METHOD AND SYSTEM FOR ONLINE CALIBRATION OF AN X-RAY MEDICAL DEVICE

Abstract

The invention relates to a system and a method for calibrating a device D comprising at least one radiation source (3) and a detector (4), the radiation source (3) and the detector (4) being mounted on at least one moving support (2), characterized in that it comprises at least the following elements: • At least one first inertial sensor (21c) disposed near the radiation source (3) and at least one second inertial sensor (22c) disposed near the detector (4), the first and second inertial sensors being adapted to measure a position Ps of the source and a position Pd of the detector, • A synchronization module (23) adapted to synchronously trigger the measurement of positions and the acquisition of images, • A processing module (25) for the measured positions,said processing module comprising an input (25e) receiving a model M of device operation and a data fusion algorithm (27) taking into account at least the two measured positions Ps, Pd, and the model M to estimate a precise position value for the source and for the detector.

Description

METHOD AND SYSTEM FOR ONLINE CALIBRATION OF A MEDICAL DEVICE A X-RAYS The invention relates in particular to a method and system for the online calibration of an X-ray medical device, and more specifically to radiology systems used for surgery or interventional radiology, such as surgical C-arms. It is notably used for the online calibration of C-arm devices. It is common to use a mobile interventional radiology system to perform surgical or interventional procedures. These systems, also called mobile C-arms (surgical C-arms or C-arm amplifiers), allow the surgeon to acquire X-ray images during the procedure and to control the positioning of instruments (catheters, needles, prostheses, etc.) in real time in a minimally invasive manner. Most of these systems can obtain two-dimensional images with a video image stream of up to thirty frames per second. These systems generally consist of an X-ray source and an image detector mounted at both ends of a C-arm, between which the object to be imaged is positioned. The images acquired by the detector are the result of projecting the object onto the detector plane. The practitioner then uses these 2D images to perform a real-time mental reconstruction of the patient's anatomy, allowing them to position the instrument in real time relative to the area to be operated on. They also mentally reorient the entire scene (patient + surgical instruments) in space to precisely control their movements. More recently, sophisticated systems have emerged. They allow for the acquisition of a 3D image of the surgical instrument during the procedure. The system rotates around the patient to obtain a series of 2D images. These 2D images are These images are then processed by an image reconstruction algorithm to obtain a 3D volumetric image. The reconstruction algorithm requires knowledge of the exact geometry of the C-arm, namely the position of the detector and the X-ray tube relative to the patient or object, for each 2D image used. Current systems offer the option of performing "offline" calibration, using, for example, a 3D calibration target to determine the projection matrices. This "offline" calibration is performed during the preventive maintenance phases of the system, for example, every six months or annually. Patent application EP3141187 relates to a calibration target for geometrically calibrating an X-ray imaging device designed to generate three-dimensional images of an object by reconstructing two-dimensional projections of said object. The calibration target comprises a volumetric support equipped with markers exhibiting radiological absorption in contrast to the volumetric support, the markers being arranged in a three-dimensional pattern. The markers are distributed into subsets along substantially parallel lines, such that cross-ratio sequences can be constructed from the respective marker subsets.Each sequence of cross-ratios comprises a unique cross-ratio for each set of four markers in which the markers are ordered according to an order dependent on the order numbers of the respective markers along the line along which they are aligned, according to a first predefined direction, said order being common to all cross-ratios. A calibration technique known from prior art, for example, uses markers positioned on a phantom that serve as reference points in space. Since the position of the markers in space is known, it is possible to deduce the geometry of the acquisition for each projection by inverting a system of equations derived from the position of the markers on the projected images. Previous systems assume that rotational acquisition is sufficiently repeatable so that the geometry determined "offline" can be applied to images acquired in real time during the procedure. The mechanics of these systems therefore had to be improved to stabilize the C-arm during the rotational acquisition of 2D images (reduced mechanical play, more rigid components, etc.). These systems are not widely used because their cost is high due to the necessary mechanical modifications. Furthermore, implementing these modifications is not straightforward for manufacturers of C-arm devices. Some known prior art solutions propose to perform "online" calibration with a method based on sensors directly integrated on the device and without resorting to the analysis of images containing radio-opaque markers. One solution uses a simple three-axis inertial sensor positioned on the detector or on the ray source. Such a method is described in the document by Grzeda Victor et al, entitled "C-arm rotation encoding with accelerometer", International Journal of Computer Assisted Radiology and Surgery, 2010, 5(4), pp: 385-391. A second solution uses two six-axis inertial measurement units and two laser rangefinders, as for example in the document by Amiri Shahram, Wilson David R., entitled "A low-cost tracked C-arm (TC-arm) upgrade system for versatile quantitative intraoperative imaging", International Journal of Computer Assisted Radiology and Surgery, 9(4), pp:695-711, 2014. In applications requiring very high precision, such as in surgery, these methods are not optimal. The precision achieved is incompatible with the desired quality of 3D reconstruction. In Amiri Sharham's document, with a single three-axis accelerometer, the relocation accuracy does not exceed one degree of accuracy, despite precise calibration of the device. With a setup comprising two six-axis inertial measurement units and two laser rangefinders, document by Grzeda Victor, the accuracies obtained are on the order of: 1.5 mm + / - 1.2 mm accuracy for isocenter location system, 2.3 mm + / - 1.1 mm accuracy for 2D-3D registration, 4.4 mm + / - 1.9 mm marker positioning accuracy in the Reconstructed scene. These accuracy values ​​are significantly below the accuracy values ​​required for precise and meticulous applications, particularly in surgical applications, or in other fields requiring precision, for example in metrology of parts at the end of manufacturing. The idea of ​​the present invention relates to a calibration method and system for estimating the online geometry of a device during operation. Calibration is performed during the device's rotational acquisition so as not to constrain device handling procedures and, in particular, to avoid the need for cumbersome and time-consuming prior (offline) calibrations. The invention finds particular application in the field of surgical operations, and more broadly in clinical use, in a non-intrusive manner and with a rapid execution time. In the description, the terms "apparatus" and "device" refer to the same object. The term "geometry" of the system refers to the set of nine intrinsic parameters (characterization of the X-ray source / detector pair) and extrinsic parameters (characterization of the geometry of the patient / X-ray pair / detector pair). The invention relates to a calibration system for a device D comprising at least one radiation source and one detector, the radiation source and the detector being mounted on at least one support made of movement, the moving support being linked to a base via at least one axis of rotation characterized in that it comprises at least the following elements: At least one initial sensor placed near the source of radiation and at least a second sensor placed near the detector, the first and second sensors being adapted to to measure a position Ps of the source and a position Pd of the detector, A synchronization module adapted to trigger in a way synchronous measurement of positions and image acquisition, A module for processing measured positions, said module of a process including an input receiving a model M of operation of the device and a data fusion algorithm taking into account at least the two measured positions Ps, Pd, and the model M to estimate a precise position value for the source and for the detector. Sensors suitable for measuring position include, for example, distributed inertial sensors, for example, along the medium between the radiation source and the detector. The system can include three inertial sensors, two of the inertial sensors being positioned at each end of the moving support and the third sensor halfway between the two ends. The system may include one encoder to measure the movement of the arm. Alternatively, it may include N encoders, with N greater than or equal to 1, each encoder being adapted to measure an angular position Pa of the arm's movement, the number N being equal to the number of degrees of freedom of the system. The system is, for example, equipped with a Kalman or extended Kalman algorithm to fuse data. The moving support is a C-shaped arm. The invention also relates to a method for calibrating a medical device D comprising at least one radiation source and a detector associated with at least one moving support characterized in that it comprises at least the following steps: o Trigger a set of position measurements at a given time t and simultaneously : Measure at least one initial position value Ps for the source and a second position value Pd for the detector, relative to a reference point, using at least two sensors adapted to measure position values, o Transmit this information, along with an M-model of operation of device D to a data fusion step in order to determine a exact position value for the source and an exact position value of the detector. The step of merging all the data is, for example, carried out using a Kalman or extended Kalman filter. The fusion step includes the fusion of measurements given by the inertial sensors with the measurement of N encoders, with N greater than or equal to 1, said encoder being adapted to measure a position Pa, the number N being equal to the number of degrees of freedom of the system. For example, an optoelectronic sensor is used to measure the mechanical deformations of the arm, or one or more Bragg gratings or interferometers are used to determine a deformation of the moving support. To measure the mechanical deformation of the support, a device chosen from the following list is used: one or more laser interferometers or rangefinders, one or more RFIP / Ultrasound sensors. Other features and advantages of the present invention will become clearer upon reading the attached descriptions of exemplary embodiments and figures, which represent: Figure 1, a schematic representation of an imaging device of the C-arm type, or C-arm Figure 2 shows a diagram of a data processing module, and Figure 3, a synoptic diagram of the steps implemented during the calibration according to the invention. To better illustrate the purpose of the present invention, the following example is given for a C-arm type device used in surgery. Without departing from the scope of the invention, the invention could be implemented in any device comprising at least one radiation source and one detector, the source and detector being mounted on one or more moving supports, the device being calibrated online (during operation) to optimize measurement results. In Figure 1, the device D shown comprises a base 1 on which a C-shaped arm 2 is mounted. The C-shaped arm supports, at one end 2, a radiation source 3, for example, an X-ray tube, and at the other end 2, an X-ray detector 4, which provides a 2D image. The C-shaped arm is positioned in a hollow arc-shaped slide 6, in which it slides in an orbital rotation R1, the axis A1 of which is the center of a circle partially represented by the arc-shaped arm. The slide is fixed to the base 1 via a retaining piece 8 and an arm 7 that provides a second angular rotation R2 about a second axis A2. For the online calibration of the device, the device includes a first sensor such as an inertial sensor 21c located near the source 3 and at least a second sensor such as an inertial sensor 22c located near the detector 4. The two position sensors 21c, 22c are adapted to measure the position Ps of the source and the position Pd of the detector according to principles known to those skilled in the art. The sensor suitable for measuring position can be an inertial sensor, an ultrasonic sensor, an optical sensor, an encoder, or any other device known to a person skilled in the art that performs the function of position measurement. The device may also include an encoder-type sensor 20 adapted to measure the angular position Pa of the orbital movement of the arm at point C at a given time t. Using an encoder allows for a simpler and more reliable measurement. Examples of sensor types will be provided later in the description. In order to carry out the measurements, the device is connected to a control and data processing module 10 (figure 2) comprising the following elements: a synchronization module 23 enabling synchronous triggering of measurements at the level of the sensors equipping the arm, a processor 24 executing the steps of the process according to the invention, a processing module 25 of the measured positions. The synchronization module 23 allows simultaneous time-stamping of image acquisitions and Ps, Pd, Pa measurements from the inertial sensors and the encoder on the arms. The measurement results are presented, for example, in the form of a table or a database containing, for a given time ti, a 2D image acquisition and three positions Ps, Pd, Pa. According to Figure 2, the processing module includes an input 25e receiving an operating model 26 of the device and a data fusion algorithm 27, of the Kalman type for example or of the extended Kalman type, taking into account at least the three measured positions Ps, Pd, Pa (measurements of the two inertial sensors and the encoder), 28, and the operating model of the device 26 to deduce a projection matrix which will be transmitted via an output 25s of the processing module to a reconstruction module. The number of encoders used is chosen, for example, based on the number of degrees of freedom (rotation-translation) of the device. The encoders' function is, in particular, to determine, with low precision but in a stable manner, the absolute position of the device relative to a reference frame attached to the device's base (the device's coordinates on the C-arm to which the casters are fixed, or a reference point). example), and thus to know a first approximation of the intrinsic and extrinsic parameters or even the associated 4x3 3D / 2D projection matrix which will be used according to techniques known to a person in the art to reconstruct a 3D image. To determine the position of the arm "C" we will use, for example, a potentiometric linear positioning rule for the "orbital" movement R1, an optical encoder wheel for the "angular" movement R2 and two rectilinear movements for the movements T1 ("forward / backward") and T2 ("up / down"), figure 1. For the movement of the arm, the latter can be equipped with a flexible toothed belt and the slide 6, support of the arm, with a toothed wheel which drives the flexible belt. Figure 3 represents a flowchart listing the sequence of steps implemented by the process. In the following example, the method uses, for instance, the encoder 20 to record the position of the arm in "C" and at least the two inertial sensors positioned at each end of the arm, 300 (Figure 3). The measurements are taken 302 synchronously thanks to the synchronization module 302, which transmits a trigger command for the measurements simultaneously to the encoder and the two inertial sensors. The measurement Pa taken by the encoder and the two measurements Ps and Pd taken by the two inertial sensors 21c, 22c, 302, are transmitted to a data processing module, 303. In this example, a Kalman filter is used. The Kalman filter, 305, receives as input a model M of the device's operation, 304, and the three positions Ps, Pd, and Pa from the data processing module. From the fusion of these data, the Kalman filter estimates a position of the X-ray source and a precise position of the detector, 306. The accuracies of the positions obtained will, for example, be in the range [100 µm–500 µm]. The 3D / 2D projection matrix will then be calculated from the precise positions. To increase the accuracy of the position obtained, the device can be equipped with an array of MEMS inertial sensors with six or nine degrees of freedom placed at chosen positions. Due to their low cost and ease of implementation, a large number of inertial sensors can be used. For example, one solution is to position inertial sensors in the following way: A first reference inertial sensor linked to the system reference frame, An inertial sensor on the rotation point of arm "C" (at the level of the mechanical coupling to achieve orbital motion), An inertial sensor! on the midpoint of arm "C", or distributed over all the way along the arm, Three or four inertial sensors at the detector and the source Rx fixed to both ends of arm “C”. The 22c sensor is, for example, an inertial sensor with a minimum of six degrees of freedom: three accelerometers and three gyroscopes, the 21c sensor is an inertial sensor with a minimum of three degrees of freedom: three accelerometers. In Figure 1, a third Inertial Sensor 23c is positioned halfway between the two inertial sensors 21c, 22c. According to an alternative embodiment, as described below, the Kalman filter can also receive additional measurements (Figure 3, 307), such as measurements of the deformations of the C-arm, obtained using one or more Bragg gratings (not shown for simplicity). These gratings can be positioned along the C-arm on its faces or edges to measure the arm's deformation. Another solution for increasing accuracy is to use one or more Mach-Zehnder or Michelson interferometers. These will allow measurements via one or more flexible rods, for example non-extensible, non-compressible, and placed in a sheath fitted along the C-shaped arm. By merging all the data: measurement data obtained by encoders, inertial sensors, Bragg gratings, interferometers, all the parameters can be obtained with better accuracy. To reduce the number of electronic sensors required to measure deformation in all spatial directions, one approach involves merging data from all sensors with the results of the mechanical deformation modeling of the "C". Indeed, depending on the design of the "C", it may be sufficient to measure only a limited number of deformations while maintaining sufficient accuracy in the estimation. According to one embodiment, the combination of one or more laser rangefinders will allow for highly precise measurement, for example on the order of microns, of the deviations in three dimensions of the displacement of the C-arm relative to the rotation mechanism of the C-arm on the orbital motion relative to its theoretical trajectory. Another variant is to use a laser scanning system that measures in real time a portion of the profile of the C-arm. Another alternative to optical measurement solutions (interferometer, laser rangefinder) is to use a combination of RFID sensor combined with ultrasonic sensors to measure the same deviations; these solutions based on relative "time of flight" measurements are less precise than optical measurements but also less expensive and therefore better suited to the targeted final system. A triangulation operation based on the measurements will then allow us to measure the displacements of the mechanical component holding the "C" arm and enabling its orbital rotation relative to the reference frame linked to the device's reference point. This movement combines the "forward / backward" and "up / down" displacements. Any other precision tracking system can be used for this additional measurement, for example, an optical system, a 2D or 3D laser scanning system, etc. The extrinsic parameters, as defined, are relative to a reference frame linked to the device's chassis and not to the table on which the patient is located. The transformation between the two reference frames can be easily performed by a person skilled in the art. The invention enables online, real-time calibration of the device, ensuring that the position of the source and detector is always precisely known. It allows for the online determination of the intrinsic and extrinsic geometric parameters of a C-arm, thus accurately deducing the 3D / 2D projection matrix. This matrix enables the creation of high-quality 3D tomographic reconstructions of a patient and also allows for precise positioning of the C-arm during surgery.

Claims

DEMANDS 1 - Calibration system for a device D comprising at least one radiation source (3) and one detector (4), the radiation source (3) and the detector (4) being mounted on at least one moving support (2), the moving support (2) being linked to a base (1) via at least one axis of rotation characterized in that it comprises at least the following elements: At least one first sensor (21c) placed near the source radiation (3) and at least a second sensor (22c) disposed at proximity of detector (4), the first two and second sensors being adapted to measure a position Ps of the source and a position Pd of the detector, A synchronization module (23) adapted to trigger in a synchronous measurement of positions and image acquisition, A processing module (25) for the measured positions, said module of processing including an input (25th) receiving a model M of operation of the device and a data fusion algorithm (27) taking into account at least the two measured positions Ps, Pd, and the M model to estimate a precise position value for the source and for the detector. 2 - System according to claim 1 characterized in that said sensors adapted to measure positions are inertial sensors (21c, 22c, 23c). 3 — System according to claim 2 characterized in that said inertial sensors are distributed along the support (2) between the radiation source (3) and the detector (4). 4 — System according to claim 3 characterized in that it comprises at least three inertial sensors (21c, 22c, 23c), two of the inertial sensors being positioned at each end (24, 22) of the moving support (2) and the third sensor (23c) halfway between the two ends (24, 22). 5 — System according to any one of the preceding claims characterized in that the moving support (2) is an arm and in that it comprises an encoder (20) for measuring the movement of the arm. 6 — System according to any one of claims 1 to 4 characterized in that the moving support (2) is an arm and in that it comprises N encoders (20), with N greater than or equal to 1, said encoder (20) being adapted to measure an angular position Pa of the movement of the arm, the number N being equal to the number of degrees of freedom of the system. — A system according to any one of the preceding claims characterized in that the data fusion algorithm is a Kalman or extended Kalman algorithm. 8 — System according to any one of the preceding claims characterized in that the moving support (2) is a C-shaped arm. 9 — Method for online calibration of a medical device D comprising at least one radiation source (3) and one detector (4) associated with at least one moving support (2) characterized in that it comprises at least the following steps: o Trigger a set of position measurements at a given time t and simultaneously : Measure at least one initial position value Ps for the source and a second position value Pd for the detector, relative to a reference point, using at least two sensors adapted to to measure a position, o Transmit this information, along with an M-model of operation of device D to a data fusion step in order to determine a exact position value for the source and an exact position value of the detector. 10 — Method according to claim 9 characterized in that an extended Kalman or Kaiman filter is used to fuse the data set. 11 - Method according to any one of claims 9 or 10 characterized in that it fuses the measurements given by the sensors with the measurement of N encoders (20), with N greater than or equal to 1, said encoder (20) being adapted to measure a position Pa, the number N being equal to the number of degrees of freedom of the system. 12 — A method according to any one of claims 9 to 11 characterized in that inertial sensors are used to measure position. 13 - Method according to any one of claims 9 to 12 characterized in that the support of the movement (2) is an arm and in that an optoelectronic sensor is used to measure the mechanical deformations of the arm. 14 — Method according to claim 9 characterized in that one or more Bragg gratings or interferometers are used to determine a deformation of the moving support. 15 - Method according to any one of claims 9 to 12 characterized in that to measure the mechanical deformation of the support (2) a device is chosen from the following list: one or more laser interferometers and rangefinders, one or more RFIP / Ultrasound sensors.