Method and system for improving calibration accuracy in image-guided surgeries

Abstract

A method 800 for improving calibration accuracy in an imaging device 102 is disclosed The method comprises acquiring at least one 2D image of an anatomical region along with a plurality of fiducial markers arranged on a calibration drum 108. The method further includes detecting the 2D positions of the fiducial markers from the acquired image and computing a error metric for each marker using either a centroid-based deviation approach, or a camera projection matrix (CPM)-based reprojection analysis. The error metric is then used to assign a weight to each fiducial marker, and a subset of fiducial markers having weights above a predefined threshold is selected. Calibration parameters of the imaging device 102 are updated based on the selected subset of fiducial markers. The method 800 enhances the accuracy and reliability of intraoperative image guidance by intelligently refining the calibration process through optimization of fiducial selection.

Description

METHOD AND SYSTEM FOR IMPROVING CALIBRATION ACCURACY IN IMAGE-GUIDED SURGERIESFIELD OF INVENTION

[0001] The present invention generally relates to the field of image-guided surgeries. More specifically, the present invention is relating to a method and a system for improving calibration accuracy in image-guided surgeries.BACKGROUND OF THE INVENTION

[0002] The subject matter discussed in the background section should not be assumed to be prior art merely because of its mention in the background section. Similarly, a problem mentioned in the background section or associated with the subject matter of the background section should not be assumed to have been previously recognized in prior art. The subject matter in the background section merely represents different approaches, which in and of themselves may also correspond to implementations of the claimed technology.

[0003] In recent years, spinal surgery has undergone significant advancements with the integration of image-guided technologies. Such innovations have proven to be particularly transformative in Minimally Invasive Surgery (MIS), where precise localization of spine and enhanced intraoperative visualization are critical for successful outcomes. The introduction of Image-Guided Surgical Systems (IGSS) has revolutionized the field of spinal surgery by providing real-time navigation and visualization capabilities during complex procedures. Such systems are specifically designed to improve the accuracy of surgical interventions, thereby enhancing patient safety, improving clinical outcomes, and optimizing the overall surgical workflow.

[0004] Traditionally, several imaging modalities have been employed in the IGSS, such as preoperative Computed Tomography (CT)-based methods, intraoperative fluoroscopy, and 3D fluoroscopy. Of these, fluoroscopy-based systems are most commonly used due to their ability to provide continuous, real-time imaging, which is indispensable during surgical interventions. A typical fluoroscopy-based system includes a C-arm, a versatile imaging device capable of rotating around the patient to capture multiple imaging angles. To enhance theaccuracy and precision of fluoroscopy-based systems, specialized accessories, such as calibration drums, are integrated into the setup.

[0005] The calibration drums are precision-engineered devices composed of parallel plastic plates embedded with metallic fiducials. Such fiducials are strategically arranged in a geometric pattern within the drum in a predefine spatial configuration and serve as crucial reference points for the system. When the C-arm captures an image, these fiducials are projected onto the fluoroscopic image as distinct markers. These markers are subsequently detected by specialized image processing algorithms, enabling the system to calculate the spatial relationship between the C-arm and the patient's anatomy, thereby enabling accurate alignment of image space with physical space.

[0006] The calibration drums play a vital role in ensuring both distortion correction and the accurate calibration of the system's parameters. Distortion correction eliminates any geometric inaccuracies inherent in the imaging process, ensuring that the captured images represent true anatomical structures. Calibration, on the other hand, involves the precise determination of both intrinsic parameters, such as focal length and optical center, and extrinsic parameters, such as the position and orientation of the C-arm. Such accurate calibration is fundamental for generating both 2D and 3D representations of the patient's anatomy, which are essential for effective navigation and surgical planning, especially in complex spinal procedures. The calibration drums, by providing a stable and consistent set of reference points, enable the system to maintain high levels of accuracy, thus improving the safety and efficacy of minimally invasive spinal surgeries.

[0007] In robot-assisted surgeries, where precision is paramount, accurate calibration becomes even more critical, as even minor deviations can lead to suboptimal or adverse surgical outcomes. The fidelity of the IGSS heavily depends on accurate detection and identification of fiducial markers. Failures in detecting these markers can result in compromised registration between the patient’s anatomy and the displayed image, leading to navigation errors and reduced surgical precision. Existing systems generally treat all fiducials with equal importance and rely on a fixed detection strategy, without accounting for markerspecific variations in detection reliability or contribution to calibration accuracy.

[0008] Traditionally, it was assumed that the inclusion of all calibration fiducials in the imaging process would minimize error metric, thereby ensuring optimal calibration accuracy.However, clinical data testing revealed a surprising outcome: in several instances, the error metric remained minimal even when not all fiducials were successfully identified in the image. Moreover, the number of unidentified fiducials varied from image to image, resulting in corresponding fluctuations in the error metric. These findings led to the hypothesis that different fiducials contribute unequally to the overall calibration accuracy. Contributing factors such as low image contrast, the presence of white patches due to air gaps over fiducials, and occlusion caused by surgical instruments further impede accurate detection. In such cases, the non-identification of certain fiducials may significantly impact system calibration and, by extension, the overall accuracy of the IGSS.

[0009] Thus, there is a need for a method to improve calibration accuracy in image-guided surgeries to overcome the above-mentioned challenges.OBJECTS OF THE INVENTION

[0010] An object of the present invention is to improve accuracy in image-guided surgeries by optimising the calibration process to adjust the impact of each fiducial based on error metric.

[0011] Another object of the present invention is to improve the accuracy of fiducial detection and the overall calibration process by assigning dynamic weights to fiducials based on their individual contributions to the error metric, allowing for a more robust and precise surgical setup.

[0012] Yet another object of the present invention is to provide a method for selecting a subset of fiducial markers with higher weights, thus improving the reliability of calibration by utilizing the most accurate fiducials during the procedure.

[0013] Yet another object of the present invention is to enable real-time correction of fiducial distortions and improve the alignment of imaging device with patient’s anatomy, ensuring that the surgical navigation system provides precise guidance throughout the procedure.

[0014] Yet another object of the present invention is to by automating fiducial weighting and selection process and reducing the need for manual intervention and improving the overall efficiency of the calibration procedure.

[0015] Yet another object of the present invention is to improve patient safety and surgical outcomes by providing accurate, dynamic calibration that adapts to changes in fiducial positioning, ensuring precise intraoperative navigation even in complex surgical scenarios.SUMMARY OF THE INVENTION

[0016] This summary is provided to introduce aspects related to the present invention of a method and a system for improving calibration accuracy in image-guided surgeries and the aspects are further described below in the detailed description. This summary is not intended to identify essential features of the claimed subject matter nor is it intended for use in determining or limiting the scope of the claimed subject matter.

[0017] In an embodiment of the present disclosure, a method for improving calibration accuracy in imaging devices is disclosed. The method comprises acquiring at least one two- dimensional (2D) image including an anatomical region of a patient and a plurality of fiducial markers by an imaging device. The plurality of fiducial markers corresponds to visual representation of plurality of fiducials embedded on a calibration drum. The method further comprises detecting 2D positions of the plurality of fiducial markers from the acquired 2D image by an image processing unit. The method further comprises calculating an error metric for each of the plurality of fiducial markers by the image processing unit using an optimization algorithm. The method further comprises assigning a weight to each of the plurality of fiducial markers based on corresponding error metric by the image processing unit. The method further comprises selecting a subset of fiducial markers having weights above a predefined threshold value by the image processing unit the predefined threshold value corresponds to a weight associated with a lower range of the error metric. The method further comprises calibrating one or more parameters of the imaging device based on the selected subset of fiducial markers to improve accuracy.

[0018] In an aspect of the present disclosure, the calibration drum is positioned within a field of view of the imaging device.

[0019] In another aspect of the present disclosure, the calibration drum comprises 81 fiducial markers, including 17 calibration fiducial markers and 64 distortion fiducial markers.

[0020] In another aspect of the present disclosure, calculating the error metric for each of the plurality of fiducial markers comprises determining a centroid of the detected 2D positionsof the plurality of fiducial markers in the at least one acquired 2D image by the image processing unit, calculating a deviation for each of the fiducial marker from the centroid of the detected 2D positions by the processing unit and determining the error metric by quantifying the deviation from the centroid by the image processing unit.

[0021] In another aspect of the present disclosure, calculating the error metric for each of the plurality of fiducial markers comprises projecting the known 3D positions of each fiducial marker into 2D positions on image plane based on the predefined spatial configuration by the image processing unit, comparing the detected 2D positions of each fiducial marker present in the acquired 2D image to the corresponding projected 2D positions of each fiducial marker by the image processing unit and determining the error metric as the difference between the detected 2D positions and the projected 2D positions of each fiducial.

[0022] In another aspect of the present disclosure, the projecting of the known 3D positions of each fiducial marker into the 2D positions on the image plane is performed using a camera projection matrix (CPM).

[0023] In another aspect of the present disclosure one or more parameters comprise one or more of intrinsic parameters, including focal length, principal point, and distortion coefficients of the imaging device and extrinsic parameters, including camera position and camera orientation of the imaging device.

[0024] In an embodiment of the present disclosure, a system to improve calibration accuracy of imaging devices is disclosed. The system comprises an image processing unit a memory coupled with the image processing unit. The memory stores program instructions. The imaging processing unit configured to acquire at least a two-dimensional (2D) image including an anatomical region of a patient and a plurality of fiducial markers. The plurality of fiducial markers corresponds to visual representation of plurality of fiducials embedded on a calibration drum. The imaging processing unit further configured to detect positions of the plurality of fiducial markers from the acquired 2D image. The imaging processing unit further configured to calculate an error metric for each of the plurality of fiducial markers. The imaging processing unit further configured to assign a weight to each the plurality of fiducial marker based on corresponding error metric. The imaging processing unit further configured to select a subset of fiducial markers having weights above a predefined threshold value. The predefined threshold value corresponds to a weight associated with a lower range of the error metric. Theimaging processing unit further configured to calibrate the one or more parameters of the imaging device based on the selected subset of fiducial markers.

[0025] In an aspect of the present disclosure, the calibration drum is positioned within a field of view of the imaging device.

[0026] In another aspect of the present disclosure, the calibration drum comprises 81 fiducial markers, including 17 calibration fiducial markers and 64 distortion fiducial markers.

[0027] In another aspect of the present disclosure, to calculate the error metric for each of the plurality of fiducial markers, the image processing unit is configured to determine a centroid of the detected 2D positions of the plurality of fiducial markers in the one or more acquired 2D images, calculate a deviation for each of the fiducial marker from the centroid position of the detected 2D positions and determine the error metric by quantifying the deviation from the centroid.

[0028] In another aspect of the present disclosure, to calculate the error metric for each of the plurality of fiducial markers, the image processing unit is configured to project the known 3D positions of each fiducial marker into 2D positions on image plane based on the predefined spatial configuration, compare the detected 2D positions of each fiducial marker present in the acquired 2D image to the corresponding projected 2D positions of each fiducial marker, determine the error metric as the difference between the detected 2D positions and the projected 2D positions of each fiducial marker.

[0029] In another aspect of the present disclosure, the projecting of the known 3D positions of each fiducial marker into the 2D positions on the image plane is performed using a camera projection matrix (CPM).

[0030] In another aspect of the present disclosure one or more parameters comprise one or more intrinsic parameters, including focal length, principal point, and distortion coefficients of the imaging device and extrinsic parameters, including camera position and camera orientation of the imaging device.

[0031] Other aspects and advantages of the invention will become apparent from the following description, taken in conjunction with the accompanying drawings, illustrating by way of example the principles of the invention.BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The accompanying drawings constitute a part of the description and are used to provide further understanding of the present invention. The drawings illustrate exemplary embodiments of the present invention and, together with the description, explain the principles of the present invention.

[0033] Fig. 1 illustrates a working environment of a system to improve calibration accuracy in an imaging device, in accordance with an embodiment of the present invention.

[0034] Fig. 2 illustrates a block diagram of the system to improve the calibration accuracy in the imaging device, in accordance with an embodiment of the present invention.

[0035] Fig. 3 illustrates a schematic diagram of a method to improve calibration accuracy in the imaging device, in accordance with an embodiment of the present invention.

[0036] Figs. 4a, 4b, 4c and 4d cumulatively illustrate the error metric analysis of the fiducial markers, in accordance with an embodiment of the present invention.

[0037] Figs. 5a and 5b illustrate tables depicting quantitative analysis of the error metric using weights obtained with fiducial co-ordinate -based optimization and CPM based optimization method respectively, in accordance with an embodiment of the present invention.

[0038] Fig. 6a and 6b cumulatively presents a comparative analysis of performance of the proposed methods in minimizing the error metric across images acquired from two different imaging devices, in accordance with an embodiment of the present invention.

[0039] Figs. 7a and 7b cumulatively illustrate the results of a clinical phantom study conducted to validate the efficacy of the proposed methods implemented, in accordance with an embodiment of the present invention.

[0040] Fig. 8 illustrates a flowchart of a method for improving calibration accuracy in the imaging device, in accordance with an embodiment of the present invention.

[0041] A more complete understanding of the present invention and its embodiments thereof may be acquired by referring to the following description and the accompanying drawings.DETAILED DESCRIPTION OF THE INVENTION

[0042] Exemplary embodiments now will be described with reference to the accompanying drawings. The disclosure may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey its scope to those skilled in art. The terminology used in the detailed description of the exemplary embodiments illustrated in the accompanying drawings is not intended to be limiting. In the drawings, like numbers refer to like elements.

[0043] It is to be noted, however, that the reference numerals used herein illustrate only typical embodiments of the present subject matter, and are therefore, not to be considered for limiting its scope, for the subject matter may admit to other equally effective embodiments.

[0044] The specification may refer to “an”, “another”, “one” or “some” embodiment s) in several locations.

[0045] This does not necessarily imply that each such reference is to the same embodiment(s), or that the feature only applies to a single embodiment. Single features of different embodiments may also be combined to provide other embodiments.

[0046] As used herein, the singular forms “a”, “an” and “the” are intended to include the plural forms as well, unless expressly stated otherwise. It will be further understood that the terms “include”, “comprises”, “including” and / or “comprising” when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It will be understood that when an element is referred to as being “connected” or “coupled” to another element, it can be directly connected or coupled to the other element or intervening elements may be present. Furthermore, “connected” or “coupled” as used herein may include operatively connected or coupled. As used herein, the term “and / or” includes all combinations and arrangements of one or more of the associated listed items.

[0047] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skills in the art to which this disclosure pertains. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent withtheir meaning in the context of the relevant art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.

[0048] The detailed description includes specific details for the purpose of providing a thorough understanding of the present invention. However, it will be apparent to those skilled in the art that the present invention may be practiced without these specific details.

[0049] The present invention provides a method and a system for enhancing the calibration accuracy of medical imaging devices, such as mobile C-arm systems, by dynamically evaluating and selecting fiducial markers based on their reliability. Traditional calibration techniques treat all detected fiducials equally, without accounting for distortions, occlusions, or spatial inconsistencies, which may introduce inaccuracies in critical image-guided surgical workflows. The invention overcomes this drawback by implementing an intelligent calibration optimization mechanism where a error metric is computed for each fiducial marker. This error metric is derived using one of two selectable methods, a centroid deviation-based technique or a reprojection error-based technique using a camera projection matrix (CPM). Based on the computed error metrics, the system selects an optimal subset of fiducials that contribute positively to calibration accuracy. These selected markers are then used to recalibrate imaging parameters. The invention ensures higher fidelity in intraoperative imaging, enabling more precise anatomical localization, and ultimately supporting safer and more effective surgical interventions.

[0050] Fig. 1 illustrates a working environment of a system 100 to improve calibration accuracy in an imaging device 102, in accordance with an embodiment of the present invention. The working environment may comprise the imaging device 102 positioned to acquire at least one Two Dimensional (2D) intraoperative or preoperative images of a patient 104 supported on a surgical table and a plurality of fiducial markers. The imaging device 102 may include a C-arm or any suitable radiological apparatus configured to capture fluoroscopic or radiographic images during a surgical procedure. The imaging device 102 is operatively coupled to the system 100 via a communication network 106 that facilitates bi-directional exchange of data and control signals.

[0051] The communication network 106 may be a wired and / or a wireless network. The communication network 106 may be implemented using communication techniques such as Visible Light Communication (VLC), Worldwide Interoperability for Microwave Access(WiMAX), Long Term Evolution (LTE), Wireless Local Area Network (WLAN), Infrared (IR) communication, Public Switched Telephone Network (PSTN), Radio waves, and other communication techniques known in the art.

[0052] The communication network 106 may enables transmission of 2D imaging data from the imaging device 102 to the system 100 for processing, and concurrently allows the system 100 to transmit calibration commands, feedback signals, or visualization overlays back to the imaging device 102. Such a networked setup ensures seamless integration between imaging acquisition and real-time optimization processes without interrupting clinical workflow.

[0053] The system 100 may comprises a memory and a processor operatively coupled to each other. The system 100 may be server implemented remotely over a cloud network or locally. The memory may store image data, imaging device parameters, calibration information, optimization algorithms, and instructions executable by the processor to perform operations including fiducial detection, parameter computation, error metric estimation, optimization-based weight estimation, weight-based fiducial selection and recalibration. The processor may be further configured to execute such instructions for enhancing the spatial accuracy of the imaging device 102.

[0054] The working environment may further include a calibration drum 108 comprising a plurality of fiducials. The plurality of fiducials may be embedded on the surface of the calibration drum 108 and are configured to be detectable by the imaging device 102 when positioned within the imaging field. The plurality of fiducials may be distributed in predefined spatial arrangements to enable precise detection and calculation of intrinsic and extrinsic parameters of the imaging device 100. The imaging device 102 captures the plurality of fiducial markers which are visual representation of the plurality of the fiducials present in the calibration drum 108 and such image data is transmitted to the system 100 for processing.

[0055] The system 100 may be communicatively coupled to a user device 110 that facilitates interaction with a user 112. The user device 110 may include a desktop, laptop, tablet, or any computing interface capable of displaying imaging outputs, system feedback, or recalibration results. The user 112 may include, without limitation, a surgeon, a radiologist, or a clinical technician responsible for overseeing imaging operations and reviewing calibration results. In one implementation the user 112 may be the patient itself. Through the user device110, the user 112 may visualize fiducial overlays, receive alerts regarding fiducial markers visibility or accuracy metrics, and interact with optimization modules for real-time calibration adjustments.

[0056] In one embodiment, the patient 102 may be a human subject undergoing a spine- related surgical procedure where imaging precision is of critical importance. The patient 104 may be positioned on a radiolucent surgical platform to facilitate unobstructed imaging. During the imaging process, the calibration drum 108 is positioned within the field of view of the imaging device 102, allowing the system 100 to analyze fiducial projections and evaluate associated error metric using one or more optimization techniques. The system 100 may further configured to assess whether the currently utilized fiducials contribute optimally to the accuracy of the imaging device 100, and selectively perform recalibration using the one or more optimization techniques. The one or more optimization techniques may include fiducial coordinates-based optimization and Camera Projection Matrix (CPM)-based optimization.

[0057] Fig. 2 illustrates a block diagram of the system 100 to improve calibration accuracy in the imaging device 102, in accordance with an embodiment of the present invention. The system 100 may be implemented as a server either locally or remotely over a networked environment, such as a hospital server infrastructure or a cloud-based imaging analytics platform. The system 100 may comprise one or more network interfaces 202 (e.g., wired Ethernet, wireless modules, etc.), the image processing unit 204, and the memory 206. Such subsystems may be interconnected via one or more internal system buses and may be powered by a dedicated or integrated power supply.

[0058] The one or more network interfaces 202 may be configured to establish communication links with external devices, including the imaging device 102 such as a C-arm, surgical planning workstations, and hospital information systems. Such interfaces may be implemented using standard communication protocols and may support interactions via Command-Line Interfaces (CLI), Graphical User Interfaces (GUI), or Application Programming Interfaces (APIs). In one implementation, the one or more network interfaces 202 facilitates bi-directional communication with the imaging device 102, enabling transfer of raw image data for calibration analysis and transmitting updated calibration parameters or alerts back to the system 100.

[0059] The image processing unit 204 may include one or more general purpose processors (e.g., INTEL® or Advanced Micro Devices® (AMD) microprocessors) and / or one or more special purpose processors (e.g., digital signal processors or Xilinx® System On Chip (SOC) Field Programmable Gate Array (FPGA) processor), MIPS / ARM-class processor, a microprocessor, a digital signal processor, an application specific integrated circuit, a microcontroller, a state machine, or any type of programmable logic array.

[0060] The memory 206 may include, but is not limited to, non-transitory machine- readable storage devices such as hard drives, magnetic tape, floppy diskettes, optical disks, Compact Disc Read-Only Memories (CD-ROMs), and magneto-optical disks, semiconductor memories, such as ROMs, Random Access Memories (RAMs), Programmable Read-Only Memories (PROMs), Erasable PROMs (EPROMs), Electrically Erasable PROMs (EEPROMs), flash memory, magnetic or optical cards, or other type of media / machine- readable medium suitable for storing electronic instructions.

[0061] The memory 206 of the system 100 may store program instructions for performing several functions associated with improving calibration accuracy in the imaging device 102. Functional code stored in the memory 206 may include program instructions to acquire a two- dimensional (2D) image including an anatomical region of a patient and a plurality of fiducial markers 208, program instructions to detect the plurality of fiducial markers from the acquired 2D image 210, program instructions to calculate a error metric for each of the plurality of fiducial markers 212, program instructions to assign a weight to each of the plurality of fiducial marker based on corresponding error metric 214, program instructions to select a subset of fiducial markers having weights above a predefined threshold value 216 and program instructions to to calibrate the one or more parameters of the imaging device based on the selected subset of fiducial markers 218.

[0062] The program instructions to acquire a two-dimensional (2D) image including an anatomical region of a patient and a plurality of fiducial markers 208 may cause the image processing unit 204 to receive the at least one 2D image including an anatomical region of a patient 104 and a plurality of fiducial markers via the network interface 202. The 2D image is acquired by the imaging device 102. In one embodiment the imaging device 102 may be a mobile C-arm imaging device. The plurality of fiducial markers visible in the 2D image corresponds to visual representation of plurality of fiducials embedded on the calibration drum 108 in the predefined spatial configuration. During the imaging process, the calibration drum108 is positioned within the field of view of the imaging device 102.Upon receiving the acquired 2D images from the imaging device 102, the image processing unit 204 may initiates a sequence of operations aimed at refining the calibration process.

[0063] The program instructions to detect the plurality of fiducial markers from the acquired 2D image 210 may cause the image processing unit 204 to detect 2D coordinates of the plurality of the fiducial markers from the received 2D image. The image processing unit 204 may identify image coordinates (i.e., pixel positions) of all visible fiducial markers in each view. For such detection purpose the image processing unit 204 may involve conventional thresholding and blob detection techniques or learning-based algorithms trained for fiducial recognition. The detected 2D coordinates are stored in the memory 206 for subsequent evaluation.

[0064] The program instructions to calculate a error metric for each of the plurality of fiducial markers 212 may cause the image processing unit 204 to determine the quality or contribution of each fiducial marker with respect to the accuracy of the calibration process.. In one implementation, the image processing unit 204 may employ fiducial coordinates-based optimization, hereinafter referred to as first optimization method. In another implementation, the image processing unit 204 may employ CPM based optimization, hereinafter referred to as second optimization method.

[0065] In one implementation, the image processing unit 204 implement the first optimization method. In such method the system 100 may determine the centroid of each fiducial marker by determining average of the 2D pixel coordinates detected in the acquired 2D image. Such calculation is performed by analyzing the intensity distribution of the fiducial marker and applying centroid localization techniques such as the center of mass or geometric centroid algorithms. Once the centroid is determined, the image processing unit 204 calculates the deviation for each fiducial marker from the centroid of the detected 2D positions. Such deviation represents the difference between the actual position of each fiducial marker and the centroid. Finally, the image processing unit 204 determines the error metric by quantifying this deviation from the centroid. The error metric serves as an indicator of the calibration accuracy. A lower centroid deviation indicates higher positional accuracy of the fiducial marker in representing the true geometric characteristics of the imaging system. Accordingly, this method enables the system 100 to analyze the spatial distribution and geometric integrity of the plurality of fiducial markers within the imaging frame and compute a corresponding errormetric for each marker. This centroid-based error analysis forms the basis for evaluating marker- specific performance during the calibration refinement. The first optimization method may be used where the user 112 don’t have any details of camera or the user don’t have the knowledge about the working of the camera internally. Such method may be also used when the user wants a quick result. Further such method is used when the image generated by the imaging device 102 is distorted or noisy.

[0066] In another implementation, the image processing unit 204 may utilize a second optimization method. In such method the system 100 estimates the error metric for each fiducial marker. The image processing unit 204 projects the known three-dimensional (3D) positions of each fiducial marker into corresponding 2D positions on the image plane. The 3D positions are predefined based on the spatial configuration of the fiducials on the calibration drum 108 and are stored in the memory 206. The projection process utilizes the current calibration parameters of the imaging device 102 to map each 3D fiducial marker coordinates to its expected 2D positions on the acquired image. Once the projected 2D positions are computed using the CPM model, the image processing unit 204 compares the detected 2D positions of each fiducial marker previously identified in the acquired image with the corresponding projected 2D positions. Such comparison allows the image processing unit 204 to determine the error metric for each fiducial marker by calculating the difference between its detected 2D position and its projected 2D position. The error metric here in this case may also refer as Reprojection error (RPE). The error metric provides a quantitative measure of how accurately the current calibration parameters model the actual imaging geometry and is used to evaluate and refine the calibration of the imaging device 102.

[0067] The program instructions to assign a weight to each of the plurality of fiducial marker based on corresponding optimisation technique 214 may cause the image processing unit 204 to compute and associate a numerical weight or reliability score for each fiducial marker. The weight may be indicative of the accuracy, stability, or suitability of the respective fiducial marker for calibration computations. In the context of the first optimisation method, the assigned weight may be inversely proportional to the centroid deviation; that is, a fiducial marker with a smaller centroid deviation is assigned a higher weight, signifying greater reliability. Conversely, markers exhibiting greater deviation are assigned lower weights, reflecting reduced spatial fidelity. In the context of the second optimisation method, the assigned weight may be inversely related to the RPE, thereby ensuring that markers withminimal reprojection discrepancy receive higher confidence ratings. The computed weights may be stored in the memory 206 for subsequent filtering and marker selection processes.

[0068] The program instructions to select a subset of fiducial markers having weights above a predefined threshold value 216 may cause the image processing unit 204 to apply a threshold-based filtering mechanism, whereby only those fiducial markers whose assigned weights exceed a predefined confidence level are retained for calibration refinement. In one implementation, the weights may be arranged in ascending order and a subset of ten fiducial markers are selected based on their weights. This process is designed to eliminate fiducial markers that may introduce error, distortion, or instability due to poor visibility, occlusion, distortion, or unfavorable imaging conditions. By retaining only high -confidence markers, the system 100 ensures that the final calibration parameters are computed using spatially consistent and geometrically reliable fiducials, thereby enhancing the precision and robustness of the overall calibration process.

[0069] The program instructions to calibrate one or more parameters of the imaging device 102 based on the selected subset of fiducial markers 218 may cause the image processing unit 204 to perform the calibration of the imaging device 102 using only the selected high-weight fiducial markers. The calibration may involve updating the intrinsic parameters, such as focal length, optical center, and distortion coefficients, as well as extrinsic parameters including position and orientation of the imaging device 102 relative to the anatomical region of the patient 104 and the calibration drum 108. By basing the recalibration on an optimally filtered subset of fiducial markers, the system 100 enhances the accuracy of spatial alignment between the imaging device 102 and the patient anatomy, mitigates the influence of unreliable fiducials, and ensures consistent image quality and surgical guidance fidelity throughout the procedure.

[0070] The system 100 may further be operably coupled to the user device 110, such as a surgical planning workstation, desktop computer, tablet, or any network-enabled interface. The user device 110 may provide a graphical user interface (GUI) that enable the user 112 to visualize the calibration results, review the selected subset of fiducial markers, and optionally adjust predefined thresholds or initiate re-calibration sequences. The interface may also allow real-time monitoring of reprojection errors or centroid-based deviations, aiding clinical decision-making. The user device 110 may retrieve processed data from the image processing unit 204 and render the visual feedback through graphical overlays or tabulated summaries, thereby facilitating transparency and control over the calibration enhancement process.

[0071] Fig. 3 illustrates a schematic diagram of a method to improve calibration accuracy in the imaging device 102, in accordance with an embodiment of the present invention. The imaging device 102, such as a C-arm, captures at least a 2D input image of the plurality of fiducials embedded on the calibration drum 108 comprising a total of 81 fiducial markers, 17 designated as calibration fiducials and 64 as distortion fiducials. The spatial distribution of the fiducial markers on the calibration drum 108 is configured to maintain non-coplanarity, which is critical for enabling accurate spatial mapping between the patient space and the image space. The image processing unit 204 extracts the coordinates of the fiducial markers from the input image { [(xi, yi)]17i=i [using a spatio-frequency based detection algorithm. These detected fiducial markers, including both calibration and distortion fiducials, serve as input to the system 100.

[0072] Accordingly, the image processing unit 204 selects the best-performing subset of the fiducials by evaluating their individual contributions to the error metric. A prior analysis is conducted to assess the impact of each fiducial marker on the error metric. Based on this assessment, the weights corresponding to the detected fiducials are computed and sorted in ascending order. Subsequently, the ten fiducials with the highest weights and the ten with the lowest weights are selected for comparative analysis. The system 100 performs error metric comparisons under different calibration conditions. Specifically, the error metric is obtained when all distortion fiducials are used in calibration is compared with the error metric obtained using the top ten fiducials, the bottom ten fiducials, and a selected set of significant fiducials. This analysis enables the identification of the optimal fiducials that result in the minimum error for the given image. The optimal set is then used in subsequent calibration procedures for accurate and reliable imaging.

[0073] The system 100 incorporates two distinct optimization techniques, both aimed at establishing a correlation between the spatial coordinates of the fiducial markers and the error metric. Such methods are designed to facilitate optimal fiducial marker selection by formulating the problem as a matrix-based error minimization task. To achieve this, the image processing unit 204 implements the optimization procedure. This tool enables the system 100 to abstract the problem as a set of weighted error choices, where each fiducial's contribution to the overall error metric is quantified and analyzed. A Least Squares optimization approach is employed to solve this problem formulation, ensuring an efficient and accurate selection of fiducials that minimize the RPE during calibration.minimize f(x) = ||Ax - b ||2(1) where x 6 Rnis obtained, A 6 Rm*nis skinny and full rank (i.e., m > n and Rank (A) = n). The solution of the least-squares problem can be expressed as:(ATA)-1X = ATb (2)The weighted norm approximation is applied as minimize Wx(||Ax - b||) (3)Here, the weight is taken to be single dimensional for a particular set of fiducial coordinates such that Wx E x 6 R". Equation (3) can be approximated as Wx||A|| representing the fiducial position with minimum error metric. The proposed methods may estimate weights regardless of whether a CPM is present. In the former, the centroid is used to determine the point with the least distortion and RPE, whereas in the latter method involving the CPM, the least RPE point is computed.

[0074] In the first optimization method (Fiducial Co-ordinates Based Optimization), The objective function in this method involves only the fiducial coordinates and the average coordinates for each fiducial. Thus, the weights are determined for the x and y coordinates of the fiducials using the following objective function:Here, (x, y) denotes the coordinates of the fiducials obtained during the fiducial extraction algorithm. The average of each fiducial’s x and y coordinates are represented by XlAverage, • • ., XnAverage and yiAverage, • • • , ynAverage respectively. The minimum RPE fiducial position is considered to be the center of all the fiducial coordinate predictions taken individually as that point has the least distortion.The above weights are estimated to optimize the coordinates together and using these weights, we can infer which fiducial markers have more error and hence contribute to reduction in accuracy. Since one of the primary reasons for the fiducial errors is due to the C- Arm device-induced distortion, the proposed weights would be apt for a particular C-Arm given a suitable amount of data is collected from that C-Arm. The above approach works on the intuition of using the centroid as the minimum error metric fiducial coordinate.

[0075] In the second optimization method (CPM based optimization), This method utilizes the reprojected fiducial coordinate using CPM as:where F(x, y) and F’(x, y) are the original points and reprojected points respectively.where PprilJrepresents the reprojected point vector RnThe objective function in this method incorporates the detected points, projected points and the original RPE. Detected points are obtained during the detection of fiducials. The projected points and RPE are obtained during the calculation of a CPM using Direct Linear Transformation (DLT).Here, (x, y) is the detected point and (x’,y’) is the projected point. RPE is the reprojection error previously obtained during calibration. The above minimization becomes more robust as the RPEminimum is computed rather than being picked on using intuition in the previous method.In Equations (4) and (8), n and m represent the number of fiducials and the number of images respectively. The weights wxi,Wx2, • • • ,wxnand wyi,wy2, . . . ,wynrepresent the multiplication factors that the optimization algorithm gives as output for the x and y coordinatesrespectively, to keep the error minimum. The weight of the fiducial is the average of these weights obtained for the x and y coordinates separately. The weights for the individual coordinates are obtained by solving the equations for all the fiducials detected in every image.

[0076] Figs. 4a, 4b, 4c and 4d cumulatively illustrates the error metric analysis of the fiducial markers, in accordance with an embodiment of the present invention. The proposed method was modeled on a dataset comprising 309 C-Arm fluoroscopic images acquired from two distinct C-Arm imaging devices. The imaging device 102 featured 81 fiducials in total, of which 17 fiducials were designated for calibration and 64 fiducials were reserved for distortion analysis. For the purpose of optimization, fiducial weights were computed using only the calibration fiducials. However, to achieve a generalized evaluation of the error metric across the entire image space, the error metric was calculated over both distortion and calibration fiducials. It is noted that non-detection or occlusion of specific fiducials often results in either an increase or a decrease in the error metric. Given that the threshold for medical-grade accuracy is below 2mm, it becomes critical to investigate the variations in resulting from the presence or absence of specific fiducials using the proposed methodology.

[0077] As shown in Fig. 4(a), an error metric analysis conducted over all images revealed that the 4th, 5th, and 7th fiducials were significant contributors to accurate calibration. The study was further narrowed to a subset of 74 images where all calibration fiducials were detected. As illustrated in Fig. 4(b), analysis of this subset demonstrated that the removal of the 4th and 7th fiducials led to a substantial increase in the error metric, thereby establishing them as the most significant fiducials. Conversely, removal of the 6th and 9th fiducials resulted in a reduction of error, classifying them as non- significant.

[0078] Further classification of these images was performed based on the C-Arm imaging device used during acquisition. In 13 images captured using C-Arm 1, as depicted in Fig. 4(c), the 4th and 5th fiducials emerged as significant, while the 8th fiducial was found to be nonsignificant. In contrast, for the 61 images acquired using C-Arm 2, illustrated in Fig. 4(d), the 4th and 7th fiducials were consistently significant, while the 6th and 9th fiducials were deemed non-significant. From this detailed analysis, it was concluded that optimal calibration could be achieved by retaining the 4th and 7th fiducials while excluding the 6th and 9th fiducials. This selection was found to minimize the error metric, thus enhancing calibration accuracy under varying image acquisition conditions.

[0079] Figs. 5a and 5b illustrate tables depicting quantitative analysis of the error metric using weights obtained with fiducial co-ordinate -based optimization and CPM based optimization method respectively, in accordance with an embodiment of the present invention. The input dataset comprises C-Arm fluoroscopic images acquired from two distinct C-Arm manufacturers. For each method, fiducial weights were computed, and the ten fiducials with the highest and lowest weights were identified. The error metric values were subsequently compared across multiple scenarios, including calibrations performed with only the top- and bottom- weighted fiducials, as well as calibrations incorporating significant fiducials while excluding non-significant ones. In each scenario, the configuration yielding the lowest RPE was selected as the optimal outcome. Fig. 5a presents the fiducial weights obtained using the coordinate-based optimization, demonstrating an error metric reduction of 6.96%. Fig. 5b presents results from the CPM-based optimization, where the error metric was reduced by 8.36%. The X parameter for regularization was varied from -200 to +200 in increments of 10 during optimization.

[0080] Fig. 6a and 6b cumulatively presents a comparative analysis of the performance of the proposed methods in minimizing the error metric across images acquired from two different imaging devices, in accordance with an embodiment of the present invention. The evaluation was carried out separately for images captured using C-Arm 1 and C-Arm 2, each containing 17 fiducials. As depicted in Figure 6a, when using C-Arm 1 images, the fiducials selected based on the weights computed using the fiducial coordinate-based optimization method resulted in a lower error metric, indicating more accurate calibration. Conversely, for images acquired using C-Arm 2, as illustrated in Fig. 6b, the CPM-based method yielded superior performance, with weights that effectively guided the selection of optimal fiducials, thereby minimizing the error metric.

[0081] Fig. 7a and 7b cumulatively illustrates the results of a clinical phantom study conducted to validate the efficacy of the proposed methods implemented, in accordance with an embodiment of the present invention. The analysis was performed on both Anteroposterior (AP) and Lateral Projection (LP) images, where fiducials were selected based on the computed optimization weights. The tracking outcomes for a pedicle screw target in the L4 and L5 regions of the spine are visualized using a color-coded tracking tool: green indicates sub-millimetre accuracy (less than 1 mm), while yellow denotes accuracy between 1 mm and 2 mm. These results demonstrate that weighted fiducial selection significantly enhances tracking precisionin IGSS. Furthermore, both optimization approaches employed offer insights into the relative contribution of each fiducial to overall accuracy, enabling the exclusion of fiducials associated with higher error. Although the current findings indicate some device-specific characteristics, the computed weights show promise for generalization across different C-Arm systems, reinforcing the adaptability of the methodology framework.

[0082] Fig. 8 illustrates a flowchart of a method 800 for improving calibration accuracy in the imaging device 102, in accordance with an embodiment of the present invention. In this regard, each block may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative implementations, the functions noted in the blocks may occur out of the order noted in the drawings. For example, two blocks shown in succession in Fig. 8 may in fact be executed substantially concurrently or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. Any process descriptions or blocks in flow charts should be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps in the process, and alternate implementations are included within the scope of the example embodiments in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved. In addition, the process descriptions orblocks in flow charts should be understood as representing decisions made by a hardware structure such as a state machine.

[0083] At step 802, at least one 2D image is acquired by the imaging device, capturing an anatomical region of interest of a patient along with the plurality of fiducial markers. The imaging device which may be a mobile C-arm or similar radiographic unit used in intraoperative settings, is configured to obtain high-resolution radiographic images during surgical procedures. In this step, the imaging device is oriented such that both the anatomical region of interest (e.g., vertebrae of the spine) and the calibration drum are within the field of view. The calibration drum comprises a physical structure embedded with a plurality of spatially distributed fiducials whose geometric configuration is predetermined. These fiducials appear as distinct fiducial markers in the acquired 2D image. The image serves as a foundational input for downstream computational steps that aim to improve the accuracy of the imaging device through calibration.

[0084] At step 804, the image processing unit may detect the 2D positions of the plurality of fiducial markers from the acquired 2D image. The image processing unit is implemented using a combination of hardware and software components, such as a central processing unit (CPU) and image analysis algorithms, that enable automatic fiducial detection. During this step, the image processing unit applies a sequence of image processing techniques such as thresholding, edge detection, morphological filtering, and template matching to locate and extract the 2D image coordinates of each visible fiducial marker. The detection is sensitive to occlusion, visibility, and contrast quality, and the accuracy of this step directly affects the fidelity of subsequent calibration operations.

[0085] At step 806, the image processing unit may calculate a error metric for each of the detected fiducial markers. The error metric serves as a quantitative measure of the reliability or accuracy of each fiducial marker in contributing to the calibration process. The calculation of the error metric depends on the selected optimization method. In one implementation (the first optimization method), the error metric is based on the centroid deviation, which measures how far the detected position of a fiducial marker deviates from its expected ideal location (centroid) on the calibration drum. In an alternative implementation (the second optimization method), the error metric is based on the RPE, which quantifies the discrepancy between the actual detected fiducial position and the reprojected position obtained through a CPM model. The purpose of computing this error metric is to identify fiducials that may have been misidentified, distorted by anatomical interference, or improperly localized due to imaging artifacts.

[0086] At step 808, the image processing unit may assign a weight to each of the plurality of fiducial markers. The assignment of weights is done such that fiducial markers with lower error metric, indicating higher accuracy are assigned higher weights, whereas markers with higher error metrics receive lower weights. The weighting function may follow an inverse relationship with the error metric or may employ a nonlinear scaling function (e.g., exponential decay or logistic regression) to emphasize the contribution of more reliable markers. The goal of this step is to prioritize fiducials markers that are most beneficial for accurate calibration and down-weight those that may introduce calibration errors. This weight value becomes a key determinant in the subsequent selection process.

[0087] At step 810, the image processing unit may select a subset of fiducial markers whose weights exceed a predefined threshold value. The threshold value is empiricallydetermined or adaptively computed to filter out fiducials markers that are likely to degrade calibration accuracy due to high error metrics. The retained subset of fiducials markers thus comprises only those markers that are considered robust, reliable, and minimally affected by visual occlusion, deformation, or poor imaging conditions. This selection step improves calibration robustness by excluding erroneous data points and ensures that only the highest- quality fiducials are used in recalibration.

[0088] At step 812, the image processing unit may perform calibration of one or more parameters of the imaging device using the selected subset of fiducial markers. The calibration process involves updating the imaging device’s intrinsic and / or extrinsic parameters to minimize spatial discrepancies and improve the geometric accuracy of the acquired images. Intrinsic parameters may include the focal length, image center (principal point), and lens distortion coefficients, while extrinsic parameters refer to the position and orientation (pose) of the imaging device relative to the patient and operating environment. The calibration utilizes the known spatial configuration of the selected fiducials on the calibration drum and compares their expected 3D positions with the corresponding 2D image coordinates to estimate camera parameters that minimize the projection error. This recalibration ensures enhanced accuracy in surgical navigation, tool tracking, and image overlay, thereby improving the safety and effectiveness of the image-guided spine surgery procedure.

[0089] The present invention offers several technical advantages in the field of image- guided spine surgeries. The invention enhances calibration accuracy by dynamically adjusting the weights of fiducial markers based on their contribution to minimize the error metric. This approach improves fiducial detection and segmentation, ensuring more reliable and accurate calibration of imaging devices. The invention further optimizes the calibration process by automating fiducial selection and weight adjustment, reducing the need for manual intervention and streamlining the surgical workflow. Additionally, the system’s capability to adapt to variations in fiducial quality and positioning ensures consistent and robust performance across diverse surgical scenarios. These features collectively contribute to enhanced navigation accuracy, improved surgical outcomes, and increased patient safety.

[0090] Although implementations of a method and a system for improving calibration accuracy in image-guided surgeries have been described in language specific to structural features and / or methods, it is to be understood that the appended claims are not necessarily limited to the specific features or methods described. Rather, the specific features and methodsare disclosed as examples of implementations of a method and a system for improving calibration accuracy in image-guided surgeries.

[0091] The invention has been described above with reference to numerous embodiments and specific examples. Many variations will suggest themselves to those skilled in this art in light of the above detailed description. All such obvious variations are within the full intended scope of the appended claims.

Claims

We Claim:

1. A method (800) for improving calibration accuracy in imaging devices, the method (800) comprising: acquiring, by an imaging device (102), at least one two-dimensional (2D) image including an anatomical region of a patient (104) and a plurality of fiducial markers, wherein the plurality of fiducial markers corresponds to visual representation of plurality of fiducials embedded on a calibration drum (108); detecting, by an image processing unit (204), 2D positions of the plurality of fiducial markers from the acquired 2D image; calculating, by the image processing unit (204), a error metric for each of the plurality of fiducial markers; assigning, by the image processing unit (204), a weight to each of the plurality of fiducial markers; selecting, by the image processing unit (204), a subset of fiducial markers having weights above a predefined threshold value, wherein the predefined threshold value corresponds to a weight associated with a lower range of the error metric; and calibrating, by the image processing unit (204), one or more parameters of the imaging device (102) based on the selected subset of fiducial markers to improve accuracy.

2. The method (800) as claimed in claim 1, wherein the calibration drum (108) is positioned within a field of view of the imaging device (102).

3. The method (800) as claimed in claim 1, wherein the calibration drum (108) comprises 81 fiducial markers, including 17 calibration fiducial markers and 64 distortion fiducial markers.

4. The method (800) as claimed in claim 1, wherein calculating the error metric for each of the plurality of fiducial markers comprises:determining, by the image processing unit (204), a centroid of the detected 2D positions of the plurality of fiducial markers present in the at least one acquired 2D image; calculating, by the image processing unit (204), a deviation for each of the fiducial marker from the centroid of the detected 2D positions; and determining, by the image processing unit (204), the error metric by quantifying the deviation from the centroid.

5. The method (800) as claimed in claim 1, wherein the calculating of the error metric for each of the plurality of fiducial markers comprises: projecting, by the image processing unit (204), the known 3D positions of the plurality of fiducial marker into 2D positions on image plane based on a predefined spatial configuration; comparing, by the image processing unit (204), the detected 2D positions of each fiducial marker present in the acquired 2D image to the corresponding projected 2D positions of each fiducial marker; and determining, by the image processing unit (204), the error metric as the difference between the detected 2D positions and the projected 2D positions of each fiducial marker.

6. The method (800) as claimed in claim 5, wherein the projecting of the known 3D positions of each fiducial marker into the 2D positions on the image plane is performed using a camera projection matrix (CPM).

7. The method (800) as claimed in claim 1, wherein one or more parameters comprise one or more of intrinsic parameters including focal length, principal point, and distortion coefficients of the imaging device (102) and extrinsic parameters including camera position and camera orientation of the imaging device (102).

8. A system (100) to improve calibration accuracy of imaging devices, the system (100) comprises: an image processing unit (204); and a memory (206) coupled with the image processing unit (204), wherein the memory (206) stores program instructions configured to:acquire at least a two-dimensional (2D) image including an anatomical region of a patient (104) and a plurality of fiducial markers, wherein the plurality of fiducial markers corresponds to visual representation of plurality of fiducials embedded on a calibration drum (108); detect positions of the plurality of fiducial markers from the acquired 2D image; calculate a error metric for each of the plurality of fiducial markers; assign a weight to each of the plurality of fiducial marker; select a subset of fiducial markers having weights above a predefined threshold value, wherein the predefined threshold value corresponds to a weight associated with a lower range of the error metric; and calibrate the one or more parameters of the imaging device (102) based on the selected subset of fiducial markers.

9. The system (100) as claimed in claim 9, wherein the calibration drum (108) is positioned within a field of view of the imaging device (102).

10. The system (100) as claimed in claim 9, wherein the calibration drum (108) comprises 81 fiducial markers, including 17 calibration fiducial markers and 64 distortion fiducial markers.

11. The system (100) as claimed in claim 9, wherein to calculate the error metric for each of the plurality of fiducial markers, the image processing unit (204) is configured to: determine a centroid of the detected 2D positions of the plurality of fiducial markers in the at least one acquired 2D image; calculate a deviation for each of the fiducial marker from the centroid position of the detected 2D positions; and determine the error metric by quantifying the deviation from the centroid.

12. The system (100) as claimed in claim 9, wherein to calculate the error metric for each of the plurality of fiducial markers, the image processing unit (204) is configured to:project the known 3D positions of the plurality of fiducial marker into 2D positions on image plane based on a predefined spatial configuration; compare the detected 2D positions of each fiducial marker present in the acquired 2D image to the corresponding projected 2D positions of each fiducial marker; and determine the error metric as the difference between the detected 2D positions and the projected 2D positions of each fiducial marker.

13. The system (100) as claimed in claim 13, wherein the projecting of the known 3D positions of each fiducial marker into the 2D positions on the image plane is performed using a camera projection matrix (CPM).

14. The system (100) as claimed in claim 9, wherein one or more parameters comprise one or more of intrinsic parameters, including focal length, principal point, and distortion coefficients of the imaging device (102) and extrinsic parameters, including camera position and orientation of the imaging device (102).

Images