Providing guidance information for implantable device extraction procedures

Abstract

A system for providing guidance information for an extraction procedure of an implantable device is provided, the system having one or more processors configured to receive attenuation data representing the implantable device in an anatomical region, the attenuation data defining x-ray attenuation within the anatomical region, analyze the attenuation data to determine an amount of adhesion between the implantable device and tissue in contact with the implantable device, and output guidance information for the extraction procedure of the implantable device based on the amount of adhesion.

Description

[Technical Field]

[0001] SUMMARY A computer-implemented method, a computer program product, and a system are disclosed that provide guidance information for an implantable device extraction procedure. [Background technology]

[0002] Medical procedures often involve the implantation of devices within the body. For example, cardiac pacemaker leads are routinely implanted in the body to transmit electrical pulses to the heart, thereby regulating the heart's beat. These leads pass through the vascular system between the pacemaker, which generates the electrical pulses, and the heart tissue, with the distal end of the lead delivering the electrical pulses. This allows cardiac pacemakers to be placed in more accessible locations within the body, typically under a flap of skin in the chest.

[0003] Implantable cardioverter-defibrillator ("ICD") leads are also routinely implanted in the body to conduct electrical pulses from the ICD to the heart. The ICD's leads pass through the vasculature between the ICD, which generates the electrical pulses, and the heart tissue in a similar manner, with the distal end of the leads similarly delivering the electrical pulses. ICDs operate in a variety of modes. For example, the ICD can deliver a rapid series of low-voltage impulses to attempt to correct the heart rhythm, or the ICD can deliver one or more small electrical shocks to the heart to attempt to convert the heart back to a normal rhythm, or the ICD can deliver one or more larger electrical shocks to the heart to attempt to convert the heart back to a normal rhythm.

[0004] Aside from cardiac pacemaker and ICD leads, other types of implantable devices can also be implanted in the body and used for other purposes, including, for example, inferior vena cava "IVC" filters.

[0005] However, implantable devices occasionally need to be extracted from the body. For example, pacemaker or ICD leads need to be extracted if the leads malfunction or if an infection occurs. When a lead is in place for a short period of time, the adhesions between the lead and any contacting tissue are relatively weak. In this situation, it is possible to extract the lead simply by pulling on the lead. However, when the lead is in place for more than several months, the adhesions between the lead and contacting tissue are relatively strong. This relatively strong adhesion, combined with the lead's tortuous path through the vasculature, makes this technique unsuitable. In this situation, various tools can be used to facilitate lead extraction. For example, a tool known as a locking stylet is used to pass through the lumen of the lead and provide additional support while pulling on the lead. The locking stylet transfers the force applied by pulling on the lead to the distal end of the lead, thereby facilitating its extraction. Alternatively, a tool containing a sheath with a cutting blade at its distal end can be passed over the lead and used to sever the lead from the tissue it contacts. An example of such a tool is the TightRail mechanical rotating dilator sheath, marketed by Philips Healthcare, Best, The Netherlands. A variation of this type of tool includes a sheath with an optical fiber instead of a cutting blade. In this case, the sheath passes over the lead while the optical fiber delivers laser light to the tissue that contacts the distal end of the sheath, which then An example of such a tool is the GlideLight Laser sheath, marketed by Philips Healthcare, Best, The Netherlands.

[0006] Thus, a variety of techniques and tools are available for use in extracting implantable devices, such as pacemaker and ICD leads, and other devices from the body, should the need arise. Summary of the Invention [Problem to be solved by the invention]

[0007] However, determining how to perform an implanted device extraction procedure remains difficult. [Means for solving the problem]

[0008] According to one aspect of the present disclosure,

[0009] A computer-implemented method for providing guidance information for an implantable device extraction procedure is provided, the method comprising: receiving attenuation data representative of an implantable device in an anatomical region, the attenuation data defining x-ray attenuation within the anatomical region; analyzing the attenuation data to determine the amount of adhesion between the implantable device and tissue in contact with the implantable device; outputting guidance information for an extraction procedure of the implantable device based on the amount of adhesion; Includes:

[0010] The method outputs guidance information. The guidance information is based on the amount of adhesion between the implantable device and tissue in contact with the implantable device. The amount of adhesion is determined from attenuation data defining X-ray attenuation in an anatomical region. Thus, the method provides a reliable, non-invasive technique for guiding extraction of the implantable device. Further aspects, features, and advantages of the present disclosure will become apparent from the following description of examples taken in conjunction with the accompanying drawings. [Brief explanation of the drawings]

[0011] [Figure 1] FIG. 1 is an example of a projection x-ray image 110 depicting multiple pacemaker leads 1201, 1202 in an anatomical region, according to some aspects of the present invention. [Figure 2]FIG. 2 is a schematic diagram illustrating an example of a device 170 for use in an implantable device extraction procedure, according to some aspects of the present disclosure. [Figure 3] FIG. 3 is a flowchart illustrating an example of a computer-implemented method for providing guidance information for an extraction procedure of an implantable device, according to some aspects of the present disclosure. [Figure 4] FIG. 4 is a schematic diagram illustrating an example of a system 200 for providing guidance information for an extraction procedure of an implantable device, according to some aspects of the present disclosure. [Figure 5] FIG. 5 is an example of a projection X-ray image 110 including guidance information for an implantable device extraction procedure in the form of an indication of the amount of adhesion 130 between the implantable device 120 and tissue 140 in contact with the implantable device 120, according to some aspects of the present disclosure. [Figure 6] FIG. 6 is a schematic diagram illustrating a display 220 including example guidance information 150, 160, 170 for an extraction procedure of an implantable device, according to some aspects of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0012] Examples of the present disclosure are provided with reference to the following description and drawings. In this description, for purposes of explanation, many specific details of particular examples are set forth. Reference herein to an "example," "embodiment," or similar terms means that a feature, structure, or characteristic described in connection with an example is included in at least that example. It should also be understood that features described in connection with one example may also be used in other examples, and that, for purposes of brevity, not all features are necessarily repeated in each example. For example, features described in connection with a computer-implemented method may be implemented in a computer program product, a system, and a corresponding method.

[0013] In the following description, reference is made to various examples of computer-implemented methods for providing guidance information for an implantable device extraction procedure. Reference is made to examples in which the implantable device is a pacemaker lead. However, it should be understood that the lead serves as an example only, and that the methods disclosed herein can also be used to provide guidance information for extracting other types of implantable devices from the body. For example, the methods disclosed herein can be used to provide guidance information for extracting an ICD lead from the body, or for extracting a pacemaker or ICD itself from the body, or for extracting an IVC filter, or for extracting a so-called leadless pacemaker or ICD from the body.

[0014] It should be noted that the computer-implemented methods disclosed herein may be provided as a non-transitory computer-readable storage medium storing computer-readable instructions that, when executed by at least one processor, cause the at least one processor to perform the method. In other words, the computer-implemented methods may be implemented in a computer program product. The computer program product may be provided by dedicated hardware or hardware capable of executing software in association with appropriate software. When provided by a processor, the functions of the method features may be provided by a single dedicated processor, a single shared processor, or multiple individual processors, some of which are shared. One or more of the functions of the method features may be provided by a processor shared within a networked processing architecture, such as a client / server architecture, a peer-to-peer architecture, the Internet, or a cloud.

[0015] Explicit use of the terms "processor" or "controller" should not be construed as exclusively referring to hardware capable of executing software, but can implicitly include, but is not limited to, digital signal processor "DSP" hardware, read-only memory "ROM," random access memory "RAM," and non-volatile storage devices for storing software. Furthermore, examples of the present disclosure may take the form of a computer-usable storage medium or computer program product accessible from a computer-readable storage medium, the computer program product providing program code for use by or in connection with a computer or any instruction execution system. For purposes of this discussion, a computer-usable storage medium or computer-readable storage medium may be any apparatus that can have, store, communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. The medium may be an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system or device, or a propagation medium. Examples of computer-readable media include semiconductor or solid-state memory, magnetic tape, removable computer diskettes, random access memory "RAM", read-only memory "ROM", rigid magnetic disks, and optical disks. Current examples of optical disks include CD-ROM, CD-R / W, Blu-Ray, and DVD.

[0016] As mentioned above, determining how to perform an implantable device extraction procedure remains difficult.

[0017] As an example, an implantable device extraction procedure will be described where the implantable device is a pacemaker lead. FIG. 1 is an example of a projection x-ray image 110 depicting multiple pacemaker leads 1201, 1202 in an anatomical region, according to some aspects of the present invention. In the example shown in FIG. 1, the anatomical region is the cardiac region, and various ribs and vertebrae are visible. Two pacemaker leads 1201, 1202 are also visible in FIG. 1. Typically, a pacemaker can include one or more such leads, the distal ends of which are implanted in cardiac tissue. In situations where a single lead is used, the distal end of that lead is typically implanted in cardiac tissue within the right ventricle. When two leads are used, the distal end of one lead is often implanted in cardiac tissue within the right atrium, and the distal end of the other lead is often implanted in cardiac tissue within the right ventricle. When three leads are used, typically the distal end of one lead is often implanted in cardiac tissue within the right atrium, the distal end of another lead is often implanted in cardiac tissue within the right ventricle, and the distal end of another lead is often implanted in cardiac tissue within the left ventricle.

[0018] As mentioned above, after a pacemaker lead is implanted in the body, it may become necessary to extract the lead. For example, a pacemaker or ICD lead may need to be extracted if the lead malfunctions or if an infection occurs. When the lead is in place for a short period of time, adhesions between the lead and any contacting tissue are relatively weak. In this situation, it is possible to extract the lead by simply pulling on the lead. However, when the lead is in place for more than several months, adhesions between the lead and contacting tissue are relatively strong. For example, over time, scar tissue develops in the cardiac tissue around the distal end of the lead, and the lead becomes strongly adhered to the scar tissue, preventing extraction. Over time, the lead may also adhere to tissue along the walls of blood vessels in the vasculature through which the lead passes, also preventing extraction. As explained in the article by Keiler, J. et al., "Neointimal fibrotic lead encapsulation - Clinical challenges and demands for implantable cardiac electronic devices," Journal of Cardiology, Vol. 70, Issue 1, July 2017, Pages 7-17, over time, new tissue may grow along the length of the lead. The adhesion of the new tissue to the lead increases friction during lead removal and impedes lead extraction. The new tissue may also attach to the vessel walls within the vasculature through which the lead passes, impeding lead extraction. Therefore, in these situations, the relatively strong adhesion between the lead and the tissue in contact with it (as seen in Figure 1), combined with the tortuous path of the lead through the vasculature, means that the technique of extracting the lead by simply pulling on the lead is inadequate.

[0019] As mentioned above, various tools are used to facilitate pacemaker lead extraction in situations where relatively strong adhesions exist between the lead and the tissue in contact with the lead. For example, a tool known as a locking stylet can be used. Alternatively, a tool including a sheath with a cutting blade at its distal end can be passed over the lead to sever it from the contacting tissue. A variation of this type of tool includes a sheath with an optical fiber instead of a cutting blade. In this case, the sheath passes over the lead while the optical fiber irradiates laser light onto the tissue in contact with the distal end of the sheath, thereby freeing the lead from the tissue. An example of this type of tool is shown in FIG. 2, which is a schematic diagram of an example device 170 for use in an implantable device extraction procedure according to some aspects of the present disclosure.

[0020] The device 170 shown in Figure 2 includes a sheath 310 having a bore 320. A plurality of optical fibers 330 1...m The distal end of the optical fiber 330 is positioned around the bore. A light source, not shown in FIG. 2, generates optical radiation that is coupled into the proximal end of the optical fiber. In the example shown in FIG. 2, the light source is an ultraviolet excimer laser having a wavelength of approximately 308 nm. In use, a pacemaker lead is passed through the bore 320, with the sheath 310 passing over the pacemaker lead while the optical fiber 330 1...m The distal end of the sheath irradiates the tissue surrounding the distal end with optical radiation from the laser. The effect of irradiating the tissue is to ablate it to a depth of approximately 50 microns, which frees the pacemaker leads from the tissue.

[0021] However, determining how to perform an implantable device extraction procedure for such leads, as well as other types of implantable devices, remains challenging. For example, in the procedure described above, the physician must decide which of the techniques described above to use and which of the available tools to use. Different techniques and tools may be appropriate for different stages of the extraction procedure. For example, one technique may be appropriate for use in extracting a lead along its length, while another technique may be appropriate for extracting the distal end of the lead. Furthermore, if a tool is used, the physician must determine in what setting to use the tool. Extraction procedures for other types of implantable devices face similar challenges.

[0022] FIG. 3 is a flowchart illustrating an example of a computer-implemented method for providing guidance information for an extraction procedure of an implantable device, according to some aspects of the present disclosure. FIG. 4 is a schematic diagram illustrating an example of a system 200 for providing guidance information for an extraction procedure of an implantable device, according to some aspects of the present disclosure. The system 200 includes one or more processors 210. It should also be noted that the operations described in connection with the method illustrated in FIG. 3 are performed by the one or more processors 210 of the system 200 illustrated in FIG. 4. Similarly, the operations described in connection with the one or more processors 210 of the system 200 are also performed in the method described with reference to FIG. 3. Referring to FIG. 3, the computer-implemented method for providing guidance information for an extraction procedure of an implantable device includes: receiving (S110) attenuation data 110 representing an implantable device 120 within an anatomical region, the attenuation data defining x-ray attenuation within the anatomical region; analyzing the attenuation data 110 to determine (S120) an amount of adhesion 130 between the implantable device 120 and tissue 140 in contact with the implantable device; outputting guidance information for an extraction procedure of the implantable device based on the adhesion amount (S130); Includes:

[0023] In the method, the guidance information is output. The guidance information is based on the amount of adhesion between the implantable device and tissue in contact with the implantable device. The amount of adhesion is determined from attenuation data defining X-ray attenuation in an anatomical region. Thus, the method provides a reliable non-invasive technique for guiding the extraction of an implantable device.

[0024] The above methods are described in more detail below.

[0025] 3, in act S110, attenuation data 110 is received. The attenuation data represents an implantable device 120 within an anatomical region and defines the x-ray attenuation within the anatomical region.

[0026] In general, the implantable device can be any type of implantable device, and the implantable device can be placed in any anatomical region.

[0027] The attenuation data 110 received in act S110 may generally represent a still image or a temporal sequence of images. In the latter case, the temporal sequence of images may be generated substantially in real time, and the method described above may be performed substantially in real time, thereby providing live guidance on the live attenuation data during the implantable device extraction procedure.

[0028] In general, the attenuation data 110 received in act S110 can be projection data, i.e., 2D data used to generate a 2D image, or volumetric data, i.e., 3D data used to generate a 3D image. In the latter case, the data can be raw data, i.e., data not yet reconstructed into a volumetric or 3D image, or image data, i.e., data already reconstructed into a 3D image. Thus, in general, the attenuation data 110 received in act S110 is generated by an X-ray projection imaging system or by a CT imaging system.

[0029] CT imaging systems generate volumetric data by rotating or stepping an x-ray source-detector arrangement around an object and acquiring attenuation data of the object from multiple rotational angles relative to the object. The volumetric data is then reconstructed into a 3D image of the object. Examples of CT imaging systems include cone-beam CT imaging systems, photon-counting CT imaging systems, dark-field CT imaging systems, and phase-contrast CT imaging systems. An example of a CT imaging system 230 used to generate the volumetric attenuation data 110 received in operation S110 is shown in FIG. 4. As an example, the volumetric attenuation data 110 is generated by a CT5000 Ingenuity CT scanner commercially available from Philips Healthcare of Vest, The Netherlands.

[0030] As described above, the attenuation data 110 received in operation S110 may alternatively be generated by an X-ray projection imaging system. An X-ray projection imaging system typically includes a support arm, a so-called “C-arm,” that supports an X-ray source and an X-ray detector. Alternatively, the X-ray projection imaging system may include a support arm having a different shape than the present example, such as an O-arm. In contrast to a CT imaging system, an X-ray projection imaging system generates attenuation data of an object with the X-ray source and X-ray detector in a stationary position relative to the object. The attenuation data, also referred to as projection data, is then used to generate a 2D image of the object. An example of an X-ray projection imaging system that may be used to generate the projection attenuation data 110 received in operation S110 is the Azurion 7 X-ray projection imaging system, commercially available from Philips Healthcare, located in Vest, The Netherlands.

[0031] Alternatively, the attenuation data 110 received in operation S110 may be volumetric attenuation data generated by a moving X-ray projection imaging system. The X-ray projection imaging system may generate the volumetric data by rotating or stepping an X-ray source and an X-ray detector around an object and acquiring projection data of the object from multiple rotation angles relative to the object. Image reconstruction techniques may then be used to reconstruct the projection data acquired from the multiple rotation angles relative to the object into a volumetric image in a manner similar to the reconstruction of a volumetric image using attenuation data acquired from a CT imaging system. Thus, the volumetric data may be generated by an X-ray projection imaging system. Thus, in examples in which volumetric attenuation data 110 is received in operation S110, the volumetric attenuation data 110 may be generated by an X-ray projection imaging system.

[0032] In some examples, the attenuation data 110 received in act S110 is spectral attenuation data. The spectral attenuation data may be divided into a plurality of different energy intervals DE 1..mdefines the x-ray attenuation in an anatomical region at m. Generally, there are two or more energy intervals, i.e., m is an integer and m≧2. 1..m The ability to generate x-ray attenuation data in 100 sq m (100 sq ft) distinguishes spectral x-ray imaging systems from conventional x-ray imaging systems. By processing data from multiple different energy intervals, media with similar x-ray attenuation values ​​when measured within a single energy interval can be distinguished, which is indistinguishable with conventional x-ray image data.

[0033] It should be noted that in examples where the attenuation data received in act S110 is spectral attenuation data, this spectral attenuation data may be provided in the form of projection data or volumetric data. Thus, the spectral attenuation data received in act S110 may be generated by a spectral X-ray projection imaging system or a spectral CT imaging system. More generally, the spectral attenuation data received in act S110 may be generated by a spectral X-ray imaging system. Examples of spectral X-ray imaging systems that may be used to generate the spectral attenuation data 110 received in act S110 include a cone-beam spectral X-ray imaging system, a photon-counting spectral X-ray imaging system, a dark-field spectral X-ray imaging system, and a phase-contrast spectral X-ray imaging system. One example of a spectral CT imaging system that may be used to generate the spectral attenuation data 110 received in act S110 is the Spectral CT 7500, available from Philips Healthcare, located in Vest, The Netherlands.

[0034] In general, spectral attenuation data is generated by a variety of different configurations of spectral X-ray imaging systems that include an X-ray source and an X-ray detector. The X-ray source of the spectral X-ray imaging system can include multiple monochromatic sources or one or more polychromatic sources, and the X-ray detector of the spectral X-ray imaging system can include a common detector for detecting multiple different X-ray energy intervals, or each detector can detect a different X-ray energy interval.1..m The detector may include a plurality of detectors that detect X-rays having energies within different X-ray energy intervals, a multi-layer detector in which X-rays having energies within different X-ray energy intervals are detected by corresponding layers, or a photon-counting detector that classifies detected X-ray photons into one of a plurality of energy intervals based on their individual energies. In a photon-counting detector, the associated energy interval for each received X-ray photon is determined by detecting pulse heights induced by electron-hole pairs generated in response to absorption of the X-ray photon in the direct conversion material.

[0035] Using the various configurations of X-ray source and detector described above, various X-ray energy intervals DE 1..m Typically, discrimination between different X-ray energy intervals at an X-ray source can be provided by temporally switching the X-ray tube potential of a single X-ray source, i.e., "rapid kVp switching," or by temporally switching or filtering the X-ray emission from multiple X-ray sources. In such configurations, a common X-ray detector can be used to detect X-rays across multiple different energy intervals, with attenuation data for each energy interval being generated in a time series. Alternatively, discrimination between different X-ray energy intervals at a detector can be provided by using a multi-layer detector, or a photon-counting detector. Such detectors can detect X-rays across multiple X-ray energy intervals DE 1..m X-rays from different X-ray energy intervals DE can be detected almost simultaneously, and therefore no time switching in the radiation source is required. In this way, a multi-layer detector or a photon counting detector can be used in combination with a polychromatic source to detect X-rays from different X-ray energy intervals DE. 1..m It is possible to generate X-ray attenuation data in

[0036] Other combinations of X-ray sources and detectors as described above can also be used to provide spectral attenuation data. For example, in a further configuration, the need to continuously switch between different X-ray sources emitting X-rays at different energy intervals is eliminated by mounting the X-ray source-detector pairs on the gantry at rotationally offset positions about the axis of rotation. In this configuration, each X-ray source-detector pair operates independently and emits X-rays at different energy intervals DE 1..m Separation of the spectral attenuation data for different energy intervals DE is facilitated by the rotational offset of the X-ray source-detector pair. To reduce the effects of X-ray scattering, an energy-selective filter is applied to the X-ray detector, and in this configuration, the spectral attenuation data for different energy intervals DE 1..m Improved separation between the spectral attenuation data for .times. ...

[0037] 1, in this example, the attenuation data 110 received in act S110 represents pacemaker leads 1201, 1202 in an anatomical region of the heart. In this example, the attenuation data 110 comprises projection data and is provided in the form of a projection x-ray image.

[0038] In general, the attenuation data 110 received in act S110 can be received via any form of data communication, including wired, optical, and wireless communication. As some examples, when wired or optical communication is used, the communication occurs via signals transmitted over electrical or optical cables, and when wireless communication is used, the communication occurs via, for example, RF or optical signals. The attenuation data 110 received in act S110 can be received from a variety of sources. For example, the attenuation data 110 is received from an imaging system, such as one of the imaging systems described above. Alternatively, the attenuation data 110 may be received from another source, such as, for example, a computer-readable storage medium, the Internet, or the cloud.

[0039] Returning to FIG. 3, in act S120, the attenuation data 110 is analyzed to determine the amount of adhesion 130 between the implantable device 120 and tissue 140 in contact with the implantable device.

[0040] As discussed above, extraction of an implantable device, such as a pacemaker lead, is hindered by adhesions between the lead and the tissue in contact with the lead. Adhesion can occur both along the length of the lead and at the distal end of the lead. The insight utilized in this disclosure is that a measure of this adhesion can be determined from the attenuation data 110 described above. In operation S120, three approaches for determining the amount of adhesion between the implantable device 120 and the tissue 140 in contact with the implantable device are described below. In a first approach, the amount of adhesion 130 between the implantable device 120 and the tissue 140 in contact with the implantable device is determined based on the attenuation value of the tissue in Hounsfield units. In a second approach, spectral attenuation data is analyzed to determine the amount of adhesion 130 between the implantable device 120 and the tissue 140 in contact with the implantable device. In a third approach, a neural network is used to predict the amount of adhesion 130 between the implantable device 120 and tissue 140 in contact with the implantable device. In the third approach, the attenuation data may be conventional X-ray attenuation data or spectral attenuation data. In either approach, segmenting the received attenuation data 110 to identify the implantable device 120 may be performed prior to analyzing the attenuation data 110 (S120) to facilitate analysis of the attenuation data. For this purpose, various segmentation algorithms may be used, such as, for example, model-based segmentation, watershed-based segmentation, region growing, level set, or graph cut. A neural network is also trained to segment the attenuation data to identify the implantable device.

[0041] In a first approach, operation S120 of analyzing the attenuation data 110 to determine the amount of adhesion 130 between the implantable device 120 and tissue 140 in contact with the implantable device includes: determining an attenuation value in Hounsfield units of tissue 140 in contact with the implantable device; determining an amount of adhesion 130 between the implantable device 120 and tissue 140 in contact with the implantable device based on the attenuation value in Hounsfield units; Includes:

[0042] In this first approach, the attenuation data 110 can be volumetric data, i.e., 3D data. Volumetric data is generated, as described above, by a CT imaging system or by rotating or stepping the X-ray source and X-ray detector of an X-ray projection imaging system around the anatomical region to acquire projection data of the anatomical region from multiple rotational angles relative to the anatomical region. As described above, the data can be conventional attenuation data or spectral attenuation data.

[0043] Attenuation values ​​in Hounsfield units are relative quantitative measurements of radiation density calculated based on a linear transformation of the linear attenuation coefficient of an X-ray beam. Distilled water at standard temperature and pressure is defined as 0 Hounsfield units, and air is defined as -1000 HU. Attenuation values ​​in Hounsfield units, HU, are calculated according to the following formula:

number

[0044] where μ water and μ airwhere σ represents the linear attenuation coefficients of water and air, respectively. Conventional CT imaging systems are routinely calibrated using water and air phantoms so that the attenuation data of these phantoms is generated essentially in Hounsfield units. Therefore, the attenuation data 110 analyzed in act S120 may be volumetric data generated by a conventional CT imaging system. Volumetric data generated by rotating or stepping the x-ray source and x-ray detector of an x-ray projection imaging system around an anatomical region may also be calibrated in a similar manner. Therefore, the attenuation data 110 analyzed in act S120 may alternatively be volumetric data generated by a conventional x-ray projection imaging system.

[0045] Alternatively, the attenuation data 110 may be provided in HU values ​​by converting volumetric data generated by a spectral CT imaging system or by converting volumetric data generated by rotating or stepping the X-ray source and X-ray detector of a spectral X-ray projection imaging system around the anatomical region. Spectral attenuation data generated by such imaging systems may be converted into Hounsfield units, often referred to as "synthetic" Hounsfield units, using various techniques. One example of such a technique is disclosed in Bornefalk, H., "Synthetic Hounsfield units from spectral CT data," Phys Med Biol., 2012 Apr 7;57(7):N83-7.

[0046] 1, in a first approach, voxels in the attenuation data 110 that are in contact with an implantable device, such as the lead 1201 shown in FIG. 1, are identified, and the amount of adhesion 130 between the implantable device 120 and the tissue 140 in contact with the implantable device is determined based on the attenuation value in Hounsfield units of that voxel. This operation may include segmenting the attenuation data 110 to identify the lead, as described above.

[0047] In a first approach, operation S120 of analyzing the attenuation data 110 to determine the amount of adhesion 130 between the implantable device 120 and the tissue 140 in contact with the implantable device includes quantifying the amount of adhesion between the implantable device and the tissue in contact with the implantable device in one or more adhesion classes or adhesion scores. This is done by assigning voxels to adhesion classes or adhesion scores using a functional relationship between HU values ​​and adhesion classes or adhesion scores. The functional relationship may be provided, for example, by a graph or a look-up table.

[0048] For example, in one example, voxels in contact with implantable devices having relatively low HU values ​​are mapped to an adhesion class or adhesion score representing a relatively low amount of adhesion, whereas voxels in contact with implantable devices having relatively high HU values ​​are mapped to an adhesion class or adhesion score representing a relatively high amount of adhesion. For example, unclotted blood typically adheres weakly to pacemaker leads and therefore does not significantly impede lead extraction from the body. unclotted blood typically has an HU value of 30 HU to 45 HU. In this example, voxels in contact with leads having HU values ​​in this range are mapped to an adhesion class or adhesion score representing a relatively low amount of adhesion. In contrast, clotted blood typically adheres more strongly to the leads. Clotted blood adheres to pacemaker leads along their length, increasing friction between the leads and the surrounding vasculature and thereby impeding their extraction. Clotted blood typically has an HU value of 60 HU to 100 HU. As a result, voxels in contact with the lead wires that have HU values ​​in this range are mapped to an adhesion class or adhesion score that represents a relatively high amount of adhesion. New tissue that grows along the length of a pacemaker lead wire over time and scar tissue that develops in the cardiac tissue around the lead wire's distal end are also typically strongly adhered to the lead wire. New tissue and scar tissue have relatively higher HU values ​​than unclotted blood. New tissue and scar tissue also interfere with lead wire extraction, as described above. As a result, in the first approach, voxels in contact with the lead wires that have HU values ​​higher than that of unclotted blood and that also represent new tissue or scar tissue are mapped to an adhesion class or score that represents a relatively high amount of adhesion.

[0049] As mentioned above, in a second approach, the spectral attenuation data is analyzed to determine the amount of adhesion 130 between the implantable device 120 and tissue 140 in contact with the implantable device. In this approach, the attenuation data 110 comprises spectral attenuation data that is analyzed over a plurality of different energy intervals DE. 1..m Defines radiographic attenuation within an anatomical region.

[0050] In this second approach, the attenuation data 110 may be volumetric data, ie, 3D data, or projection data, ie, 2D data.

[0051] Referring to FIG. 1 , in a second approach, voxels or pixels in the attenuation data 110 that are in contact with an implantable device, such as the electrical lead 1201 shown in FIG. 1 , are identified, and the amount of adhesion 130 between the implantable device 120 and the tissue 140 in contact with the implantable device is calculated by the effective atomic number Z of the voxel / pixel. eff or based on the type of tissue represented by the voxel / pixel. This operation includes segmenting the attenuation data 110 to identify leads, as described above.

[0052] The effective atomic number of a voxel / pixel provides a rough indication of the strength of adhesion of the material represented by the voxel / pixel to the implantable device. Thus, voxels / pixels in contact with the implantable device that have a relatively low value of effective atomic number tend to represent materials that are relatively weakly adhered to the implantable device. For example, unclotted blood has a relatively low value of effective atomic number and is typically weakly adhered to the implantable device. In contrast, voxels / pixels in contact with the implantable device that have a relatively high value of effective atomic number tend to represent materials that are relatively strongly adhered to the implantable device. For example, clotted blood, new tissue, and scar tissue have a relatively high value of effective atomic number and are typically more strongly adhered to the implantable device. The effective atomic number is calculated from the attenuation data 110 using various techniques. An example of a suitable technique is disclosed in Saito, M. et al., "A simple formulation for deriving effective atomic numbers via electron density calibration from dual-energy CT data in the human body," Medical Physics, Vol. 44, Issue 6, June 2017, pages 2293-2303.

[0053] In a second approach, operation S120 of analyzing the attenuation data 110 to determine the amount of adhesion 130 between the implantable device 120 and the tissue 140 in contact with the implantable device includes quantifying the amount of adhesion between the implantable device and the tissue in contact with the implantable device in one or more adhesion classes, or adhesion scores, which may be calculated using an effective atomic number Z eff This is done by assigning voxels / pixels to adhesion classes or adhesion scores using a functional relationship between the value of σ and the adhesion class or adhesion score, which may be provided, for example, by a graph or a look-up table.

[0054] For example, in one embodiment, a relatively low effective atomic number Z eff Voxels / pixels in contact with the implantable device having a value of 0.001 will be mapped to an adhesion class or adhesion score that represents a relatively small amount of adhesion, whereas v ... high amount of adhesion. eff Voxels / pixels in contact with the implantable device having a value of .gtoreq.1 are mapped to an adhesion class or adhesion score that represents a relatively high amount of adhesion.

[0055] In a second approach, the adhesion volume 130 between the implantable device 120 and the tissue 140 in contact with the implantable device can alternatively be calculated as the effective atomic number Z of the voxel / pixel: eff Instead of being determined based on the spectral attenuation data, the tissue type represented by the voxel / pixel may be determined based on the tissue type represented by the voxel / pixel, in which case the tissue type represented by the voxel / pixel may be determined by applying a material decomposition algorithm to the spectral attenuation data.

[0056] In this manner, since different types of tissue are known to adhere to implantable devices with different strengths, the type of tissue in contact with the implantable device can be used to reliably determine the amount of adhesion 130 between the implantable device 120 and the tissue 130 in contact with the implantable device. In this manner, operation S120 of analyzing the attenuation data 110 to determine the amount of adhesion 130 between the implantable device 120 and the tissue 140 in contact with the implantable device includes: applying a material decomposition algorithm to the spectral attenuation data to identify tissue types in contact with the implantable device; determining an amount of adhesion 130 between the implantable device 120 and tissue 130 in contact with the implantable device based on the type of tissue; Includes:

[0057] In this approach, operation S120 of analyzing the attenuation data 110 to determine the amount of adhesion 130 between the implantable device 120 and the tissue 140 in contact with the implantable device includes quantifying the amount of adhesion between the implantable device and the tissue in contact with the implantable device in one or more adhesion classes, or adhesion scores.

[0058] The functional relationship between tissue type and adhesion class or adhesion score can be used to assign voxels / pixels in the attenuation data 110 to adhesion classes or adhesion scores. The functional relationship may be provided, for example, by a graph or a look-up table.

[0059] For example, in one example, voxels / pixels in contact with an implantable device representing the tissue type "soft tissue" are mapped to an adhesion class or adhesion score representing a relatively low amount of adhesion, whereas voxels / pixels in contact with an implantable device representing the tissue type "dense tissue" are mapped to an adhesion class or adhesion score representing a relatively high amount of adhesion.

[0060] According to this example, various material decomposition algorithms can be used to identify tissue types. One example of a material decomposition algorithm that can be used is disclosed in Brendel, B. et al., "Empirical, projection-based basis-component decomposition method," Medical Imaging 2009, Physics of Medical Imaging, edited by Ehsan Samei and Jiang Hsieh, Proc. of SPIE Vol. 7258, 72583Y. Another example of a material decomposition algorithm that can be used is disclosed in Roessl, E. and Proksa, R., "K-edge imaging in X-ray computed tomography using multi-bin photon counting detectors," Phys Med Biol. 2007 Aug 7, 52(15):4679-96. Another example of a material decomposition algorithm that can be used is disclosed in published PCT patent application WO / 2007 / 034359A2. Another example of a material decomposition algorithm that can be used is disclosed in Silva, AC et al., "Dual-energy (spectral) CT: applications in abdominal imaging", RadioGraphics 2011; 31(4):1031-1046.

[0061] As described above, in a third approach, a neural network is used to predict the amount of adhesion 130 between the implantable device 120 and the tissue 140 in contact with the implantable device. In this approach, operation S120 of analyzing the attenuation data 110 to determine the amount of adhesion 130 between the implantable device 120 and the tissue 140 in contact with the implantable device includes: inputting the received attenuation data 110 into a neural network; using the neural network to predict the amount of adhesion between the implantable device and tissue in contact with the implantable device; wherein the neural network is trained to predict the amount of adhesion between an implantable device and a tissue in contact with the implantable device using training attenuation data having a plurality of training images representing the implantable device in an anatomical region and ground truth data having ground truth adhesion values ​​corresponding to the training images, the ground truth adhesion values ​​representing the amount of adhesion between the implantable device and a tissue in contact with the implantable device.

[0062] In this approach, the attenuation data input to the neural network can be conventional X-ray attenuation data or spectral attenuation data. The attenuation data can be volumetric data or projection data. The attenuation data can represent a still image or a temporal sequence of images. The neural network can be implemented by a variety of architectures, such as a convolutional neural network (CNN), a recurrent neural network (RNN), or a transformer.

[0063] In one example, the neural network is trained to predict the amount of adhesion 130 between the implantable device 120 and the tissue 140 in contact with the implantable device based further on one or more of age, sex, comorbidities, and age of the implantable device. The use of such additional information improves the predictions of the neural network.

[0064] The neural network receiving training attenuation data; inputting the training attenuation data into the neural network; For each of the training images, using a neural network to predict a value of adhesion volume between the implantable device and tissue in contact with the implantable device; adjusting parameters of the neural network based on a difference between the predicted adhesion amount and a ground truth adhesion amount; repeating said predicting and said adjusting until a stopping criterion is met; is trained to predict the amount of adhesion between an implantable device and tissue in contact with the implantable device.

[0065] In this example, the training attenuation data includes a plurality of training images representing an implantable device in an anatomical region and ground truth data having ground truth adhesion values ​​corresponding to the training images, the ground truth adhesion values ​​representing the amount of adhesion between the implantable device and tissue in contact with the implantable device.

[0066] In this example, the training attenuation data defines x-ray attenuation within an anatomical region. The training attenuation data includes training images representing the same type of implantable device, e.g., pacemaker leads, for which the neural network is making predictions. The training attenuation data may be generated by the same type of x-ray imaging system, i.e., a volumetric, projection, or spectral imaging system, as the attenuation data 110 for which the neural network is making predictions. Alternatively, the training attenuation data may be simulated attenuation data generated by a different type of imaging system, which is processed to resemble data generated by the type of x-ray imaging system for which the neural network is making predictions. For example, volumetric "CT" images may be projected to resemble projection x-ray images. Training images may be curated from records of past implantable device extraction procedures. The training data may be generated from tens, hundreds, or thousands of such procedures, and the training data may represent subjects of various ages, genders, and body mass index "BMI" values.

[0067] The corresponding ground truth data can be generated automatically or by an expert, determined automatically by quantifying the amount of adhesion between the implantable device and the tissue in contact with the implantable device using the techniques described above, with adhesion classes or adhesion scores, or the expert can label regions of images in the training attenuation data with ground truth values ​​that represent the amount of adhesion between the implantable device and the tissue in contact with the implantable device.

[0068] For example, if additional information such as age, sex, comorbidities, and age of implanted device is input into the neural network to generate its predictions, the corresponding values ​​of this information are also input into the neural network during training.

[0069] Generally, training a neural network involves inputting a training dataset into the neural network and iteratively adjusting the neural network's parameters until the trained neural network provides accurate outputs. Training is often performed using a graphics processing unit (GPU) or a dedicated neural processor, such as a neural processing unit (NPU) or a tensor processing unit (TPU). Training often employs a centralized approach, using cloud- or mainframe-based neural processors to train the neural network. Following training with the training dataset, the trained neural network is deployed to a device for analyzing new input data during inference. Processing requirements during inference are significantly less than those required during training, allowing neural networks to be deployed on a variety of systems, such as laptop computers, tablets, and mobile phones. Inference is performed on a server or in the cloud by, for example, a central processing unit (CPU), GPU, NPU, or TPU.

[0070] Thus, the process of training the neural network described above involves adjusting the neural network's parameters. The parameters, or more specifically, the weights and biases, control the behavior of the activation function in the neural network. In supervised learning, the training process automatically adjusts the weights and biases so that when presented with input data, the neural network accurately provides the corresponding expected output data. To do this, a loss function or error value is calculated based on the difference between the predicted output data and the expected output data. The value of the loss function can be calculated using functions such as, for example, negative log-likelihood loss, mean absolute error (or L1 norm), mean squared error, root mean squared error (or L2 norm), Huber loss, or (binary) cross-entropy loss. During training, the value of the loss function is typically minimized, and training terminates when the value of the loss function meets a stopping criterion. In some cases, training terminates when the value of the loss function meets one or more of several criteria.

[0071] Various methods are known for solving loss minimization problems, such as gradient descent and quasi-Newton algorithms. Various algorithms have been developed to implement these methods and variations of them, including, but not limited to, stochastic gradient descent (SGD), batch gradient descent, mini-batch gradient descent, Gauss-Newton algorithm, Levenberg-Marquardt algorithm, momentum methods, Adam, Nadam, Adagrad, Adadelta, RMSProp, and Adamax "optimizers." These algorithms use the chain rule to calculate the derivatives of the loss function with respect to the model parameters. This process is called backpropagation because derivatives are calculated starting from the last layer, or output layer, and moving toward the first layer, or input layer. These derivatives inform the algorithm how the model parameters must be adjusted to minimize the error function. That is, adjustments to the model parameters are made starting from the output layer and working backward through the network until the input layer is reached. In the first training iterations, the initial weights and biases are often randomized. The neural network then predicts output data, which is also random. Backpropagation is then used to adjust the weights and biases. The training process is iterative, adjusting the weights and biases at each iteration. Training ends when the error, or difference, between the predicted output data and the expected output data is within an acceptable range for the training or validation data. The neural network is then deployed, and the trained neural network makes predictions for new input data using the trained values ​​of its parameters. If the training process is successful, the trained neural network accurately predicts the expected output data from the new input data.

[0072] In any of the three techniques described above for determining the amount of adhesion between an implantable device and tissue in contact with the implantable device, the amount of adhesion is determined for one or more locations on the surface of the device. For example, in the example of a pacemaker lead, the amount of adhesion is determined at the distal end of the lead or for multiple locations on the surface of the distal end of the pacemaker lead. Some examples of implantable devices, such as pacemaker leads, have an elongated shape, and adhesions at locations along the length of the implantable device are relevant to determining guidance information for extracting the implantable device. Thus, in one example, operation S120 of analyzing attenuation data 110 to determine the amount of adhesion 130 between an implantable device 120 and tissue 140 in contact with the implantable device is performed for multiple locations along the length of the implantable device.

[0073] In a related example, a total amount of adhesion between an implantable device and tissue in contact with the implantable device is determined. In this example, the method described with reference to FIG. 3 includes determining a total amount of adhesion between an implantable device 120 and tissue 140 in contact with the implantable device based on an amount of adhesion 130 between the implantable device 120 and tissue 140 in contact with the implantable device at multiple locations along the implantable device, and outputting S130 guidance information for an extraction procedure of the implantable device is based on the total amount of adhesion.

[0074] In this example, the total adhesion volume is determined, for example, by integrating the adhesion volumes at multiple locations along the implantable device, and provides an indication of the overall difficulty of the implantable device extraction procedure.

[0075] Returning to FIG. 3 , in operation S130, guidance information is output. This operation is performed for each of the three techniques described above for determining the amount of adhesion between the implantable device and tissue in contact with the implantable device. Various types of guidance information are output in operation S130, and the guidance information may be output in various ways. In some examples, the guidance information is output in a visual format. For example, in one example, the guidance information is output to a display. The guidance information may be output, for example, to the display 220 shown in FIG. 4 . The guidance information may alternatively be output audibly, or in other ways, including output to a printer, or output to a computer-readable storage medium.

[0076] As mentioned above, various types of guidance information can be output in operation S130. As some examples, outputting guidance information for an implantable device extraction procedure (S130) can include: the amount of adhesion 130 between the implantable device 120 and the tissue 140 in contact with the implantable device; the anticipated difficulty of the implantable device extraction procedure; the expected duration of the implantable device extraction procedure; Outcome metrics for implantable device extraction procedures; Recommended Device 170 for Use in Implantable Device Extraction Procedures; Recommended device settings for use in implantable device extraction procedures; and Recommended procedural steps for use during implantable device extraction procedures outputting one or more indicators of:

[0077] FIG. 5 is an example of a projection X-ray image 110 including guidance information for an implantable device extraction procedure in the form of an indication of the amount of adhesion 130 between an implantable device 120 and tissue 140 in contact with the implantable device, according to some aspects of the present disclosure.

[0078] 5, the implantable device is a pacemaker lead, and guidance information is superimposed on the projection x-ray 110. In the example shown, the projection x-ray image 110 is generated from attenuation data 110 received in act S110. However, the projection x-ray image 110 may alternatively be generated from other types of attenuation data representing the implantable device in the anatomical region. For example, volumetric image data may be projected, with guidance information superimposed thereon, to provide the projected image.

[0079] In the example shown in FIG. 5 , guidance information is superimposed on a projection X-ray image 110 adjacent a pacemaker lead, and the guidance information includes an indication of the amount of adhesion 130 between the pacemaker lead 120 and tissue 140 in contact with the implantable device. In FIG. 5 , the amount of adhesion is indicated at multiple locations along the length of the pacemaker lead. In this example, the guidance information notifies the physician of segments of the pacemaker lead that will be difficult to extract. Referring to the adhesion scale shown in FIG. 5 , the example shown indicates that segments BC and DE have a relatively higher amount of adhesion between the pacemaker lead and the vessel wall than segments AB, CD, and EF. Using this guidance information, the physician can, for example, require additional lead extraction time for segments BC and DE. Based on this information, the physician can also decide to select a specific device, such as a sheath with a cutting blade or laser at its distal end, rather than a locking stylet, to extract the pacemaker lead for these segments.

[0080] As mentioned above, other types of guidance information may alternatively or additionally be output in operation S 130. Figure 6 is a schematic diagram illustrating a display 220 including example guidance information 150, 160, 170 for an extraction procedure of an implantable device, according to some aspects of the present disclosure.

[0081] In one example, the act of outputting guidance information for an implantable device extraction procedure S130 includes outputting an indication of the expected difficulty of the implantable device extraction procedure.

[0082] The expected difficulty of the implantable device extraction procedure is determined based on the adhesion volume 130 using a functional relationship. For example, the functional relationship between the expected difficulty and the adhesion volume 130 can be used to assign an expected difficulty to a portion of the implantable device or to the entire implantable device. The functional relationship may be provided, for example, by a graph, a look-up table, or a decision tree. For example, an indication of the expected difficulty for the entire implantable device can be obtained by mapping the average adhesion volume 130 along the length of the pacemaker lead to a difficulty scale of 1 to 10 using the functional relationship in graphical form.

[0083] Alternatively, the expected difficulty of the implantable device extraction procedure can be determined based on the adhesion volume 130 using a neural network that predicts the adhesion volume between the implantable device and the tissue in contact with the implantable device. In this case, the neural network can use the expected adhesion volume between the implantable device and the tissue in contact with the implantable device as an input for further predicting the difficulty. Alternatively, the neural network can directly predict the difficulty, in which case the difficulty is implicitly predicted based on the adhesion volume because the neural network is trained using training attenuation data that includes ground truth values ​​representing the adhesion volume between the implantable device and the tissue in contact with the implantable device. When a neural network is used to determine the expected difficulty of the implantable device extraction procedure, additional training data in the form of ground truth values ​​for the difficulty are also used to train the neural network. The ground truth values ​​may be provided by an expert, for example, a physician who has performed previous procedures for which training attenuation data was curated.

[0084] In another example, the operation S130 of outputting guidance information for the implantable device extraction procedure includes outputting an indication of an expected duration 160 of the implantable device extraction procedure. The expected duration 160 of the implantable device extraction procedure is determined in a manner similar to the expected difficulty. In other words, in one example, the expected duration 160 of the implantable device extraction procedure is determined based on the adhesion volume 130 using a functional relationship. For example, an indication of the expected duration 160 is obtained by mapping the integral of the adhesion volume 130 along the length of the pacemaker lead to the duration of the procedure using the functional relationship in graphical form.

[0085] Alternatively, the expected duration 160 can be determined based on the adhesion volume 130 using a neural network that predicts the amount of adhesion between the implantable device and the tissue in contact with the implantable device. In this case, the neural network can use the predicted adhesion volume between the implantable device and the tissue in contact with the implantable device as input to further predict the expected duration 160. Alternatively, the neural network can directly predict the expected duration 160, in which case the expected duration 160 is implicitly predicted based on the adhesion volume because the neural network is trained using training attenuation data that includes ground truth values ​​representing the amount of adhesion between the implantable device and the tissue in contact with the implantable device. When a neural network is used to determine the expected duration 160, additional training data in the form of ground truth values ​​for the duration are also used to train the neural network. The ground truth values ​​can be known from previous procedures in which training attenuation data was curated.

[0086] In another example, the operation S130 of outputting guidance information for the implantable device extraction procedure includes outputting an indication of an outcome metric for the implantable device extraction procedure. The outcome metric can represent various factors. Examples of outcome metrics output in this example include the probability of the procedure succeeding or failing, or the probability of a medical complication resulting from the procedure. The outcome metric is determined in a manner similar to the expected difficulty. In other words, in one example, the outcome metric is determined based on the adhesion volume 130 using a functional relationship. For example, the indicator of the outcome metric is obtained by mapping the maximum value of the adhesion volume 130 along the length of the pacemaker lead to the outcome metric using the functional relationship in the form of a look-up table. The maximum value can represent the most difficult part of the extraction procedure and, therefore, can be the indicator of the outcome.

[0087] Alternatively, the outcome metric can be determined based on the adhesion volume 130 using a neural network that predicts the amount of adhesion between the implantable device and the tissue in contact with the implantable device. In this case, the neural network can use the predicted adhesion volume between the implantable device and the tissue in contact with the implantable device as input for further predicting the outcome metric. Alternatively, the neural network can predict the outcome metric directly, in which case the outcome metric is implicitly predicted based on the adhesion volume because the neural network is trained using training attenuation data that includes ground truth values ​​representing the amount of adhesion between the implantable device and the tissue in contact with the implantable device. When a neural network is used to determine the outcome metric, additional training data in the form of ground truth values ​​of the outcome metric are also used to train the neural network. The ground truth values ​​can be known from previous procedures in which training attenuation data was curated.

[0088] In another example, the operation S130 of outputting guidance information for the implantable device extraction procedure includes outputting an indication of a recommended device 170 for use in the implantable device extraction procedure. Examples of the recommended device output in this example include a recommendation to use the laser device shown in FIG. 2 or the mechanical rotation dilator sheath described above. The recommended device 170 is determined in a manner similar to the expected difficulty. In other words, in one example, the recommended device 170 is determined based on the adhesion amount 130 using a functional relationship. For example, the indication of the recommended device 170 is obtained using a functional relationship in the form of a look-up table or a decision tree by mapping the maximum adhesion amount 130 along the length of the pacemaker lead or the entire length of the pacemaker lead where the adhesion amount 130 exceeds a threshold to the recommended device 170.

[0089] Alternatively, the recommended device 170 is determined based on the adhesion volume 130 using a neural network that predicts the adhesion volume between the implantable device and the tissue in contact with the implantable device. In this case, the neural network can use the predicted adhesion volume between the implantable device and the tissue in contact with the implantable device as input for further predicting the recommended device 170. Alternatively, the neural network can directly predict the recommended device 170, in which case the recommended device 170 is implicitly predicted based on the adhesion volume because the neural network is trained using training attenuation data that includes ground truth values ​​representing the adhesion volume between the implantable device and the tissue in contact with the implantable device. When a neural network is used to determine the recommended device 170, additional training data in the form of ground truth values ​​for the recommended device 170 are also used to train the neural network. The ground truth values ​​may be known from previous procedures in which training attenuation data was curated.

[0090] In another example, the operation S130 of outputting guidance information for an implantable device extraction procedure includes outputting an indication of recommended device settings for use in the implantable device extraction procedure. Examples of recommended device settings output in this example include the power level of the device's laser shown in FIG. 2 or the rotation speed of the mechanical rotary dilator sheath described above. The recommended device settings are determined in a manner similar to the expected difficulty. In other words, in one example, the recommended device settings are determined based on the adhesion amount 130 using a functional relationship. For example, the indication of recommended device settings is obtained by mapping the adhesion amount 130 along the length of the pacemaker lead to the recommended device settings using a functional relationship in graphical form.

[0091] Alternatively, the recommended device settings may be determined based on the adhesion volume 130 using a neural network that predicts the amount of adhesion between the implantable device and the tissue in contact with the implantable device. In this case, the neural network can use the predicted adhesion volume between the implantable device and the tissue in contact with the implantable device as input to further predict the recommended device settings. The neural network can directly predict the recommended device settings, in which case the recommended device settings are implicitly predicted based on the adhesion volume because the neural network was trained using training attenuation data that includes ground truth values ​​representing the amount of adhesion between the implantable device and the tissue in contact with the implantable device. When a neural network is used to determine the recommended device settings, additional training data in the form of ground truth values ​​of the recommended device settings are also used to train the neural network. The ground truth values ​​may be known from previous procedures in which training attenuation data was curated.

[0092] In another example, the operation S130 of outputting guidance information for an implantable device extraction procedure includes outputting an indication of recommended procedure steps for use in the implantable device extraction procedure. Examples of recommended procedure steps output in this example include "advance extraction device at (specified) speed" and "withdraw extraction device 5 centimeters, rotate (specified) angle, then advance tool again." The recommended procedure steps are determined in a manner similar to the recommended device settings. In other words, in one example, the recommended procedure steps are determined based on adhesion volume 130 using a functional relationship. For example, an indication of extraction device advancement speed can be obtained by mapping values ​​of adhesion volume 130 along the length of the pacemaker lead to a recommended advancement speed using a functional relationship in graphical form.

[0093] Alternatively, the recommended procedural steps are determined based on the adhesion volume 130 using a neural network that predicts the adhesion volume between the implantable device and the tissue in contact with the implantable device. In this case, the neural network can use the predicted adhesion volume between the implantable device and the tissue in contact with the implantable device as input for further predicting the recommended procedural steps. Alternatively, the neural network can directly predict the recommended procedural steps, in which case the recommended procedural steps are implicitly predicted based on the adhesion volume because the neural network is trained using training attenuation data that includes ground truth values ​​representing the adhesion volume between the implantable device and the tissue in contact with the implantable device. When a neural network is used to determine the recommended procedural steps, additional training data in the form of ground truth values ​​of the recommended procedural steps are also used to train the neural network. The ground truth values ​​can be known from previous procedures in which training attenuation data was curated.

[0094] Other types of guidance information may also be provided. In another example, an indication of the value of a risk metric is output. In this example, the method described with reference to FIG. analyzing the attenuation data 110 to determine a value of a risk metric 150, the value of the risk metric representing a risk of damaging the tissue represented in the attenuation data 110 and / or a risk of fracturing the implantable device; Outputting guidance information for an implantable device extraction procedure based on adhesion volume 130 (S130) includes outputting an indication of the value of risk metric 150.

[0095] In this example, the tissue represented in the attenuation data 110 may be, for example, a lung, a heart valve, or a blood vessel wall. A value of a risk metric representing the risk of damaging the tissue represented in the attenuation data 110 is determined based, for example, on the type of tissue surrounding the implantable device 120 and the distance between the implantable device and the tissue. In this example, a list of tissue types is identified and labeled with a degree of susceptibility to damage. The value of the risk metric may be determined by dividing the degree of susceptibility to damage by the distance to the tissue. The distance is measured in an image generated from the attenuation data 110. For example, the distance may be measured in a projection X-ray image or a volumetric X-ray image. Thus, nearby tissues considered to be highly susceptible to damage are associated with a high risk metric. In one example, one or more organs at risk, such as the lungs, are identified in the attenuation data 110 using segmentation techniques. If a tissue represented in the attenuation data 110 is located between the implantable device and the organ at risk, the tissue is assigned a relatively higher value of the risk metric 150 than if the implantable device is located between the tissue and the organ at risk. As a result, for example, the risk of damaging a blood vessel wall is deemed higher for the portion of the blood vessel wall on the lung-facing side of a pacemaker lead than for the portion of the blood vessel wall on the opposite side of the pacemaker lead. Tissue types are identified, for example, by segmenting the attenuation data 110 and labeling the tissue type using an anatomical diagram. Alternatively, if the attenuation data 110 is spectral attenuation data, different tissue types can be identified by applying a material decomposition algorithm to the attenuation data 110. The risk metric can also be predicted using the neural network described above. In this case, additional training data in the form of ground truth values ​​for the risk metric are also used to train the neural network. For example, an expert can label training images in the training data with ground truth values ​​for the risk metric.

[0096] Alternatively, in this example, a value of a risk metric representing the risk of fracturing the implantable device can be determined. Over time, an implantable device, such as a pacemaker lead, weakens and, as a result, becomes more susceptible to fracturing during device removal. In this case, the risk metric is determined from the attenuation data, for example, by identifying cracks in the implantable device and / or significant kinking along the length of the implantable device. The value of the risk metric can also be predicted using the neural network described above. In this case, additional training data in the form of ground truth values ​​for the risk metric are also used to train the neural network. For example, an expert can label portions of the implantable device in the training data that fractured during device removal.

[0097] In another example, implantable device-induced inflammation in an anatomical region is identified. Identifying such inflammation is relevant to implantable device extraction procedures because it informs the physician of areas that require additional attention during extraction, as well as the urgency of the procedure. In the example of a pacemaker lead, it can also indicate the point at which the lead needs to be extracted.

[0098] In this example, the method described with reference to FIG. analyzing the attenuation data 110 to determine the presence of implantable device-induced inflammation in the anatomical region; outputting an indication of the presence of inflammation induced by the implantable device; Includes:

[0099] In this example, the presence of implantable device-induced inflammation is determined based on measurements made within images generated from the attenuation data. For example, inflammation is determined based on pathological observations such as swollen intimal or wall thickening accompanied by increased X-ray attenuation. Image-based measurements of the size of such areas are compared to baseline values ​​to determine the presence of such inflammation. In examples where the attenuation data 110 includes spectral attenuation data, so-called "Z" attenuation data is used.eff An "image," i.e., an image derived from the spectral CT data representing effective atomic number, is also generated and used to identify inflammation based on the presence of areas with abnormally high values ​​of effective atomic number.

[0100] In another example, the method described with reference to FIG. analyzing the attenuation data 110 to determine a stability metric for the implantable device; Outputting an index of this stability metric Includes:

[0101] In this example, the stability metric represents the mechanical stability of the implantable device. In the example of a pacemaker lead, the mechanical stability of the lead is affected by cracks in the lead, as described above, and also by the strength of the lead's attachment to the cardiac tissue at its distal end. Mechanical stability also affects the probability of a successful extraction procedure and the selection of extraction devices and techniques. The stability metric is determined by analyzing images generated from the attenuation data 110 to identify such cracks, as described above, or alternatively, by determining the depth of insertion of the distal end of the pacemaker lead into the cardiac tissue. For example, a pacemaker lead that has a crack or is only slightly inserted into the cardiac tissue is considered unstable. Similarly, an IVC filter may have cracks around its periphery. The stability metric is used by a physician to determine where additional attention should be paid along the length of the pacemaker lead or around the IVC filter.

[0102] In some situations, different techniques or extraction tools may be better suited to removing different segments of a pacemaker lead. For example, a rotating dilator sheath may be better suited to removing a length of pacemaker lead along the vasculature, while a locking stylet may be better suited to removing the distal end of the pacemaker lead from cardiac tissue. In other situations, a pacemaker lead that is difficult to extract may be abandoned. For example, if the risks of removing the lead outweigh the benefits, the pacemaker lead may be abandoned entirely, or cut along its length so that a portion of the lead is abandoned within the body. In such situations, it is useful to recommend the length of the pacemaker lead to be extracted. In one example, the method described with reference to FIG. 3 includes: calculating a recommended length for the implantable device (120) to be extracted during an implantable device extraction procedure, the recommended length being calculated based on the amount of adhesion (130) between the implantable device (120) and tissue (140) in contact with the implantable device, and / or based on a stability metric and / or the presence of inflammation induced by the implantable device; outputting an indication of the length of the recommendation; Includes:

[0103] In this example, different tools may be recommended for implantable device removal based on the amount of adhesion 130. For example, if the amount of adhesion 130 between the implantable device and the contacting tissue is low, a single tool may be recommended for extraction of the entire device. In contrast, if the amount of adhesion 130 along the length of the pacemaker lead is relatively low and the amount of adhesion 130 at the distal end of the pacemaker lead is relatively high, a first tool may be recommended for extraction of the length of the pacemaker lead and a second tool may be recommended for extraction of the distal end. Alternatively, a segment of the pacemaker lead with weak or moderate adhesion 130 between the implantable device 120 and the tissue 140 contacting the implantable device may be recommended for extraction, while a segment of the pacemaker lead with strong adhesion 130 between the implantable device 120 and the tissue 140 contacting the implantable device may be recommended for discarding. The recommended length of the implantable device may be shown, for example, as an overlay on an image of the implantable device. The stability metric and the presence of inflammation are also factors that influence the recommended length of lead extraction, for example, it may be recommended to remove a portion of the lead close to the inflammation, or not to remove a portion of the lead if the lead is deemed unstable.

[0104] In one example, the output adhesion amount 130 between the implantable device 120 and the tissue 140 in contact with the implantable device is used to generate a control signal for a medical robot to perform an implantable device extraction procedure. The adhesion amount 130 is used to generate a control signal for guiding and / or actuating an interventional tool controlled by the medical robot. The interventional tool may be, for example, the TightRail mechanical rotational dilator sheath described above or the GlideLight Laser sheath described above. The control signal may, for example, guide the interventional tool to a region of tissue where the adhesion amount 130 between the implantable device 120 and the tissue 140 in contact with the implantable device exceeds a threshold. The control signal may, for example, control actuation of the interventional tool such that the interventional tool separates the implantable device from the tissue in contact with the implantable device. The control signal may provide a relatively higher level of actuation in areas where the tissue has a relatively greater amount of adhesion 130 between the implantable device 120 and the tissue 140 in contact with the implantable device than in areas where the tissue has a relatively lesser amount of adhesion 130 between the implantable device 120 and the tissue 140 in contact with the implantable device. The magnitude of actuation may be modulated based on the value of the risk metric to reduce the risk of damaging the tissue.

[0105] In another example, a computer program product is provided having instructions that, when executed by one or more processors, cause the one or more processors to perform a procedure for providing guidance information for an extraction procedure of an implantable device, the method comprising: receiving (S110) attenuation data 110 representing an implantable device 120 in an anatomical region, the attenuation data defining x-ray attenuation within the anatomical region; Analyzing the attenuation data 110 to determine (S120) an amount of adhesion 130 between the implantable device 120 and tissue 140 in contact with the implantable device; outputting guidance information for an implantable device extraction procedure based on the adhesion amount (S130); It has.

[0106] In another example, a system 200 for providing guidance information for an implantable device extraction procedure is provided, the system comprising: receiving (S110) attenuation data 110 representing an implantable device 120 in an anatomical region, the attenuation data defining x-ray attenuation within the anatomical region; analyzing the attenuation data 110 to determine (S120) an amount of adhesion 130 between an implantable device 120 and tissue 140 in contact with the implantable device; outputting guidance information for an implantable device extraction procedure based on the adhesion amount (S130); The one or more processors 210 are configured to:

[0107] An example of a system 200 is shown in Figure 4. It should be noted that the system 200 may include one or more of an imaging system for generating the attenuation data 110 received in operation S110, such as, for example, a CT imaging system 230 shown in Figure 4, a display 220 for displaying output guidance information, etc., a patient bed 240, and a user input device configured to receive user input, such as, for example, a keyboard, a mouse, a touch screen, etc.

[0108] An enumerated list of embodiments of the present disclosure is provided below.

[0109] Example 1 1. A computer-implemented method for providing guidance information for an implantable device extraction procedure, the method comprising: receiving S110 attenuation data 110 representing an implantable device 120 in an anatomical region, the attenuation data defining x-ray attenuation within the anatomical region; analyzing the attenuation data to determine an amount of adhesion between the implantable device and tissue in contact with the implantable device; outputting guidance information for an extraction procedure of the implantable device based on the adhesion amount 130; 10. A computer-implemented method comprising:

[0110] Example 2 Analyzing the attenuation data 110 to determine S120 an amount of adhesion 130 between an implantable device 120 and tissue 140 in contact with the implantable device includes: determining an attenuation value in Hounsfield units of tissue 140 in contact with the implantable device; determining an amount of adhesion 130 between the implantable device 120 and tissue 140 in contact with the implantable device based on the attenuation value in Hounsfield units; 2. The computer-implemented method according to Example 1,

[0111] Example 3 The attenuation data 110 comprises spectral attenuation data, the spectral attenuation data being arranged in a plurality of different energy intervals DE 1...m 2. The computer-implemented method of Example 1, further comprising: defining the x-ray attenuation within the anatomical region as:

[0112] Example 4 Analyzing the attenuation data 110 to determine S120 an amount of adhesion 130 between an implantable device 120 and tissue 140 in contact with the implantable device includes: applying a material decomposition algorithm to the spectral attenuation data to identify tissue types in contact with the implantable device; determining an amount of adhesion 130 between the implantable device 120 and tissue 130 in contact with the implantable device based on the tissue type; 4. The computer-implemented method according to Example 3,

[0113] Example 5 5. The computer-implemented method of any one of Examples 1 to 4, wherein analyzing the attenuation data 110 to determine S120 the amount of adhesion 130 between an implantable device 120 and tissue 140 in contact with the implantable device is performed for multiple locations along the length of the implantable device.

[0114] Example 6 The method includes determining a total amount of adhesion between the implantable device 120 and the tissue 140 in contact with the implantable device based on an amount of adhesion 130 between the implantable device 120 and the tissue 140 in contact with the implantable device at the plurality of locations along the implantable device; and outputting S130 guidance information for an extraction procedure of the implantable device based on the total adhesion volume. The computer-implemented method according to Example 5.

[0115] Example 7 The method further comprises analyzing the attenuation data 110 to determine a value of a risk metric 150, the value of the risk metric representing a risk of damaging tissue represented in the attenuation data 110 and / or a risk of fracturing the implantable device; 7. The computer-implemented method of any one of Examples 1 to 6, wherein outputting guidance information for an extraction procedure for the implantable device based on the adhesion volume 130 S130 comprises outputting an indication of a value of the risk metric 150.

[0116] Example 8 The method comprises: Analyzing the attenuation data 110 to determine S120 an amount of adhesion 130 between the implantable device 120 and tissue 140 in contact with the implantable device includes: inputting the received attenuation data 110 into a neural network; and using the neural network to predict the amount of adhesion 130 between the implantable device 120 and tissue 140 in contact with the implantable device; 2. The computer-implemented method according to Example 1, wherein the neural network is trained to predict an amount of adhesion between the implantable device and a tissue in contact with the implantable device using attenuation data having a plurality of training images representing the implantable device in the anatomical region and ground truth data including ground truth adhesion values ​​corresponding to the training images, the ground truth adhesion values ​​representing an amount of adhesion between the implantable device and a tissue in contact with the implantable device.

[0117] Example 9 outputting guidance information for an extraction procedure for the implantable device S130; the amount of adhesion 130 between the implantable device 120 and tissue 140 in contact with the implantable device; the anticipated difficulty of the extraction procedure for said implantable device; the expected duration of the implantable device extraction procedure 160; outcome metrics for the implantable device extraction procedure; a recommended device 170 for use in the implanted device extraction procedure; a recommended device configuration for use in the implantable device extraction procedure; and Suggested procedural steps for use in an extraction procedure for said implantable device. 9. The computer-implemented method of any one of Examples 1 to 8, further comprising outputting one or more indicators of:

[0118] Example 10 10. The computer-implemented method of any one of Examples 1 to 9, wherein analyzing the attenuation data 110 to determine an amount of adhesion 130 between the implantable device 120 and tissue 140 in contact with the implantable device S120 comprises quantifying the amount of adhesion between the implantable device and tissue in contact with the implantable device in one or more adhesion classes or adhesion scores.

[0119] Example 11 The method comprises: analyzing the attenuation data 110 to determine the presence of implantable device-induced inflammation in the anatomical region; outputting an indication of the presence of inflammation induced by said implantable device; 11. The computer-implemented method of any one of Examples 1 to 10, further comprising:

[0120] Example 12 The method comprises: analyzing the attenuation data 110 to determine a stability metric for the implantable device; outputting an indication of said stability metric; 12. The computer-implemented method of any one of Examples 1 to 11, further comprising:

[0121] Example 13 The method comprises: calculating a recommended length for the implantable device 120 to be extracted during the implantable device extraction procedure, the recommended length being calculated based on the amount of adhesion 130 between the implantable device 120 and tissue 140 in contact with the implantable device, and / or based on the stability metric, and / or the presence of inflammation induced by the implantable device; outputting an indication of the length of the recommendation; 13. The computer-implemented method of Example 12, further comprising:

[0122] Example 14 14. The computer-implemented method of any one of Examples 1 to 13, wherein the method further comprises segmenting the received attenuation data 110 and identifying at least the implantable device 120 before analyzing the attenuation data 110 in S120.

[0123] Example 15 15. The computer-implemented method of any one of Examples 1 to 14, wherein the implantable device 120 comprises a pacemaker lead, or an implantable cardioverter-defibrillator lead, or a pacemaker, or an implantable cardioverter-defibrillator, or an IVC filter.

[0124] The above examples should be understood as illustrative of the present disclosure, not limiting. Further examples are contemplated. For example, examples described in connection with a computer-implemented method may be provided in a corresponding manner by a computer program product, by a computer-readable storage medium, or by one or more processors of system 200. It should be understood that features described with respect to any one example may be used alone or in combination with other described features, and may be used in combination with one or more features of another example or in combination with other examples. Furthermore, equivalents and modifications not described above may also be used without departing from the scope of the invention as defined in the appended claims. In the claims, the word "comprising" does not exclude other elements or operations, and a lack of a plurality does not exclude a plurality. The mere fact that certain features are recited in mutually different dependent claims does not indicate that a combination of these features cannot be advantageously used. Any reference signs in the claims should not be construed as limiting their scope.

Claims

1. 1. A system for providing guidance information for an implantable device extraction procedure, the system comprising: receiving attenuation data representative of an implantable device in an anatomical region, the attenuation data defining x-ray attenuation within the anatomical region; analyzing the attenuation data to determine the amount of adhesion between the implantable device and tissue in contact with the implantable device; outputting guidance information for an extraction procedure of the implantable device based on the amount of adhesion; 1. A system having one or more processors configured to:

2. Analyzing the attenuation data to determine an amount of adhesion between the implantable device and tissue in contact with the implantable device includes: determining an attenuation value in Hounsfield units of tissue in contact with the implantable device; determining an amount of adhesion between the implantable device and the tissue in contact with the implantable device based on the attenuation value in Hounsfield units; The system of claim 1 , comprising:

3. the attenuation data comprises spectral attenuation data; the spectral attenuation data defining x-ray attenuation within the anatomical region at a plurality of different energy intervals; The system of claim 1 .

4. Analyzing the attenuation data to determine an amount of adhesion between the implantable device and tissue in contact with the implantable device includes: applying a material decomposition algorithm to the spectral attenuation data to identify tissue types in contact with the implantable device; determining an amount of adhesion between the implantable device and tissue in contact with the implantable device based on the tissue type; The system of claim 3 , comprising:

5. 5. The system of claim 1, wherein analyzing the attenuation data to determine the amount of adhesion between the implantable device and tissue in contact with the implantable device is performed for multiple locations along the length of the implantable device.

6. the one or more processors are configured to determine a total amount of adhesion between the implantable device and tissue in contact with the implantable device based on the amount of adhesion between the implantable device and tissue in contact with the implantable device at the multiple locations along the implantable device; outputting guidance information for an extraction procedure of the implantable device based on the total adhesion volume; The system of claim 5.

7. the one or more processors are further configured to analyze the attenuation data to determine a value of a risk metric, the value of the risk metric representing a risk of damaging tissue represented in the attenuation data and / or a risk of fracturing the implantable device; and outputting guidance information for an extraction procedure of the implantable device based on the amount of adhesions, the guidance information including outputting an indication of the value of the risk metric.

7. The system according to claim 1, comprising:

8. Analyzing the attenuation data to determine an amount of adhesion between the implantable device and tissue in contact with the implantable device includes: inputting the received attenuation data into a neural network; using the neural network to predict the amount of adhesion between the implantable device and tissue in contact with the implantable device; 2. The system of claim 1, wherein the neural network is trained to predict an amount of adhesion between the implantable device and tissue in contact with the implantable device using training attenuation data comprising a plurality of training images representing an implantable device in an anatomical region and ground truth data having ground truth adhesion values ​​corresponding to the training images, the ground truth adhesion values ​​representing an amount of adhesion between the implantable device and tissue in contact with the implantable device.

9. outputting the guidance information for an extraction procedure of the implantable device, the amount of adhesion between the implantable device and tissue in contact with the implantable device; the anticipated difficulty of the implantable device extraction procedure; the expected duration of the implantable device extraction procedure; outcome metrics for the implantable device extraction procedure; a recommended device for use in the implantable device extraction procedure; recommended device settings for use in the implantable device extraction procedure; and Suggested procedural steps for use in an implantable device extraction procedure.

9. The system of claim 1, further comprising outputting one or more indicators of:

10. 10. The system of claim 1, wherein analyzing the attenuation data to determine the amount of adhesions between the implantable device and tissue in contact with the implantable device comprises quantifying the amount of adhesions between the implantable device and tissue in contact with the implantable device in one or more adhesion classes or adhesion scores.

11. the one or more processors analyzing the attenuation data to determine the presence of implantable device-induced inflammation in the anatomical region; outputting an indication of the presence of inflammation induced by said implantable device; The system of claim 1 , further configured to:

12. the one or more processors analyzing the attenuation data to determine a stability metric for the implantable device; outputting an indication of said stability metric; The system of claim 1 , further configured to:

13. the one or more processors calculating a recommended length for extracting the implantable device during the extraction procedure, the recommended length being calculated based on the amount of adhesion between the implantable device and tissue in contact with the implantable device, and / or based on the stability metric and / or the presence of inflammation induced by the implantable device; outputting an indication of the length of the recommendation; The system of claim 12 , further configured to:

14. 14. The system of claim 1, wherein the one or more processors are further configured to segment the received attenuation data to identify at least the implantable device before analyzing the attenuation data.

15. 15. The system of claim 1, wherein the implantable device comprises a pacemaker lead, or an implantable cardioverter-defibrillator lead, or a pacemaker, or an implantable cardioverter-defibrillator, or an IVC filter.

16. When executed by one or more processors, the one or more processors:

1. A computer program product having instructions for causing a procedure to be performed to provide guidance information for an extraction procedure of an implantable device, the method comprising: receiving attenuation data representative of an implantable device in an anatomical region, the attenuation data defining x-ray attenuation within the anatomical region; analyzing the attenuation data to determine the amount of adhesion between the implantable device and tissue in contact with the implantable device; outputting guidance information for an extraction procedure of the implantable device based on the amount of adhesion; 1. A computer program product comprising:

17. 1. A computer-implemented method for providing guidance information for an implantable device extraction procedure, the method comprising: receiving attenuation data representative of an implantable device in an anatomical region, the attenuation data defining x-ray attenuation within the anatomical region; analyzing the attenuation data to determine the amount of adhesion between the implantable device and tissue in contact with the implantable device; outputting guidance information for an extraction procedure of the implantable device based on the amount of adhesion; 10. A computer-implemented method comprising:

Images