural network that is used to determine whether a subject is a likely candidate for an ablation procedure.

[0033] The method may include pre-processing a plurality of MRI images of at least a portion of a spine of a plurality of patients to make the MRI images more uniform for training. The method may also include pre-processing a plurality of MRI images of at least a portion of a spine of a plurality of patients to make the MRI images more uniform for training. Basilar nerve The method may include identifying indicators of back pain that are likely to be successfully treated by an ablation procedure. Basilar nerveIn some embodiments, the method may include identifying images from the plurality of MRI images that were successfully treated by the ablation procedure. Basilar nerve This may include comparing images before and after the ablation procedure treatment.

[0034] According to some embodiments, certain subjects are Basilar nerve A system for quantitatively predicting likelihood of responding favorably to ablation, upon execution of instructions stored in a non-transitory computer-readable storage medium, includes receiving one or more images (e.g., MRI) of at least a portion of a spine (e.g., lumbosacral region) of a particular subject, applying a pre-processing imaging technique to the one or more images to provide one or more image uniformities for feature detection, detecting features from the one or more images to identify a plurality of indicators of back pain (e.g., chronic back pain), quantifying the identified plurality of indicators based on the severity of the plurality of indicators, and determining whether the particular subject is experiencing back pain (e.g., chronic back pain) based on the quantification of the plurality of indicators. Basilar nerve The server or computing system may include one or more hardware processors configured to determine an objective score indicative of the likelihood of responding favorably to the ablation procedure. The plurality of indicators may include, for example, bone marrow intensity changes and / or characteristics of vertebral endplate defects or vertebral endplate degeneration. The range of indicators may include a quantity, severity, and / or spatial assessment of the indicator.

[0035] In some embodiments, the one or more hardware processors are further configured to detect features from one or more images (e.g., MRIs) by applying the trained neural network to the one or more images to identify multiple indicators of back pain (e.g., chronic low back pain) and / or determine an objective score based on the identified multiple indicators.

[0036] In some embodiments, the system includes an imaging scanner or system (such as an MRI scanner) that can retrieve and store images.

[0037] When executed by one or more processors, the method includes receiving one or more images (e.g., MRI) of at least a portion of a spine of a particular subject; applying a pre-processing imaging technique to the one or more images; extracting features from the one or more images to identify a plurality of indicators of back pain (e.g., chronic low back pain); and, based on the extraction, determining whether the particular subject is suffering from back pain. Basilar nerve and determining an objective score indicative of likelihood of responding favorably to ablation, wherein the plurality of indicators include at least one of (i) bone marrow intensity changes and (ii) characteristics of vertebral endplate defects or vertebral endplate degeneration.

[0038] The process may further include quantifying the identified plurality of back pain indicators. Quantifying the identified plurality of indicators may include one or more of determining an amount of bone marrow intensity alteration and / or vertebral endplate defect, determining a level of severity of bone marrow intensity alteration and / or vertebral endplate defect, and quantifying the identified fat fraction alteration.

[0039] In some embodiments, extracting features comprises applying a trained neural network to the one or more images to automatically identify multiple indicators of back pain. In some embodiments, determining the objective score comprises applying a trained neural network to the one or more images to automatically calculate the objective score based on the extraction of the features.

[0040] According to some embodiments, a method for detecting and treating back pain in a subject includes identifying a candidate vertebral body for treatment based on a determination that the vertebral body exhibits one or more symptoms or defects associated with vertebral disc degeneration; and treating the identified candidate vertebral body by applying a thermal treatment dose of at least 240 cumulative equivalent minutes ("CEM") to a location within the vertebral body using a CEM model of 43 degrees Celsius or an equivalent thermal treatment dose using another model, such as the Arrhenius model. Basilar nerveThe one or more conditions associated with degeneration or loss of the vertebral endplate include pre-Modic change characteristics.

[0041] In some embodiments, the determination is based on images (e.g., MRI images, CT images, X-ray images, fluoroscopic images, ultrasound images) of the candidate vertebral body. In some aspects, the determination is based on obtaining biomarkers from the subject. Biomarkers can be obtained, for example, from one or more serum samples (e.g., plasma). Biomarkers can be obtained over an extended period of time (e.g., over a period of days, weeks, or months) or at a single instance in time.

[0042] According to some embodiments, target or candidate vertebrae for treatment can be identified prior to treatment. Target or candidate vertebrae can be identified based on identifying various types or factors associated with endplate degeneration and / or defects (e.g., focal defects, erosive defects, rim defects, and angle defects, all of which can be considered pre-Modic change characteristics). For example, one or more imaging modalities (e.g., MRI, CT, X-ray, fluoroscopic imaging) can be used to determine whether a vertebral body or vertebral endplate exhibits active Modic or "pre-Modic change" characteristics (e.g., features likely to result in Modic change, such as Type 1 Modic change, which includes evidence of inflammation and edema, or Type 2 Modic change, which includes bone marrow (e.g., fibrosis) changes and increased visceral fat content). For example, images obtained via MRI (e.g., IDEAL MRI) can be used to identify early signs or precursors of edema or inflammation in vertebral endplates (e.g., via application of one or more filters) prior to formal characterization or diagnosis as Type 1 Modic change. Examples of pre-Modic change characteristics can include mechanical characteristics (e.g., loss of soft nuclear material in the disc adjacent to the vertebral body, loss of disc height, loss of hydrostatic pressure, microfractures, focal endplate defects, erosive endplate defects, rim endplate defects, corner endplate defects, osteitis, spondylodiscitis, Schmorl's nodes) or bacterial characteristics (e.g., detection of bacteria entering the disc adjacent to the vertebral body, disc herniation or annular tears that may allow bacteria to enter the disc, inflammation or new capillary formation that may be caused by bacteria), or other pathogenic mechanisms that provide early signs or precursors of potential Modic change or vertebral endplate degeneration or loss.

[0043] Thus, vertebral bodies may be identified as potential targets for treatment before Modic changes occur (or before painful symptoms appear in the patient), allowing patients to be proactively treated before chronic back pain occurs, preventing or reducing the likelihood of chronic back pain. In this way, patients do not need to suffer from debilitating back pain for a period of time before treatment. Modic changes may or may not be correlated with endplate defects and may or may not be used to select or screen candidates. In some embodiments, Modic changes are not assessed, and only vertebral endplate degeneration and / or defects (e.g., changes characteristic of pre-Modic changes prior to the onset or ability to identify Modic changes) are identified. The rostral and / or caudal endplates may be assessed for pre-Modic changes (e.g., endplate defects that appear before Modic changes, which may affect the subchondral and vertebral bone marrow adjacent to the vertebral endplate).

[0044] In some embodiments, levels of one or more biomarkers (e.g., substance P, cytokines, high sensitivity C-reactive protein, or other compounds associated with inflammatory processes and / or pain and / or correlated with degeneration or loss of vertebral endplates (e.g., pre-Modic changes) or pathophysiological processes associated with Modic changes such as disc resorption, degradation and formation of types III and IV collagen, or myelofibrosis) are obtained from the patient (e.g., via a blood sample (e.g., serum) or cerebrospinal fluid sample) to determine whether the patient is Basilar nerveCandidates for ablation therapy can be determined, for example, whether one or more candidate vertebrae exhibit factors or symptoms associated with endplate degeneration or loss (e.g., pre-Modic alteration characteristics). Cytokine biomarker samples (e.g., pro-angiogenic serum cytokines such as vascular endothelial growth factor (VEGF)-C, VEGF-D, tyrosine-protein kinase receptor 2, VEGF receptor 1, intercellular adhesion molecule 1, and vascular cell adhesion molecule 1) can be obtained from multiple different discs or vertebral bodies or foramina of a patient and compared to each other to determine which vertebrae should be targeted for treatment. Other biomarkers, such as neoepitopes of type III and type IV procollagen (e.g., PRO-C3, PRO-C4) and type III and type IV collagen degradation neoepitopes (e.g., C3M, C4M), can also be evaluated.

[0045] In some embodiments, samples are obtained over a period of time and compared to determine changes in levels over time. For example, biomarkers can be measured weekly, bimonthly, monthly, quarterly, or every six months over a period of time and compared to analyze trends or changes over time. If significant changes are observed between biomarker levels (e.g., changes indicative of endplate degeneration or loss (e.g., characteristic of pre-Modic changes) or Modic changes as described above), treatment can be recommended and implemented to prevent or treat back pain. Biomarker levels (e.g., substance P, cytokine protein levels, PRO-C3, PRO-C4, C3M, C4M levels) can be measured using a variety of in vivo or in vitro kits, systems, and techniques (e.g., radioimmunoassay kits / methods, enzyme-linked immunosorbent assay kits, immunohistochemistry techniques, array-based systems, bioassay kits, in vivo injection of anti-cytokine immunoglobulins, multiplexed fluorescent microsphere immunoassays, homogeneous time-resolved fluorescent assays, bead-based technologies, interferometry, flow cytometry, etc.). Cytokine proteins can be measured directly or indirectly, such as by measuring mRNA transcripts.

[0046] Identifying pre-Modic alteration characteristics may include determining a quantitative or qualitative endplate score based on the severity, extent, and / or amount of the identified pre-Modic alteration characteristic (e.g., vertebral endplate defect), and vertebrae having a quantitative endplate score above a threshold are identified as potential candidates for treatment (e.g., Basilar nerve Pre-Modic alteration characteristics may be considered as a treatment (e.g., Basilar nerve In identifying candidate patients and / or vertebral bodies for endoscopic ablation (e.g., MRI), age, sex, body mass index, bone mineral density measurements, back pain history, and / or other known risk factors for vertebral endplate degeneration or loss (e.g., smoking, occupational or recreational physical demands or status), the risk factors may be combined with other known risk factors for vertebral endplate degeneration or loss (e.g., smoking, occupational or recreational physical demands or status) in identifying candidate patients and / or vertebral bodies for endoscopic ablation (e.g., MRI).

[0047] According to some embodiments, a method for detecting and treating back pain in a subject includes acquiring images of a vertebral body of the subject and analyzing the images to determine whether the vertebral body exhibits one or more symptoms associated with pre-Modic changes. The method also includes analyzing intraosseous nerves (e.g., Basilar nerve ), including adjusting (e.g., ablation, denervation, stimulation).

[0048] The images may be obtained using, for example, an MRI imaging modality, a CT imaging modality, an X-ray imaging modality, an ultrasound imaging modality, or fluoroscopy. The one or more symptoms associated with pre-Modic changes may include features likely to result in Modic changes (e.g., type 1 Modic changes, type 2 Modic changes). The one or more symptoms associated with pre-Modic changes may include early signs or precursors of edema or inflammation in the vertebral endplate prior to formal characterization or diagnosis as a Modic change. The one or more symptoms may include edema, inflammation, and / or tissue changes within the vertebral body or along a portion of the vertebral endplate of the vertebral body. The tissue changes may include tissue lesions or changes in the tissue type or characteristics of the endplate of the vertebral body, and / or tissue lesions or changes in the tissue type or characteristics of the bone marrow of the vertebral body. The one or more symptoms may include focal defects, erosive defects, marginal defects, and corner defects in the vertebral endplate of the vertebral body.

[0049] Spinal therapy procedures can include modulation of nerves within or around the bones of the spine (e.g., vertebral bodies). As used herein, the terms "modulation" or "neuromodulation" should be given their ordinary meaning and should also include ablation, permanent denervation, temporary denervation, destruction, blockage, inhibition, electroporation, therapeutic stimulation, diagnostic stimulation, blockage, necrosis, desensitization, or other effects on tissue. Neuromodulation refers to the modulation of nerves (structural and / or functional) and / or nerve conduction. Modulation is not necessarily limited to nerves, but can include effects on other tissues, such as tumors or other soft tissues.

[0050] The specific spinal neuromodulation procedure performed involves intravertebral Basilar nerve using a radiofrequency energy delivery device. Basilar nerveThe method can include applying energy (e.g., radiofrequency energy, ultrasound energy, microwave energy) to a target treatment area within the vertebral body sufficient to denervate (e.g., ablate, electroporate, dissociate, necrosing) the target treatment area. Denervation can alternatively or additionally include applying an ablation fluid (e.g., steam, chemical, cryoablation fluid) to the target treatment area within the vertebral body.

[0051] Any of the method steps described herein may be performed by one or more hardware processors (e.g., of a server) by executing program instructions stored on a non-transitory computer-readable medium.

[0052] Some embodiments of the present invention have one or more of the following advantages: (i) improved treatment accuracy; (ii) increased efficacy results; (iii) improved efficiency; (iv) increased patient satisfaction; (v) an increase in the number of people receiving a particular back pain treatment procedure who would not have been previously identified; and / or (vi) a reduction in treated patients who would not be successful with a particular back pain treatment procedure due to back pain from other causes.

[0053] For purposes of summarizing the disclosure, certain aspects, advantages, and novel features of embodiments of the disclosure are described herein. It is to be understood that not necessarily all such advantages may be achieved in accordance with any particular embodiment of the disclosure provided herein. Thus, the embodiments disclosed herein may be implemented or performed to achieve or optimize one advantage or advantages as taught or suggested herein, without necessarily achieving other advantages as taught or suggested herein.

[0054] While the methods summarized above and described in more detail below describe specific actions taken by a practitioner, it should be understood that they may also include direction of those actions by other parties. Thus, for example, an action such as "applying thermal energy" includes "directing the application of thermal energy." Further aspects of embodiments of the present disclosure are described below in this specification. With respect to the drawings, elements from one figure may be combined with elements from other figures. [Brief explanation of the drawings]

[0055] Some embodiments of the present disclosure may be more fully understood by reference to the following drawings, which are for illustrative purposes only.

[0056] [Figure 1] 1 illustrates an embodiment of a computing environment including a quantitative patient candidate diagnosis (QPCD) system that enables clinicians to quantitatively analyze patient candidates for basilar nerve ablation procedures.

[0057] [Figure 2] 1 illustrates an embodiment of a process for generating an objective or quantitative prediction of the likelihood that a potential patient will respond favorably to a basilar nerve ablation procedure or other treatment. [Figure 3] 1 illustrates an embodiment of a process for generating an objective or quantitative prediction of the likelihood that a potential patient will respond favorably to a basilar nerve ablation procedure or other treatment. [Figure 4] 1 illustrates an embodiment of a process for generating an objective or quantitative prediction of the likelihood that a potential patient will respond favorably to a basilar nerve ablation procedure or other treatment.

[0058] [Figure 5] 10 illustrates an example of a pre-processing and / or feature extraction step to facilitate identification and quantitative assessment of multiple indicators of back pain. [Figure 6A] 10 illustrates an example of a pre-processing and / or feature extraction step to facilitate identification and quantitative assessment of multiple indicators of back pain. [Figure 6B]10 illustrates an example of a pre-processing and / or feature extraction step to facilitate identification and quantitative assessment of multiple indicators of back pain. [Figure 6C] 10 illustrates an example of a pre-processing and / or feature extraction step to facilitate identification and quantitative assessment of multiple indicators of back pain. [Figure 6D] 10 illustrates an example of a pre-processing and / or feature extraction step to facilitate identification and quantitative assessment of multiple indicators of back pain. [Figure 7] 10 illustrates an example of a pre-processing and / or feature extraction step to facilitate identification and quantitative assessment of multiple indicators of back pain.

[0059] [Figure 8] 1 shows a schematic flow diagram of an embodiment for training a neural network and then using the neural network to perform quantitative predictions of the likelihood that a potential patient will respond favorably to a basilar nerve ablation procedure or other treatment. DETAILED DESCRIPTION OF THE INVENTION

[0060] Introduction Back pain (e.g., chronic lower back pain) can be caused by many causes, including vertebral endplate defects or degeneration, bone marrow strength changes such as Modic changes, ligament sprains, facet joint pain, muscle strain, muscle atrophy, spinal tendon injuries, spinal nerve compression, herniated discs, slipped discs, degenerative disc disease, sacroiliac joint dysfunction, bacterial or fungal infections, vertebral fractures, osteoporosis, and / or spinal tumors. It can be difficult for clinicians to reliably and accurately identify the exact cause of back pain by visually inspecting images obtained from one or more imaging modalities and / or by reviewing subjective patient pain scores (e.g., Oswestry Disability Index ("ODI") scores or visual analog score ("VAS") pain scores, quality of life measures, patient-reported outcome measures). As a result, patients with back pain (e.g., chronic lower back pain) may be treated with a particular treatment that does not successfully relieve the patient's back pain because the particular treatment does not effectively treat the actual cause of the back pain or does not treat all of the actual causes of the back pain.

[0061] For example, certain back pain treatment procedures are designed to treat back pain (e.g., chronic lower back pain) originating from one or more vertebral bodies or vertebral endplates. Basilar nerve It may be an ablation procedure, and the actual source of pain is Basilar nerve It may also be or include discogenic pain originating from one or more intervertebral discs that may not be effectively treated by an ablation procedure. As another example, a patient may receive a back pain treatment procedure intended to treat discogenic pain or pain originating from a source other than one or more vertebral bodies or vertebral endplates, when the actual source of the pain originates from one or more vertebral bodies or vertebral endplates. Thus, clinicians may perform ineffective or unsuccessful treatments, and patients may experience continued pain and reduced satisfaction, which may result in poor feedback or patient reviews for the particular clinician or hospital or treatment center or company providing the technology used in the treatment.

[0062] According to some embodiments, the systems and methods disclosed herein are suitable for use in certain back pain treatment procedures (e.g., Basilar nerve The systems and methods disclosed herein also provide more reliable prediction of specific causes or types of back pain (e.g., chronic low back pain) that may be effectively treated by ablation procedures (e.g., ablation procedures). The systems and methods disclosed herein also provide more reliable prediction of specific causes or types of back pain (e.g., chronic low back pain) that may not have been previously identified based on visualization of images by a clinician or subjective or qualitative factors or input from the patient. Basilar nerve Advantageously, the prediction can provide an increase in the number of patients identified as potential candidates for a particular back pain treatment procedure (e.g., ablation procedures). Basilar nerveThe present invention may include the generation (e.g., fully or partially automated automatic calculation) of an objective or quantitative score, value, or other output based on a combination (e.g., a weighted combination) of indicators of the causes of particular types of back pain that can be effectively treated by a specific back pain treatment procedure (e.g., an ablation procedure). Predictions based on multiple indicators can provide increased accuracy, reliability, and confidence, and reduce false positives and false negatives. Furthermore, the present invention may also include the generation (e.g., fully or partially automated automatic calculation) of an objective or quantitative score, value, or other output based on a combination (e.g., a weighted combination) of indicators of the causes of particular types of back pain that can be effectively treated by a specific back pain treatment procedure (e.g., an ablation procedure). Predictions based on multiple indicators can provide increased accuracy, reliability, and confidence, and reduce false positives and false negatives. Basilar nerve The systems and methods disclosed herein can advantageously increase the rate of successful treatment of certain back pain treatment procedures (e.g., ablation procedures), resulting in increased patient satisfaction and reduced costs. Basilar nerve The parameters (e.g., positioning, duration, targeting) of the ablation procedure may be adjusted or tailored to allow for more effective treatment of the actual cause of the back pain (e.g., chronic lower back pain).

[0063] Exemplary QPCD System FIG. 1 provides a clinician with access to a QPCD system 120 to identify patients who are likely to have pain originating from one or more vertebral bodies or vertebral endplates and therefore may be offered a spinal neuromodulation treatment (e.g., Intracept®, commercially available from Relievant Medsystems, Inc.) that targets that specific source of back pain (e.g., chronic low back pain). Basilar nerve Ablation procedures, etc. Basilar nerve 1 illustrates an embodiment of a computing environment 100 for determining patient candidates who are likely to respond favorably to a spinal neuromodulation procedure (e.g., an ablation procedure) based, at least in part, on analyzing a plurality of indicators identified from images of at least a portion of the patient's spine obtained using one or more imaging modalities (e.g., MRI, CT, SPECT, X-ray, etc.). Basilar nerveThe computing environment 100 may include a clinician system 108 that can access the QPCD system 120, and the clinician system may determine a patient's likelihood of back pain resulting from a particular cause (e.g., one or more vertebral endplates or vertebral bodies) and therefore a specific spinal neuromodulation procedure (e.g., ablation procedure). Basilar nerve The present invention may include one or more modules for determining the likelihood that a patient will respond favorably to a biopsy (ablation procedure).

[0064] The QPCD system 120 may include an image retriever module 122 that can retrieve images corresponding to scans of at least a portion of the spine (e.g., lumbar, sacral, thoracic, cervical, or a combination of two or more of these spinal regions) of a particular patient or multiple subjects. In one embodiment, the image retriever 122 can receive raw images directly from the imaging scanner 106 (e.g., an MRI scanner). In other embodiments, the image retriever 122 can receive images from a picture archiving and communication system (PACS) repository 102. The image retriever module 122 can also receive images from a storage medium, such as a compact disc (CD), a portable hard drive, cloud storage, a server, or other storage database or medium. The PACS repository 102 can store images, for example, in the Digital Imaging and Communication in Medicine (DICOM) format. The PACS repository 102 can also include other non-image data about the patient (e.g., age, gender, body mass index, bone mineral density measurements, pain scores, quality of life measurements, patient-reported outcomes, whether the patient received spinal neuromodulation therapy, and whether the treatment was successful). The image retriever module 122 can also receive images in different formats (e.g., jpeg, png, pdf, bmp, CT scanner raw files, MRI raw files, PET raw files, X-ray raw files, etc.). In one embodiment, the image retriever module 122 retrieves images from the PACS repository 102 or an imaging scanner wirelessly over the network 104. In another embodiment, the image retriever module 122 retrieves images via a local wired or integrated connection. The image retriever module 122 can receive images from the PACS repository 102 in response to input from the clinician system 108.

[0065] The QPCD system 120 may include an image processing module 124 for performing preprocessing and / or analysis (e.g., feature extraction or detection) of the images retrieved by the image retriever module 122. The image processing module 124 may process the images and identify one or more indicators of back pain originating from one or more vertebral bodies or vertebral endplates from the images, as described in more detail below. The indicators may include one or more of bone marrow intensity changes, vertebral endplate defects or degeneration, features of paraspinal musculature (e.g., multifidus muscular atrophy), bone turnover, disc calcification indicators, etc. The image processing module 124 may preprocess the received images (e.g., by rotating, resizing, contrast altering, image enhancement, or performing other image processing and cleanup techniques) to prepare the images for feature extraction or feature detection to identify indicators. The image processing module 124 may also perform feature extraction or feature detection to identify one or more indicators of back pain originating from one or more vertebral bodies or vertebral endplates from the images. Feature extraction may include identification (e.g., alphanumeric text labels) of each vertebral level shown in the image (as shown in FIG. 5). Image processing module 124 can use information obtained from one image to process other images of the same patient or future patients. Image processing module 124 may incorporate pre-trained neural networks to perform pre-processing and / or feature extraction on the images.

[0066] The QPCD system 120 may also quantify the multiple indicators identified by the image processing module 124 and, based on an objective or quantitative score, value, or other output, recommend a specific spinal neuromodulation treatment (e.g., Basilar nerve The present invention may include a quantifier / score calculator module 126 for generating an objective or quantitative score, value, or other output indicative of the likelihood that a patient will respond favorably to a particular spinal neuromodulation procedure (e.g., ablation procedure). In some embodiments, the quantifier / score calculator module 126 may estimate the likelihood of a patient responding favorably to a particular spinal neuromodulation procedure (e.g., ablation procedure) based on the objective or quantitative score or other value. Basilar nerveThe quantifier / score calculator module 126 may generate a Yes or No output or a binary output indicating a recommendation to proceed with an ablation procedure. The quantifier / score calculator module 126 may also include post-processing checks aimed at increasing the reliability of the score or value or binary Yes / No output (e.g., reducing false positives or false negatives). For example, an objective or quantitative score, value, or other output may be based on an analysis of a combination (e.g., a weighted combination) of one or more indicators (e.g., first-stage indicators such as bone marrow intensity changes and / or vertebral endplate loss or degeneration), and the post-processing checks may be based on an analysis of one or more additional indicators (e.g., second-stage indicators such as paraspinal muscle characteristics, bone turnover determined from SPECT imaging, etc.). The quantitative score or other value and / or binary output may be stored in the patient data repository 140 or the PACS repository 102 along with other patient data. The score or other quantitative value and / or binary output may also be transmitted to the clinician system 108 via a wired or wireless network. The quantifier / score calculator module 126 may also apply a pre-trained algorithm or neural network.

[0067] The image processing module 124 can store the analyzed image in the patient data repository 140 or send it back to the PACS repository 102. In some embodiments, the image processing module 124 can include internal checks to ensure that the image corresponds to a spine or portion of a spine. The user interface module 128 can interact with one or more other modules of the QPCD system 120 to generate one or more graphical user interfaces. In some embodiments, the graphical user interfaces can be one or more web pages or electronic documents. The user interface module 128 can also receive data, such as patient information, from the clinician system 108. In some cases, the user interface module 128 can receive commands from the clinician system 108 to initiate one or more functions of the QPCD system 120.

[0068] The QPCD system 120 can be implemented in computer hardware and / or software. The QPCD system 120 can run on one or more computing devices, such as one or more physical server computers. In implementations in which the QPCD system 120 is implemented on multiple servers, these servers can be co-located or geographically separated (e.g., separate data centers). Furthermore, the QPCD system 120 can be implemented on one or more virtual machines running on a physical server or group of servers. Furthermore, the QPCD system 120 can be hosted in a cloud computing environment, such as Amazon Web Services (AWS) Elastic Compute Cloud (EC2) or the Microsoft® Windows® Azure Platform. The QPCD system 120 can also integrate with the scanner 106 via a software or hardware plug-in or an API (application programming interface). In some embodiments, the clinician system 108 can implement some or all of the modules of the QPCD system 120. For example, the clinician system 108 can implement the user interface generation module 128, with the remaining modules implemented remotely on a server. In other embodiments, plug-ins to the QPCD system 120 may be installed in third-party tools. The QPCD system 200 may include multiple engines or modules for performing the processes and functions described herein, such as the modules described above. The engines or modules may include programmed instructions for performing the processes as described herein. The programming instructions may be stored in memory. The programming instructions may be implemented in C, C++, JAVA, or any other suitable programming language. In some embodiments, some or all of the portions of the QPCD system 120, including the engines or modules, may be implemented in application-specific circuits such as ASICs and FPGAs.Although shown as separate engines or modules, the functionality of the engines or modules described herein need not necessarily be separated.

[0069] The clinician system 108 can remotely access the QPCD system 120 on these servers via the network 104. The clinician system 108 can include thick client software or thin client software that can access the QPCD system 120 on one or more servers via the network 104. The network can be a local area network (LAN), a wide area network (WAN) such as the Internet, a combination thereof, or the like. For example, the network 104 can include a private intranet of a hospital or other institution, the public Internet, or a combination thereof. In some embodiments, the user software on the clinician system 108 can be browser software or other application software. The clinician system 108 can access the QPCD system 120 via the browser software or other application software.

[0070] In general, the clinician system 108 can include any type of computing device capable of running one or more applications and / or accessing network resources. For example, the clinician system 108 can be a desktop, laptop, netbook, tablet computer, smartphone, smartwatch, augmented reality wear, PDA (personal digital assistant), server, e-reader, video game platform, television set-top box (or simply a television with computing capabilities), kiosk, combinations thereof, etc. The clinician system 108 includes software and / or hardware for accessing the QPCD system 120, such as a browser or other client software.

[0071] Example of a quantitative forecasting process FIG. 2 illustrates a potential patient receiving a specific spinal neuromodulation procedure (e.g., Basilar nerve 1 illustrates an embodiment of a process 200 for generating an objective or quantitative prediction of the likelihood of responding favorably to a radiological ablation procedure (e.g., ablation treatment). The objective or quantitative prediction may be a numerical, graphical, or textual indicator (or a combination thereof). For example, the objective or quantitative prediction may include a percentage, a score on a scale, a binary Yes or No, and / or a color. The quantitative prediction process 200 may be performed by the QPCD system 120 described above. For illustrative purposes, the quantitative prediction process 200 is described as being performed by components of the computing environment 100 of FIG. 1. The entire process 200, or portions of the process 200, may be automated by one or more hardware processors executing stored program instructions stored on a non-transitory computer-readable medium.

[0072] The quantitative prediction process 200 begins at block 202 with receiving images of a patient candidate (e.g., from the PACS 102 or from the imaging scanner 106). The image retriever module 122 may receive image data corresponding to an MRI, CT, SPECT, PET, X-ray, or other imaging scan of at least a portion of the patient's spine. The MRI image data may include T1-weighted MRI images, T2-weighted MRI images, fat-suppressed MRI images, UTE MRI sequence images, IDEAL MRI sequence images, fast spin-echo MRI images, T1ρ-weighted images, and / or other MRI images obtained using other MRI sequences, pulses, weightings, or techniques. The received images may include one or more regions of the patient's spine (e.g., lumbar, sacral, thoracic, cervical, or a combination of two or more of these spinal regions). The images may include sequential images over a period of time or images at a single time point.

[0073] In block 204, the QPCD system 120 can analyze the received images to identify and quantify one or more indicators of back pain within or from the images (e.g., vertebral endplate loss or degeneration, bone marrow intensity changes, features of the paraspinal musculature (e.g., multifidus muscular atrophy), active bone turnover, indicators of disc calcification, vertebral fat fraction). The indicators of back pain can be indicators that correlate with back pain originating from one or more vertebral bodies or vertebral endplates and / or one or more adjacent discs. For example, the image processing module 124 can perform image processing techniques to automatically identify or detect one or more indicators (e.g., via feature extraction), and the quantifier / score calculator module 126 can analyze (e.g., quantify) the identified one or more indicators. The quantifier / score calculator module 126 then determines in block 206 whether the patient has pain originating from one or more vertebral bodies or vertebral endplates and is a candidate for spinal neuromodulation treatment (e.g., Basilar nerve An objective prediction (e.g., a quantitative score) of the likelihood of a patient responding favorably to ablation treatment (e.g., a pulmonary artery ablation procedure) can be generated.

[0074] In some embodiments, only indicators of back pain known to correlate with pain originating from one or more vertebral bodies or vertebral endplates are identified and assessed, and indicators of discogenic back pain (pain originating from the intervertebral disc) or other pain sources are not identified or assessed. Indicators may be identified or determined, and / or objective scores may be generated or calculated by application of a trained algorithm or trained neural network.

[0075] The QPCD system 120 can optionally generate a confidence level or perform an additional validation step in block 208 to reduce false positives or false negatives in the objective prediction (e.g., a quantitative score or a binary YES / NO output). The validation or confidence level generation step may include identifying and / or quantifying one or more additional indicators (e.g., indicators known to have a strong correlation or sensitivity to) of a particular back pain source (e.g., chronic low back pain resulting from one or more vertebral bodies or vertebral endplates) not used in the previous step. For example, while the multiple indicators identified and quantified in the previous step may include vertebral endplate loss or degeneration and / or bone marrow intensity changes, the one or more indicators used in the validation or confidence level generation step of block 208 may include features of the paraspinal musculature (e.g., multifidus muscle characteristics), active bone turnover, intervertebral disc calcification, or other indicators.

[0076] In some embodiments, the indices used in blocks 204 and 206 can be considered first-tier or more reliable / accurate indicators of a particular source of back pain, and the indices used in block 208 can be considered second-tier indicators correlating with a particular source of back pain. In other embodiments, the indices used to determine a quantitative score or other output may be better accepted by clinicians at that time as correlating with a particular cause of back pain (e.g., chronic low back pain resulting from one or more vertebral bodies or vertebral endplates). The indices identified and quantified in block 208 can be identified based on the same images as in blocks 204 and 206 or based on different images (e.g., SPECT images, CT images, different MRI images). In one embodiment, the images used in blocks 204 and 206 are only MRI images, but can comprise different types of MRI images (e.g., T1-weighted images, T2-weighted images, fat-suppressed images, UTE images, IDEAL images). In some embodiments, both first-tier and second-tier indices are used to determine a quantitative score or other output.

[0077] FIG. 3 illustrates a patient candidate receiving spinal neuromodulation treatment (e.g., Basilar nerve1 illustrates another embodiment of a process 300 for generating an objective or quantitative prediction of the likelihood of responding favorably to a radiological procedure (ablation procedure). Similar to quantitative prediction process 200, quantitative prediction process 300 may be performed by the QPCD system 120 described above. For illustrative purposes, quantitative prediction process 300 is described as being performed by components of computing environment 100 of FIG. 1. The entire process 300, or portions of process 300, may be automated by one or more hardware processors executing stored program instructions stored on a non-transitory computer-readable medium. Any of the steps of process 300 may include the application of a trained algorithm or a trained neural network.

[0078] In block 302, the QPCD system 120 (e.g., image retriever module 122) performs a spinal neuromodulation procedure (e.g., Basilar nerve Receive images of a patient candidate for a radiological ablation procedure. The images may correspond to MRI, CT, SPECT, PET, X-ray, or other imaging scans of the patient's spine. The MRI images may include T1-weighted MRI images, T2-weighted MRI images, fat-suppressed MRI images, UTE MRI images, and / or IDEAL MRI images. The received images may include one or more regions of the patient's spine (e.g., lumbar, sacral, thoracic, cervical, or a combination of two or more of these spinal regions). The images may include sequential images over a period of time or images at a single time point.

[0079] In block 304, the image processing module 124 may apply preprocessing to the image. Preprocessing may include analog or digital image processing techniques. Preprocessing may include rotation, cropping, zooming in, zooming out, noise removal, segmentation, smoothing, contrast or color enhancement, and / or other image processing techniques. Preprocessing may also include spatial orientation identification, vertebral level identification, general anatomical feature identification, etc. In some embodiments, preprocessing may be performed by running the image through a pre-trained neural network that is trained to clean up, enhance, reconstruct, or improve the quality of images, such as noisy MRI images.

[0080] In block 306, the image processing module 124 may perform feature extraction on the preprocessed images. Feature extraction, if not performed in preprocessing, may include spatial orientation identification, vertebral level identification, general anatomical feature identification, etc. Feature extraction may also include identification of indicators of back pain within the images (e.g., vertebral endplate defects or degeneration, bone marrow intensity changes, features of the paraspinal musculature (e.g., multifidus muscular atrophy), bone turnover, vertebral bone marrow fat fraction, intervertebral disc calcification, etc.).

[0081] The QPCD system 120 may then analyze the extracted features in block 308. The analysis may apply one or more rules to the extracted features to identify identified indicators of back pain and whether patients with the identified indicators are likely to be candidates for a particular spinal neuromodulation treatment (e.g., Basilar nerve The method may include assessing (e.g., quantifying) the likelihood of the patient responding favorably to ablation treatment (e.g., ablation procedure).

[0082] Analysis of vertebral endplate defects or degeneration can include spatial analysis and quantitative analysis. Spatial analysis can include, for example, identifying the location or positions along the vertebral endplate where the defect or degeneration occurs. Analysis of vertebral endplate defects or degeneration can include, for example, identifying various subclassifications of defects (e.g., focal defects, erosive defects, rim defects, corner defects), identifying defects relative to the normal continuous lining of the vertebral endplate, identifying irregularities in the endplate lining, assessing the volume or amount of the defect, assessing the extent or severity of the defect (e.g., width, depth, total area or volume, percentage of the total), assessing the contour profile of the vertebral endplate (e.g., jaggedness, depth), and identifying the defect as a specific phenotypic subtype of vertebral endplate defect. Contour profiles can be generated, for example, by hypo- and hyper-signal differentiation on T1-weighted or T2-weighted images.

[0083] Analysis of bone marrow intensity changes may include classification of changes as Type 1 or Type 2 Modic changes based on conventional Modic change classification schemes. Analysis of bone marrow intensity changes may also include spatial and / or extent or severity analysis of changes. For example, analysis may identify where bone marrow intensity changes occur within the vertebral body and / or the extent (height, volume, location) of bone marrow intensity changes. Ring-like nucleus border bone marrow intensity changes may be more significant than bone marrow intensity changes in the center of the vertebral body, for example, or vice versa. In some embodiments, Modic changes may be classified using T1-weighted, T2-weighted, or fat-suppressed MRI images. For example, Type 1 Modic changes may be identified as white swelling or inflammation on T2-weighted MRI images and reduced bright spots on T1-weighted MRI images. Type 2 Modic changes may be identified as bright spots on both T1- and T2-weighted MRI images. In some embodiments, analysis of bone marrow intensity changes may incorporate the use of a UTE MRI sequence or an IDEAL sequence.

[0084] In some embodiments, analysis of bone marrow intensity changes can include assessment of vertebral fat fraction. Vertebral fat fraction (e.g., water-to-fat conversion in bone marrow) can include analysis of IDEAL MRI images. Bone marrow intensity changes can be identified in both the vertebral body and one or more adjacent vertebral endplates. Bone marrow intensity changes can include, for example, bone marrow edema, bone marrow inflammation, bone marrow lesions, and / or conversion of normal red hematopoietic bone marrow to yellow fatty bone marrow, which can be identified from the received images.

[0085] Bone marrow intensity changes may also include pre-Modic change characteristics, which provide early signs or precursors of edema or inflammation at the vertebral endplate prior to formal characterization or diagnosis as Modic type 1. Examples of pre-Modic change characteristics include mechanical characteristics (e.g., loss of soft nuclear material in the disc adjacent to the vertebral body, loss of disc height, loss of hydrostatic pressure, microfractures, fissures, spondylodiscitis, Schmorl's nodes, osteitis) or bacterial characteristics (e.g., detection of bacteria entering the disc adjacent to the vertebral body, disc herniation or annular tears that may allow bacteria to enter the disc, inflammation or new capillary formation that may be caused by bacteria), or other pathogenic mechanisms that provide early signs or precursors of potential Modic change. The rostral and / or caudal endplates may be evaluated for pre-Modic changes (e.g., endplate defects that appear before Modic change, which may affect the subchondral and vertebral bone marrow adjacent to the vertebral endplate).

[0086] After analysis of the extracted features in block 308, the QPCD system 120 (e.g., quantifier / score calculator module 126) may determine whether the patient is eligible for spinal neuromodulation treatment (e.g., spinal cord neuromodulation treatment) based on the analysis of the extracted features, similar to that described in connection with block 206 of the quantitative prediction process 200. Basilar nerve An objective prediction (e.g., a quantitative score or other output) of the likelihood that a patient will respond favorably to a spinal neuromodulation procedure (e.g., a spinal cord ablation procedure) can be generated. Basilar nerveThe output may be a binary YES or NO output regarding whether the patient is likely to respond favorably to ablation (ablation procedure). The output may be based on analysis of only tier 1 indicators, or a combination (e.g., a weighted combination) of two, three, four, or more than four indicators, which may include both tier 1 and tier 2 indicators.

[0087] The quantitative prediction process 300 may optionally include post-processing refinement at block 312. Post-processing refinement may function as a check, for example, to reduce false positives or false negatives or to increase the confidence in the quantitative prediction. Post-processing refinement may include identifying and analyzing one or more additional indicators of back pain, as described in connection with block 208 of the quantitative prediction process 200. Post-processing refinement may provide an additional level of confidence in the determination at block 310. In some embodiments, post-processing refinement is not performed. For example, if the quantitative score or other value is above a certain predetermined threshold, post-processing refinement may not be performed, increasing processing time if post-processing refinement is not necessary or desired.

[0088] As described above in connection with block 208 of quantitative prediction process 200, additional indicators identified and analyzed in post-processing refinement in block 312 may include paraspinal musculature features (e.g., multifidus muscle atrophy) and may include image-based analysis of the cross-sectional area (diameter, size) of atrophy and / or analysis of the fat fraction (e.g., percentage or ratio) within the muscle tissue. Paraspinal musculature features may be identified, for example, in T1-weighted MRI images and / or T2-weighted fast spin-echo MRI images. Analysis and quantification of paraspinal musculature features may include spatial analysis (e.g., location or position of lipoatrophic changes in muscle composition). For example, lipoatrophic changes in muscle composition of paraspinal musculature (e.g., multifidus muscle tissue) may be identified as areas of hyperintensity medial and / or deep along the multifidus muscle fascia sheath. Analysis and quantification of the characteristics of the paraspinal musculature may include quantifying the degree or severity of changes in the characteristics of the musculature (e.g., the degree of fatty infiltration measured as a percentage of the total cross-sectional area of ​​the musculature).

[0089] Additional indicators identified and analyzed in post-processing refinement of block 312 may also include detection of active bone turnover (inflammatory response) based on SPECT images. For example, inflamed bone turns over faster than normal bone and can be identified and quantified. Patient candidates with vertebral bodies with active bone turnover may be more likely to undergo certain spinal neuromodulation treatments (e.g., Basilar nerve are more likely to respond favorably to treatment.

[0090] In some embodiments, the additional indices (e.g., second layer indices) can include one or more indices of discogenic pain resulting from the disc (e.g., disc calcification, disc biochemical composition (e.g., proteoglycan and collagen content) or morphology, annular tears, Pfirrman grade score, etc.). Such additional indices can be, for example, indices of pain resulting from a particular spinal neuromodulation treatment (e.g., Basilar nerve Treatment) may be used when the treatment is likely to be effective in treating discogenic back pain in addition to pain originating from one or more vertebral bodies or vertebral endplates. However, in some embodiments, indicators of discogenic pain (or at least only discogenic pain) are not identified or analyzed.

[0091] In some embodiments, the additional indices can include indices (e.g., biomarkers) that may not be discernible from images. Biomarkers can include, for example, substance P, cytokines, high-sensitivity C-reactive protein, or other compounds associated with inflammatory processes and / or pain and / or correlate with pathophysiological processes associated with Modic changes, such as degeneration or loss of vertebral endplates (e.g., pre-Modic changes) or Modic changes, such as disc resorption, degradation and formation of type III and type IV collagen, or myelofibrosis. Biomarkers can be obtained from the patient (e.g., via a blood (e.g., serum) or cerebrospinal fluid sample). Cytokine biomarker samples (e.g., pro-angiogenic serum cytokines such as vascular endothelial growth factor (VEGF)-C, VEGF-D, tyrosine-protein kinase receptor 2, VEGF receptor 1, intercellular adhesion molecule 1, vascular cell adhesion molecule 1, etc.) can be obtained from multiple different discs or vertebral bodies or foramina of a patient and compared to each other to determine which vertebral bodies to target for treatment. Other biomarkers, such as neoepitopes of type III and type IV procollagen (e.g., PRO-C3, PRO-C4) and type III and type IV collagen degradation neoepitopes (e.g., C3M, C4M), can also be assessed.

[0092] Biomarkers may include genetic markers, products of gene expression, autoantibodies, cytokines / growth factors, proteins or enzymes (such as heat shock proteins), and / or acute phase reactants. Biomarkers may include compounds that correlate with back pain, such as inflammatory cytokines, interleukin-1-beta (IL-1-beta), interleukin-1-alpha (IL-1-alpha), interleukin-6 (IL-6), IL-8, IL-10, IL-12, tumor necrosis factor-alpha (TNF-alpha), granulocyte-macrophage colony-stimulating factor (GM-CSF), interferon gamma (IFN-gamma), and prostaglandin E2 (PGE2). Biomarkers may also indicate the presence of tumor cells or tissue if tumor tissue is targeted by a specific treatment. Biomarkers may be found in serum / plasma, urine, synovial fluid, tissue biopsies, foramina, intervertebral disc, cerebrospinal fluid, or cells from blood, body fluids, lymph nodes, and / or tissues. In some embodiments, the biomarkers can be indicators identified from images.

[0093] FIG. 4 illustrates a patient candidate with back pain arising from one or more vertebral bodies or vertebral endplates and therefore a candidate for a particular spinal neuromodulation procedure (e.g., Basilar nerve4 illustrates one embodiment of a specific implementation of process 400 for generating an objective or quantitative prediction of the likelihood that a subject will respond favorably to a radiological ablation procedure (e.g., ablation treatment). The entire process 400, or portions of process 400, can be automated by one or more hardware processors executing stored program instructions stored on a non-transitory computer-readable medium. Any of the steps of process 400 can include the application of a trained algorithm or a trained neural network. The quantitative prediction process 400 first includes identifying vertebral endplate defects and / or degeneration in block 402. The quantitative prediction process 400 then includes identifying bone marrow intensity changes in block 404. It should be understood that these two steps may be performed in the opposite order. The identification steps in blocks 402 and 404 can be performed by the image processing module 124, for example, by applying preprocessing and feature extraction techniques such as those described above in connection with FIGS. 2 and 3. Turning to block 406, the quantitative prediction process 400 then includes analyzing the defects and / or changes identified in blocks 402 and 406. At block 408, the quantitative prediction process 400 determines whether a particular patient candidate is likely to respond to a particular spinal neuromodulation treatment (e.g., Basilar nerve The analysis and generation steps of blocks 406 and 408 may be performed by, for example, the quantifier / score calculator module 126, as described above in connection with FIGS.

[0094] Any of the quantitative prediction processes 200, 300, 400 can further include displaying a quantitative score, value, or other output (e.g., a binary YES / NO output) on a display (e.g., on the clinician system 108) for viewing by a clinician. The display of the output may be performed or executed by the user interface module 128 of the QPCD system 120. The clinician can decide whether to proceed with treatment for a particular patient based on the output. The treatment protocol can also be adjusted based on the output.

[0095] Figures 5, 6A-6D, and 7 illustrate examples of preprocessing and / or feature extraction steps that may be performed by the QPCD system 120 to facilitate the identification and quantitative assessment of multiple indicators of back pain. Figure 5 illustrates an example of vertebral level identification on an MRI image of a patient's lumbosacral region of the spine (identified L1-S2 levels). The identification may include alphanumeric text labels, as shown in Figure 5, for example. Figures 6A-6D illustrate examples of vertebral endplate defect or degeneration identification on various MRI images. White arrows overlaid on the image identify the vertebral endplate defect. Figure 6A illustrates a normal, healthy body; therefore, no indicators are identified. Figure 6B identifies a focal defect in the vertebral endplate. Figure 6C identifies an angular defect in the vertebral endplate. Figure 6D identifies an erosive defect in the vertebral endplate. Figure 7 illustrates an example of bone marrow intensity changes on an MRI image. Bone marrow intensity changes are identified by white arrows overlaid on the image. Bone marrow intensity changes can appear as hyperintense and / or hypointense tissue areas depending on the type of relaxation or MRI signal and sequencing used (e.g., T1-weighted or T2-weighted MRI signal).

[0096] Vertebral endplate defects and / or bone marrow intensity changes can be identified by the image processing module 124 of the QPCD system 120, as described above. For example, vertebral endplate defects and / or bone marrow intensity changes can be identified and extracted as features to be analyzed using image processing and feature extraction or feature detection techniques. Vertebral endplate defects and / or bone marrow intensity changes can be identified by, for example, pixel / voxel color value comparison techniques, pixel / voxel signal intensity comparison, cluster analysis techniques, image comparison techniques by comparing with images of normal healthy patients without back pain indicators, etc.

[0097] Training a neural network According to some embodiments, one or more steps of the processes described herein can be performed using machine learning techniques (e.g., using trained artificial neural networks that include deep learning algorithms). Machine learning or deep learning algorithms can be trained using supervised or unsupervised training. The processes disclosed herein can use machine learning modeling in conjunction with signal processing techniques, as described above, to analyze images to identify indicators of back pain and determine a quantitative prediction or score. The use of machine learning can advantageously increase the reliability or accuracy of predictions, and can be used to improve the accuracy of certain spinal neuromodulation treatments (e.g., Basilar nerve This can shorten the time to identify patients who are likely to respond favorably to spinal neuromodulation treatments (e.g., ablation procedures) and reduce false positive predictions based on human error. According to some embodiments, applying machine learning algorithms to a large number of images of healthy subjects without back pain and images of patients with back pain can predict the likelihood of receiving spinal neuromodulation treatments (e.g., ablation procedures). Basilar nerve This may enable reliably accurate and extremely rapid identification of potential patients who are likely to respond favorably to a particular quantitative prediction of ablation procedures.

[0098] Machine learning modeling and signal processing techniques include, but are not limited to, supervised and unsupervised algorithms for regression and classification. Specific classes of algorithms include, for example, artificial neural networks (perceptrons, backpropagation, convolutional neural networks (e.g., fast domain convolutional neural networks), recurrent neural networks, long-short-term memory networks, and deep belief networks), Bayesian (naive Bayes, multinomial Bayes, and Bayesian networks), clustering (k-means, expectation-maximization, and hierarchical clustering), ensemble methods (classification and regression tree variants and boosting), single or multiple linear regression, wavelet analysis, fast Fourier transform, instance-based (k-nearest neighbors, self-organizing maps, and support vector machines), regularization (elastic nets, ridge regression, and least absolute shrinkage and selection operators), and dimensionality reduction (principal component analysis variants, multidimensional scaling, discriminant analysis variants, and factor analysis). In some embodiments, any number of the aforementioned algorithms are not included. In some embodiments, the TensorFlow open-source software library can be used to implement machine learning algorithms. The neural network can be trained, stored, and implemented on a QPCD system 120, such as an image processing module 124 and / or a quantifier / score calculator module 126.

[0099] 8 shows a schematic flow diagram of one embodiment of training a neural network for use in then using the neural network in performing one or more of the process steps described herein (e.g., identifying and quantifying metrics and determining a quantitative score or other output). The neural network can be trained using spine images of hundreds or thousands of subjects. The images may be from a database of stored images accessible by QPCD system 120 via network 104. The spine images may include images of all or a portion of the spinal anatomy (e.g., one or more regions of the spine or spine, such as the lumbosacral region).

[0100] Spinal imaging can be used to visually or manually identify indicators of back pain (e.g., specific causes or types of back pain, such as chronic low back pain) and to identify specific spinal neuromodulation procedures (e.g., the INTRACEPT® procedure commercially offered by Relievant Medsystems, Inc.). Basilar nerve The spine images for training may also include images from previous patients who have been successfully or unsuccessfully treated with ablation procedures. Basilar nerve The images may include images from patients treated with spinal procedures for the treatment of back pain other than ablation procedures (e.g., fusion, vertebral tumor resection, vertebral fracture repair, discectomy, or diskectomy). In some examples, these other spinal procedures may also include stimulation of vertebral endplates, which may result in biomarkers, such as the biomarkers or indicators described herein, or other indicators of back pain (e.g., chronic low back pain). The images may also include images from healthy (e.g., initial) subjects without identified indicators of back pain (e.g., a specific cause or type of back pain). In some embodiments, the images are MRI images (e.g., T1-weighted MRI images, T2-weighted MRI images, fat-suppressed MRI images, UTE MRI images, IDEAL MRI images). In some embodiments, the images may also include images acquired by other modalities (e.g., CT, SPECT, PET, X-ray, and / or others). The images for each subject may include sequential images over a period of time or images at a single time point. Training may be performed on subjects undergoing a spinal procedure (e.g., MRI images, T2-weighted MRI images, fat-suppressed MRI images, UTE MRI images, IDEAL MRI images) to provide training on variables that may change before and after treatment. Basilar nerve This may include comparing images of the patient taken before and after the ablation procedure (before and after).

[0101] Training can include applying preprocessing techniques to images to facilitate feature extraction or detection. For example, MRI images may be grainy, noisy, or blurry, at least in some portions (e.g., due to artifacts caused by patient motion or metallic elements, differences in setup parameters within the MRI sequence, differences in Tesla magnetic field strength, insufficient spatial resolution or image contrast, insufficient signal-to-noise or contrast-to-noise ratios, improper signal weighting, truncation artifacts, aliasing, chemical shift artifacts, crosstalk, etc.). Preprocessing techniques can include, for example, rotation, alignment, resizing, cropping, denoising (e.g., removing artifacts, noise, and grain), segmentation, smoothing, contrast or color enhancement, making intensity levels more uniform or consistent, applying filters, cleanup, image reconstruction, and / or other image processing techniques. Rotation and alignment can be performed on MRI images because images can depend on the patient's orientation within the MRI machine, as well as other factors. Resizing may be necessary to zoom in on areas of the image where features are most likely to occur and crop areas of the image that are not related to features. Preprocessing may also include dividing the images into a grid of uniform nodes or regions among each training image, which may be numbered to facilitate image feature extraction and comparison. Preprocessing may also include spatial orientation identification, vertebral level identification, general anatomical feature identification, etc. According to some embodiments, preprocessing techniques advantageously result in more uniform images to improve neural network training speed and accuracy.

[0102] In some embodiments, preprocessing may be directed only to portions of the image deemed to be of interest (e.g., portions of the vertebral anatomy likely to exhibit indicators of back pain that may be effectively treated by a particular spinal neuromodulation procedure). According to some embodiments, if preprocessing is not performed on an image (e.g., an MRI image), the output may be less accurate due to less than ideal feature extraction or detection.

[0103] Training can further include performing automated feature extraction or detection techniques. Training can include performing an object detection task to recognize objects and an object localization task to estimate the coordinates of a bounding box in which the object is located within the image. For example, feature extraction can include pixel / voxel color value comparison techniques, pixel / voxel signal intensity comparison techniques, variance analysis techniques, cluster analysis techniques, image comparison techniques by comparing with images of normal, healthy patients without back pain indicators, and / or other feature detection techniques. In some embodiments, feature extraction or detection can be performed partially or fully manually by one or more users (e.g., using a pen mouse or other user interface or user input tool to draw the boundaries of bounding boxes around specific features or labeling features within the image). In some embodiments, the training images can be provided with annotation data or tags (e.g., a Comma-Separated Values ​​(CSV) file) with information regarding vertebral level identification, the presence of back pain indicators (e.g., vertebral endplate defect or degeneration, bone marrow intensity changes, or other indicators described herein), indicator location, indicator orientation, indicator severity, patient-reported outcomes before and after treatment (e.g., VAS score, ODI score, quality of life measure such as QoL or EQ score, patient-reported outcome measure, etc.). In some embodiments, the annotation data can include tags that identify what the output should be for that particular image (e.g., a quantitative or objective score, value, or other output indicating whether a particular spinal neuromodulation procedure is likely to be successful). The annotation data can also include tags that identify a binary classification output of YES or NO regarding whether a particular spinal neuromodulation procedure was effective or successful for the patient associated with the image. The annotation data may be provided by multiple clinicians to generate more reliable scores.

[0104] Unsupervised neural networks can be used to identify patterns and classify or extract features. For example, neural networks can use classification algorithms, including clustering (k-means, expectation-maximization, and hierarchical clustering), ensemble methods (variations of classification and regression trees and boosting), instance-based methods (k-nearest neighbors, self-organizing maps, and support vector machines), regularization (elastic nets, ridge regression, and least absolute shrinkage and selection operators), and dimensionality reduction (variants of principal component analysis, multidimensional scaling, discriminant analysis, and factor analysis) to classify or extract features that may correlate with indicators of back pain (e.g., specific types or causes of back pain). Neural networks can also use TensorFlow software code modules. While primarily described in the context of back pain (e.g., chronic low back pain), the neural network training and quantitative prediction techniques described herein can also be applied to other types of back pain (e.g., middle or upper back pain), neck pain, shoulder pain, and peripheral nerve pain (e.g., pain in the wrist, arm, elbow, leg, knee, or ankle). The images processed include an image of each anatomical part, and landmarks corresponding to each bone involved are identified.

[0105] Spinal Neuromodulation Procedures Any of the processes described herein may also be used in conjunction with certain spinal neuromodulation treatments (e.g., Basilar nerve This may include treating the patient by performing certain spinal neuromodulation procedures (e.g., Basilar nerveThe treatment device (e.g., treatment probe) used to perform an ablation procedure can be any device capable of modifying tissue (e.g., nerve, tumor, bone tissue). Any energy delivery device capable of delivering energy can be used (e.g., radiofrequency energy delivery device, microwave energy delivery device, laser device, infrared energy device, resistive heating device, other electromagnetic energy delivery device, ultrasound energy delivery device, etc.). The treatment device can be an RF energy delivery device. The RF energy delivery device can include a bipolar electrode pair at the distal end of the device. The electrodes of the bipolar pair can include an active tip electrode and a return ring electrode spaced apart from the active tip electrode. The RF energy delivery device can include one or more temperature sensors (e.g., thermocouples, thermistors) disposed on the exterior surface of or embedded within the shaft of the energy delivery device. According to some embodiments, the RF energy delivery device may not use internal circulatory cooling.

[0106] In some embodiments, a water jet cutting device can be used to modulate (e.g., denervate) a nerve. In some implementations, a chemical neuromodulation tool injected into a vertebral body or endplate can be used to ablate or otherwise modulate a nerve or other tissue. For example, a chemical neuromodulation tool can be configured to selectively bind to a nerve or endplate. In some embodiments, a local anesthetic (e.g., a liposomal local anesthetic) can be used inside or outside a vertebral body or other bone to denervate or block a nerve. In some implementations, brachytherapy can be used to place a radioactive material or implant within a vertebral body to deliver radiation therapy sufficient to ablate or denervate the vertebral body. Phototherapy can be used to ablate or otherwise modulate a nerve after a chemical or targeted agent has bound to a specific nerve or vertebral endplate.

[0107] According to some embodiments, thermal energy may be applied within the cancellous bone portion of the vertebral body (e.g., by one or more radio frequency (RF) energy delivery devices coupled to one or more RF generators). The thermal energy may be conducted by heat transfer to the surrounding cancellous bone, thereby heating the cancellous bone portion. According to some embodiments, the thermal energy is applied within a specific frequency range and extends through the cancellous bone of the vertebral body. Basilar nerve The temperature and duration of the treatment is sufficient to heat the cancellous bone such that tissue necrosis is regulated. In some embodiments, the regulation comprises permanent ablation or denervation or cell perforation (e.g., electroporation). In some embodiments, the regulation comprises temporary denervation or inhibition. In some embodiments, the regulation comprises stimulation or denervation without tissue necrosis.

[0108] In the case of thermal energy, the temperature of the thermal energy can be in the range of about 60 to about 115 degrees Celsius (e.g., about 60 to about 80°C, about 70 to about 90°C, about 75 to about 90°C, about 65 to about 75°C, about 68 to about 78°C, about 83 to about 87°C, about 80 to about 100°C, about 85 to about 95°C, about 90 to about 110°C, about 95 to about 115°C, about 70 to about 115°C, or overlapping ranges thereof). The temperature gradient can be in the range of 0.1 to 5 degrees Celsius per second (e.g., 0.1 to 1.0°C / second, 0.25 to 2.5°C / second, 0.5 to 2.0°C / second, 1.0 to 3.0°C / second, 1.5 to 4.0°C / second, or 2.0 to 5.0°C / second). Treatment times can range from about 10 seconds to about 1 hour (e.g., 10 seconds to 1 minute, 1 minute to 5 minutes, 5 minutes to 10 minutes, 5 minutes to 20 minutes, 8 minutes to 15 minutes, 10 minutes to 20 minutes, 15 minutes to 30 minutes, 20 minutes to 40 minutes, 30 minutes to 1 hour, 45 minutes to 1 hour, or overlapping ranges thereof). Pulsed energy may be delivered as an alternative to continuous energy, or sequentially with continuous energy. For radiofrequency energy, the applied energy can range from 350 kHz to 650 kHz (e.g., 400 kHz to 600 kHz, 350 kHz to 500 kHz, 450 kHz to 550 kHz, 500 kHz to 650 kHz, overlapping ranges thereof, or any value within a recited range, such as 450 kHz ± 5 kHz, 475 kHz ± 5 kHz, 487 kHz ± 5 kHz, etc.). The radio frequency energy output can be in the range of 5W to 100W (e.g., 5W to 15W, 5W to 20W, 5W to 30W, 8W to 12W, 10W to 25W, 15W to 25W, 20W to 30W, 8W to 24W, 5W to 50W, 10W to 20W, 20W to 50W, 25W to 75W, 50W to 100W, and overlapping ranges thereof, or any value within the recited range).

[0109] According to some embodiments, the thermal treatment dose (e.g., using a 43°C thermal dose calculation metric model in cumulative equivalent units (CEM)) is between 200 and 300 CEM (e.g., 200-240 CEM, 230 CEM-260 CEM, 240 CEM-280 CEM, 235 CEM-245 CEM, 260 CEM-300 CEM), or greater than a predetermined threshold (e.g., greater than 240 CEM), or is the thermal treatment dose equivalent using the Arrhenius model. The CEM number can represent the average thermal cumulative dose value at the target treatment region or location, and can represent a number that represents the desired dose for a particular biological endpoint. Thermal damage can occur via necrosis or apoptosis.

[0110] Cooling may optionally be provided to prevent heating of surrounding tissue during neuromodulation procedures. Cooling fluid may be circulated internally through the delivery device from and to a fluid reservoir in a closed-circuit manner (e.g., using inflow and outflow lumens). The cooling fluid may include pure water or saline having a temperature sufficient to cool the electrodes (e.g., 2-70°C, 2-10°C, 5-10°C, 5-15°C, 20-50°C, 40-70°C, overlapping ranges thereof, or any value within the recited ranges). Cooling may be provided by the same device (e.g., thermal) or a separate device used to deliver thermal energy. In some implementations, cooling is delivered to the region (e.g., cooling fluid exits the fluid delivery device). According to some embodiments, no cooling is used.

[0111] In some embodiments, ablative cooling can be applied to nerve or bone tissue instead of heat (e.g., for cryoneurolysis or cryoablation applications). The temperature and duration of cooling can be sufficient to modulate intraosseous nerves (e.g., ablation or localized freezing due to excessive cooling). Low temperature can destroy the myelin capsule or sheath surrounding the nerve. Low temperature can also advantageously reduce the sensation of pain. Cooling can be delivered using a hollow needle under fluoroscopy or other imaging modality.

[0112] In some embodiments, one or more fluids or agents can be delivered to the target treatment site to modulate the nerve. The agent can include, for example, a bone morphogenetic protein. In some embodiments, the fluid or agent can include a chemical agent for modulating the nerve (e.g., a chemical disruptant, alcohol, phenol, a nerve inhibitor, or a nerve stimulant). The fluid or agent can be delivered using a hollow needle or injection device under fluoroscopy or other imaging modality. While spinal neuromodulation procedures are specifically discussed herein, other neuromodulation procedures (e.g., peripheral neuromodulation procedures) can also be performed.

[0113] term In some implementations, the system includes various features that exist as a single feature (rather than multiple features). For example, in one embodiment, the system includes a single radiofrequency generator, a single introducer cannula with a single stylet, a single radiofrequency energy delivery device or probe, and a single bipolar electrode pair. A single thermocouple (or other means for measuring temperature) may also be included. In alternative embodiments, multiple features or components are provided.

[0114] In some embodiments, the system includes one or more of: a means for quantitatively predicting a scored indicator that indicates the likelihood that a patient will respond favorably to treatment; a means for tissue modulation (e.g., an ablation or other type of modulation catheter or delivery device); a means for imaging (e.g., MRI, CT, fluoroscopy); a means for access (e.g., an introducer assembly, curved cannula, drill, curette); and the like.

[0115] The terms used herein are for the purpose of describing particular embodiments only and are not intended to be limiting of the present invention. For example, as used herein, the singular forms "a," "an," and "the" are intended to include the plural forms unless the context clearly dictates otherwise. It can be further understood that the terms "comprises" and / or "comprising," as used herein, specify the presence of stated features, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, elements, components, and / or groups thereof. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items and can be abbreviated as " / ."

[0116] Spatially relative terms such as "under," "below," "lower," "over," and "upper" may be used herein for ease of description to describe the relationship of one element or feature to another, as shown in the figures. It will be understood that the spatially relative terms are intended to encompass different orientations of the device in use or operation in addition to the orientation depicted in the figures. For example, if the device in the figures is inverted, elements described as being "below" or "beneath" another element or feature would then be oriented "above" the other element or feature. Thus, the exemplary term "under" can encompass both an orientation of above and below. The device may be otherwise oriented (rotated 90 degrees or at other orientations), and the spatially relative descriptors used herein would be interpreted accordingly.

[0117] Although the terms "first" and "second" may be used herein to describe various features / elements (including steps), these features / elements should not be limited by these terms unless the context indicates otherwise. These terms may be used to distinguish one feature / element from another. Thus, a first feature / element described below could be referred to as a second feature / element, and similarly, a second feature / element described below could be referred to as a first feature / element without departing from the teachings of the present invention.

[0118] Throughout this specification and the claims that follow, unless the context requires otherwise, the word "comprise," and variations such as "comprises" and "comprising," mean that various components may be used jointly in methods and articles (e.g., compositions and apparatuses that include devices and methods). For example, the term "comprising" is understood to imply the inclusion of any stated element or step, but not the exclusion of any other elements or steps.

[0119] As used in this specification and claims, including when used in the examples, unless expressly specified otherwise, all numbers can be read as if preceded by the word "about" or "approximately," even if the term does not explicitly appear. The phrase "about" or "approximately" can be used when describing a magnitude and / or location to indicate that the stated value and / or location is within a reasonably expected range of value and / or location. For example, a numerical value can have a value of + / - 0.1% of the stated value (or range of values), + / - 1% of the stated value (or range of values), + / - 2% of the stated value (or range of values), + / - 5% of the stated value (or range of values), + / - 10% of the stated value (or range of values), etc. Any numerical value set forth herein should also be understood to include approximately that value unless the context dictates otherwise. For example, if the value "70" is disclosed, then "about 70" is also disclosed. Any numerical range recited herein is intended to include all subranges subsumed therein. As will be appreciated by those of ordinary skill in the art, when a value is disclosed, it is understood that "less than or equal to" that value, "greater than or equal to" that value, and possible ranges between those values ​​are also disclosed. For example, if a value "X" is disclosed, "less than or equal to X" and "greater than or equal to X" (e.g., where X is a number) are also disclosed. It is also understood that throughout this application, data is provided in several different formats, and this data represents endpoints and starting points, as well as ranges for any combination of the data points. For example, when a specific data point "10" and a specific data point "15" are disclosed, it is understood that greater than, greater than, less than, less than, and equal to 10 and 15, as well as between 10 and 15, are considered to be disclosed. It is also understood that each unit between two specified units is disclosed. For example, if 10 and 15 are disclosed, then 11, 12, 13, and 14 are also disclosed.

[0120] While various exemplary embodiments have been described above, any of several modifications can be made to the various embodiments without departing from the scope of the invention as set forth in the claims. For example, the order in which various described method steps are performed may often be changed in alternative embodiments, and in other alternative embodiments, one or more method steps may be skipped entirely. Optional features of the various apparatus and system embodiments may be included in some embodiments and not in other embodiments. Accordingly, the foregoing description has been provided primarily for illustrative purposes and should not be construed as limiting the scope of the invention as set forth in the claims.

[0121] The examples and illustrations contained herein indicate, by way of illustration and not limitation, specific embodiments in which the subject matter may be practiced. As noted above, other embodiments may be utilized and derived therefrom, such that structural and logical substitutions and changes may be made without departing from the scope of the present disclosure. Such embodiments of the inventive subject matter may be individually or collectively referred to herein by the term "invention," where more than one is actually disclosed, merely for convenience and without any intention to intentionally limit the scope of the present application to any single invention or inventive concept. Thus, although specific embodiments have been illustrated and described herein, any configuration calculated to achieve the same purpose may be substituted for the specific embodiment illustrated. The present disclosure is intended to cover any and all adaptations or modifications of various embodiments. Section headings used herein are provided merely to enhance readability and are not intended to limit the scope of embodiments disclosed in a particular section to the features or elements disclosed in that section. Combinations of the above embodiments, as well as other embodiments not specifically described herein, may be apparent to those skilled in the art upon reviewing the above description. The methods disclosed herein include specific actions performed by a practitioner. However, they may also include, explicitly or implicitly, any third-party instructions for their operation. The term "embodiment" should not be limited to interpretation as an "invention" and may mean a non-limiting example, implementation, or aspect.

Claims

1. 1. A computer-implemented method, implemented by one or more processors, for a medical image analysis system that quantitatively predicts the likelihood that a particular subject will respond favorably to basilar nerve ablation for treating back pain, comprising: the one or more processors automatically identifying a plurality of indicators of back pain based on an automated analysis of one or more magnetic resonance images (MRIs) of at least a portion of the particular subject's spine, the step including at least the steps of: (i) automatically identifying one or more bone marrow intensity changes; and (ii) automatically identifying one or more features of vertebral endplate defects or vertebral endplate degeneration. automatically quantifying the identified metrics; calculating, by the one or more processors, an objective score indicative of the likelihood that the particular subject will respond favorably to basilar nerve ablation based on the automated quantification; A method comprising:

2. The step of automatically quantifying the identified plurality of indicators comprises: determining the amount of bone marrow intensity change and / or vertebral endplate defect; determining the level of the bone marrow intensity change and / or the extent of the vertebral endplate defect; and Quantifying the identified fat fraction changes; The method of claim 1 , comprising one or more of:

3. 10. The method of claim 1, wherein identifying a plurality of indicators of back pain comprises applying a trained neural network to the one or more MRIs to automatically identify the plurality of indicators of back pain.

4. The method of claim 1 , wherein at least some of the MRIs include T1-weighted MRI, T2-weighted MRI, and / or fat-suppressed MRI.

5. 5. The method of claim 1, wherein at least some of the MRIs are generated using ultrashort echo arrival time MRI techniques.

6. 5. The method of claim 1, wherein at least some of the MRIs are generated using least squares estimation MRI techniques.

7. 5. The method of claim 1, wherein identifying one or more bone marrow intensity changes comprises identifying the one or more bone marrow intensity changes as either a type 1 Modic change or a type 2 Modic change.

8. The step of identifying one or more characteristics of vertebral endplate defects or vertebral endplate degeneration includes: identifying irregularities or deviations from the normal continuous lining of the vertebral endplates; identifying deviations from the normal contour profile of the vertebral endplate; Identifying fat fraction changes; and identifying one or more phenotypic subtypes of vertebral endplate defects; The method of claim 1 , comprising one or more of the following:

9. 5. The method of claim 1, further comprising determining a level of confidence in the objective score based on one or more additional indicators of back pain.

10. The one or more additional indicators are: Changes in multifidus muscle properties, Bone turnover in single-photon emission computed tomography images, and the pain score obtained for said particular subject; The method of claim 9 , comprising one or more of:

11. The method of claim 1 , wherein the method is performed by applying a machine learning algorithm.

12. 5. The method of claim 1, further comprising the step of displaying the objective score output on a display.

13. The method of claim 1 , wherein the objective score is used to generate a treatment recommendation.

14. 1. A computer-implemented method implemented by one or more processors in a medical imaging system for quantitatively predicting the likelihood that a particular subject will respond favorably to basilar nerve ablation for treating back pain, comprising: receiving, by the one or more processors, one or more images of at least a portion of the spine of the particular subject; automatically applying a pre-processing imaging technique to the one or more images; the one or more processors automatically extracting features from the one or more images to identify a plurality of indicators of back pain; Based on the extraction, the one or more processors automatically determine an objective score indicative of the likelihood that the particular subject will respond favorably to basilar nerve ablation; 10. A computer-operated method comprising:

15. 15. The computer-implemented method of claim 14, wherein the plurality of indices comprises at least one of (i) bone marrow intensity changes and (ii) characteristics of vertebral endplate defects or vertebral endplate degeneration.

16. 15. The computer-implemented method of claim 14, wherein the plurality of indices includes both (i) bone marrow intensity changes and (ii) features of vertebral endplate defects or vertebral endplate degeneration.

17. 17. The computer-operated method of any one of claims 14 to 16, wherein extracting features from the one or more images to identify the plurality of indicators of back pain comprises applying a trained neural network to the one or more images to automatically identify the plurality of indicators of back pain.

18. 17. The computer-implemented method of any one of claims 14 to 16, wherein the one or more images comprise images obtained from one or more of the following imaging modalities: magnetic resonance imaging ("MRI"), ultrashort echo time ("UTE") MRI sequence imaging, iterative decomposition of water and fat with echo asymmetry and least squares estimation ("IDEAL") MRI sequence imaging, fast spin-echo MRI sequence imaging, computed tomography ("CT") imaging, positron emission tomography ("PET") bone imaging, x-ray imaging, and fluoroscopy.

19. 15. The computer-implemented method of claim 14, wherein determining an objective score comprises automatically quantifying the plurality of indicators based on ranges of the plurality of indicators.

20. 20. The computer-implemented method of claim 19, wherein the range may include an amount, a severity, and / or a spatial rating.

21. The plurality of indicators are: Changes in multifidus muscle properties, Bone turnover in SPECT images, and The pain score obtained for the particular subject The computer-implemented method of any one of claims 14 to 16, further comprising one or more of:

22. A computer-implemented method according to any one of claims 14 to 16, wherein at least part of the method is carried out by application of a machine learning algorithm.

23. A computer-implemented method according to any one of claims 14 to 16, further comprising the step of displaying the objective score output on a display.

24. 1. A computer-implemented method, executed by one or more processors, of a computer system for quantitatively predicting the likelihood that a particular subject will respond favorably to basilar nerve ablation for treating chronic low back pain, comprising: receiving, by the one or more processors, one or more magnetic resonance images (MRIs) of at least one lumbosacral region of a spine of a particular subject; the one or more processors automatically applying pre-processing imaging techniques to the one or more MRIs to provide uniformity of the one or more MRIs for feature detection; the one or more processors automatically detecting features from the one or more MRIs to identify a plurality of indicators of chronic low back pain, the plurality of indicators including bone marrow intensity changes and features of vertebral endplate defects or vertebral endplate degeneration; the one or more processors automatically quantifying the identified indicators based on a range of the indicators, the range may include an amount, a severity, and / or a spatial assessment; Based on the quantification, the one or more processors automatically determine an objective score indicative of the likelihood that the particular subject will respond favorably to a basilar nerve ablation procedure; 10. A computer-operated method comprising:

25. The computer-operated method of claim 24, wherein the step of the one or more processors automatically detecting features from the one or more images to identify the plurality of indicators of chronic low back pain includes the step of applying a trained neural network to the one or more images to automatically identify the plurality of indicators of chronic low back pain.

26. 1. A computer-implemented method for training a neural network implemented on one or more processors to determine whether a particular subject is a likely candidate for a successful outcome from a spinal basilar nerve ablation procedure, comprising: collecting a set of magnetic resonance images from a database, each magnetic resonance image comprising an image of at least a portion of a spine in a subject having at least one indicator of chronic low back pain; applying one or more transforms to each magnetic resonance image to create a modified set of magnetic resonance images, wherein applying one or more transforms includes applying image processing transforms to make the magnetic resonance images more uniform for training the neural network; creating a first training set including the set of acquired magnetic resonance images, the set of corrected magnetic resonance images, and a set of magnetic resonance images of at least a portion of the spine in one or more subjects without indicators of chronic low back pain; training the neural network in a first stage using the first training set to identify indicators of chronic low back pain in magnetic resonance images; creating a second training set for a second stage of training that includes the first training set and magnetic resonance images of at least a portion of the spine in one or more subjects without indicators of chronic low back pain who are erroneously determined by the neural network after the first stage of training to have at least one indicator of chronic low back pain; training the neural network in a second stage using the second training set to reduce false positive determinations; 10. A computer-implemented method comprising:

27. 27. The method of claim 26, wherein applying one or more transformations comprises preprocessing the set of acquired magnetic resonance images to make the magnetic resonance images more uniform for training.

28. 27. The method of claim 26, further comprising identifying indicators of back pain likely to be successfully treated by the basilar nerve ablation procedure in at least a portion of the set of acquired magnetic resonance images.

29. 27. The method of claim 26, further comprising identifying images in the set of acquired magnetic resonance images in which the subject was successfully treated with the basilar nerve ablation procedure.

30. 27. The method of claim 26, wherein the set of acquired magnetic resonance images includes magnetic resonance images of a subject who has previously undergone spinal fusion or discectomy.

31. 1. A computer-implemented method for training a neural network to determine whether a particular subject is a likely candidate for a successful outcome from a basilar nerve ablation procedure, comprising: collecting a set of digital images from a database, each digital image comprising a digital image of at least a portion of a spine in a subject having at least one indicator of chronic low back pain; applying one or more transformations to each digital image to create a set of modified digital images, wherein applying one or more transformations includes applying image processing transformations to make the digital images more uniform for training the neural network; creating a first training set including the set of collected digital images, the set of corrected digital images, and a set of digital images of at least a portion of the spine in one or more subjects without indicators of chronic low back pain; training the neural network in a first stage using the first training set to identify indicators of chronic low back pain in digital images; creating a second training set for a second stage of training that includes the first training set and digital images of at least a portion of the spine in one or more subjects without indicators of chronic low back pain who are erroneously determined by the neural network after the first stage of training to have at least one indicator of chronic low back pain; training the neural network in a second stage using the second training set to reduce false positive determinations; 10. A computer-implemented method comprising:

32. 1. A system for quantitatively predicting the likelihood that a particular subject will respond favorably to basilar nerve ablation for treating chronic low back pain, comprising: a server that, upon execution of instructions stored on a non-transitory computer-readable storage medium, automatically receiving one or more magnetic resonance images (MRIs) of at least one lumbosacral region of the spine of a particular subject; automatically applying pre-processing imaging techniques to the one or more MRIs to render the one or more MRIs uniform for feature detection; automatically detecting features from the one or more MRIs to identify a plurality of indicators of chronic low back pain; the plurality of indicators include bone marrow intensity changes and characteristics of vertebral endplate defects or vertebral endplate degeneration; automatically quantifying the identified indicators based on a range of the indicators, the range may include an amount, a severity, and / or a spatial assessment; automatically determining an objective score indicative of the likelihood that the particular subject will respond favorably to a basilar nerve ablation procedure based on the quantification of the plurality of indices. wherein the system operates as an automated medical image analyzer.

33. 33. The system of claim 32, wherein the one or more hardware processors are further configured to detect features from the one or more MRIs to identify the plurality of indicators of chronic low back pain by applying a trained neural network to the one or more MRIs to automatically identify the plurality of indicators of chronic low back pain.

34. 34. The system of claim 32 or 33, further comprising an MRI scanner configured to acquire the one or more MRIs.

35. When executed by one or more processors of a medical image analysis system, the one or more processors automatically receiving one or more images of at least a portion of a spine of a particular subject; the one or more processors automatically applying a pre-processing imaging technique to the one or more images; the one or more processors automatically extracting features from the one or more images to identify a plurality of indicators of back pain, the plurality of indicators including at least one of (i) bone marrow intensity changes and (ii) features of vertebral endplate defects or vertebral endplate degeneration; the one or more processors automatically determining an objective score based on the extraction that indicates the likelihood that the particular subject will respond favorably to basilar nerve ablation; 1. A non-transitory physical computer storage medium comprising stored computer-executable instructions configured to perform a process comprising:

36. The process further includes automatically quantifying the identified plurality of indicators of back pain, the step of automatically quantifying the identified plurality of indicators comprising: determining the amount of bone marrow intensity change and / or vertebral endplate defect; determining the level of the bone marrow intensity change and / or the extent of the vertebral endplate defect; and Quantifying the identified changes in fat fraction; 36. The non-transitory physical computer storage medium of claim 35, comprising one or more of:

37. 36. The non-transitory physical computer storage medium of claim 35, wherein the back pain is chronic lower back pain.

38. 36. The non-transitory physical computer storage medium of claim 35, wherein extracting features comprises applying a trained neural network to the one or more images to automatically identify the plurality of indicators of back pain.

39. 1. A computer-implemented method executed by one or more processors for quantitatively predicting the likelihood that a particular subject will respond favorably or adversely to neuromodulation, comprising: the one or more processors automatically identifying a plurality of indicators of back and / or peripheral nerve pain, the automatically identifying a plurality of indicators including at least the steps of: (i) automatically identifying one or more bone marrow intensity changes; and (ii) automatically identifying one or more features of vertebral endplate defects or vertebral endplate degeneration; the one or more processors automatically quantifying the identified plurality of metrics; the one or more processors automatically calculating an objective score based on the quantification, the objective score indicating the likelihood that the particular subject will respond favorably or adversely to the neuromodulation; A method comprising:

40. 40. The method of claim 39, wherein the plurality of indicia are identified based on imaging data.

41. 40. The method of claim 39, wherein the plurality of indices are identified based on magnetic resonance imaging (MRI).

42. 40. The method of claim 39, wherein the plurality of indicia are identified based on scanned data.

43. 43. The method of any one of claims 39 to 42, wherein the plurality of indicia are identified based on acoustic data.

44. 43. The method of any one of claims 39 to 42, wherein the neuromodulation comprises denervation.

45. 43. The method of any one of claims 39 to 42, wherein the neuromodulation comprises denervation or ablation of the basal spinal nerve.

46. 43. The method of any one of claims 39 to 42, further comprising classifying a plurality of subjects based on the objective scores calculated for the subjects.

Images