Systems and methods for assisting a surgeon and producing patient-specific medical devices
Systems and methods for assisting a surgeon with implant during a surgery are disclosed. A method includes defining areas of interest in diagnostic data of a patient and defining an implant type. Post defining the areas of interest, salient points are determined for the areas of interest. Successively, an XZ angle, an XY angle, and a position entry point for an implant are determined based on the salient points of the areas of interest. In spinal procedures, a maximum screw diameter and a length of the spinal screw are successively determined based on the salient points. Based on determined length and diameter, a spinal screw and a matching screw guide is determined. Thereafter, the spinal screw and the screw guide is printed using a Three-Dimensional (3D) printer. Such printed spinal screw and screw guide could be used by the surgeon during the spinal surgery.
1 . A non-transitory computer-readable storage medium storing instructions that, when executed by a computing system, cause the computing system to perform operations comprising:
analyzing, using at least one trained machine learning model, diagnostic data of a patient to identify a section of a spine of the patient for surgical correction, the section of the patient's spine including at least a first vertebral level and a second vertebral level;
designing, based on the identified section of the patient's spine and/or the diagnostic data, a first patient-specific implant configured to be implanted at the first vertebral level and a second patient-specific implant configured to be implanted at the second vertebral level, wherein
the first patient-specific implant includes a first patient-specific cage, plate, or disk,
the second patient-specific implant includes a second patient-specific cage, plate, or disk, and
the first patient-specific implant and the second patient-specific implant are configured to provide a target correction to the patient's spine when implanted in the patient;
generating fabrication data for manufacturing the first patient-specific implant and the second patient-specific implant; and
after the first patient-specific implant and the second patient-specific implant are implanted in the patient,
determining post-surgery measurements for a post-operative correction of the patient based on post-surgery anatomical images of the patient,
determining an outcome comparison by comparing measurements of the target correction to the post-surgery measurements,
determining an outcome score associated with the target correction based on the outcome comparison, and
retraining the at least one trained machine learning model based on the outcome score.
2 . The non-transitory computer-readable storage medium of claim 1 , wherein the first patient-specific implant includes a patient-specific cage, and wherein the second patient-specific implant includes a patient-specific plate.
3 . The non-transitory computer-readable storage medium of claim 1 , wherein the first patient-specific implant includes a first patient-specific disk, and wherein the second patient-specific implant includes a second patient-specific disk.
4 . The non-transitory computer-readable storage medium of claim 1 , wherein the first patient-specific implant includes a first patient-specific plate, and wherein the second patient-specific implant includes a second patient-specific plate.
5 . The non-transitory computer-readable storage medium of claim 1 , wherein analyzing the diagnostic data includes:
segmenting the section of the patient's spine; and
analyzing the section of the patient's spine to determine boundaries of the first vertebral level and the second vertebral level.
6 . The non-transitory computer-readable storage medium of claim 5 , wherein analyzing the diagnostic data includes using the at least one trained machine learning model to analyze the diagnostic data.
7 . The non-transitory computer-readable storage medium of claim 1 , wherein designing the first and second patient-specific implants includes using the at least one trained machine learning model to design the first and second patient-specific implants.
8 . The non-transitory computer-readable storage medium of claim 1 , wherein the operations further comprise:
generating a virtual model of the first patient-specific implant and the second patient-specific implant;
receiving user feedback on the first patient-specific implant and/or the second patient-specific implant; and
before generating the fabrication data, adjusting one or more characteristics of the first patient-specific implant and/or the second patient-specific implant based at least in part on the user feedback.
9 . The non-transitory computer-readable storage medium of claim 1 , further comprising manufacturing the first patient-specific implant and the second patient-specific implant.
10 . The non-transitory computer-readable storage medium of claim 1 , wherein the operations further comprise:
inputting a set of post-surgery anatomical images of the patient into the at least one trained machine learning model for retraining.
11 . A system for generating patient-specific medical devices, the system comprising:
one or more processors; and
a memory storing instructions that, when executed by the one or more processors, cause the system to perform operations comprising:
analyzing, using at least one trained machine learning model, diagnostic data of a patient to identify a section of a spine of the patient for surgical correction, the section of the patient's spine including at least a first vertebral level and a second vertebral level;
designing, based on the identified section of the patient's spine and/or the diagnostic data, a first patient-specific implant configured to be implanted at the first vertebral level and a second patient-specific implant configured to be implanted at the second vertebral level, wherein
the first patient-specific implant includes a first patient-specific cage, plate, or disk,
the second patient-specific implant includes a second patient-specific cage, plate, or disk, and
the first patient-specific implant and the second patient-specific implant are configured to provide a target correction to the patient's spine when implanted in the patient;
generating fabrication instructions for manufacturing the first patient-specific implant and the second patient-specific implant; and
after the first patient-specific implant and the second patient-specific implant are implanted in the patient,
determining post-surgery measurements for a post-operative correction of the patient based on post-surgery anatomical images of the patient;
determining an outcome comparison by comparing measurements of the target correction to the post-surgery measurementsa,
determining an outcome score associated with the target correction based on the outcome comparison, and
retraining the at least one trained machine learning model based on the outcome score.
12 . The system of claim 11 , wherein the first patient-specific implant includes a patient-specific cage, and wherein the second patient-specific implant includes a patient-specific plate.
13 . The system of claim 11 , wherein the first patient-specific implant includes a first patient-specific disk, and wherein the second patient-specific implant includes a second patient-specific disk.
14 . The system of claim 11 , wherein the first patient-specific implant includes a first patient-specific plate, and wherein the second patient-specific implant includes a second patient-specific plate.
15 . The system of claim 11 , wherein the operation of analyzing the diagnostic data includes:
segmenting the section of the patient's spine; and
analyzing the section of the patient's spine to determine boundaries of the first vertebral level and the second vertebral level.
16 . The system of claim 15 , wherein the operation of analyzing the diagnostic data includes using the at least one trained machine learning model to analyze the diagnostic data.
17 . The system of claim 11 , wherein the operation of designing the first and second patient-specific implants includes using the at least one trained machine learning model to design the first and second patient-specific implants.
18 . The system of claim 11 , wherein the operations further comprise:
generating a virtual model of the first patient-specific implant and the second patient-specific implant;
receiving user feedback on the first patient-specific implant and/or the second patient-specific implant; and
before generating the fabrication instructions, adjusting one or more characteristics of the first patient-specific implant and/or the second patient-specific implant based at least in part on the user feedback.
19 . The system of claim 11 , wherein the operations further comprise manufacturing the first patient-specific implant and the second patient-specific implant.
20 . A non-transitory computer-readable storage medium storing instructions that, when executed by a computing system, cause the computing system to perform operations comprising:
analyzing, using at least one trained machine learning model, diagnostic data of a patient to identify a section of a spine of the patient for surgical correction, the section of the patient's spine including at least a first vertebra and a second vertebra;
designing, using the at least one trained machine learning model, based on the identified section of the patient's spine and/or the diagnostic data, a first patient-specific implant configured to contact the first vertebra and a second patient-specific implant configured to contact the second vertebra, wherein designing the first patient-specific implant and the second patient-specific implant includes specifying a first configuration of the first patient-specific implant and a second configuration of the second patient-specific implant;
manufacturing the first patient-specific implant and the second patient-specific implant; and
after the first patient-specific implant and the second patient-specific implant are implanted in the patient,
determining post-surgery measurements for a post-operative correction of the patient based on post-surgery anatomical images of the patient,
determining an outcome comparison by comparing measurements of the surgical correction to the post-surgery measurements,
determining an outcome score associated with the surgical correction based on the outcome comparison, and
retraining the at least one trained machine learning model based on the outcome score.
21 . The non-transitory computer-readable storage medium of claim 20 , wherein the operations further comprise:
generating a virtual model of the first patient-specific implant and the second patient-specific implant;
receiving user feedback on the first patient-specific implant and/or the second patient-specific implant; and
before manufacturing the first patient-specific implant and the second patient-specific implant, adjusting one or more characteristics of the first patient-specific implant and/or the second patient-specific implant based at least in part on the user feedback.
22 . A personalized spinal implant system having (i) a first patient-specific cage, plate, or disk configured to be implanted at a first vertebral level and (ii) a second patient-specific cage, plate, or disk configured to be implanted at a second vertebral level, wherein the first patient-specific cage, plate, or disk and the second patient-specific cage, plate, or disk have different configurations so as to cooperate to provide a patient-specific multi-level spinal correction, wherein the personalized spinal implant system is designed by a process comprising:
analyzing, using at least one trained machine learning model, diagnostic data of a patient to identify a section of a spine of the patient for surgical correction, the section of the patient's spine including at least the first vertebral level and the second vertebral level;
designing, based on the identified section of the patient's spine and/or the diagnostic data, the first patient-specific cage, plate, or disk configured to be implanted at the first vertebral level and the second patient-specific cage, plate, or disk configured to be implanted at the second vertebral level, wherein the first patient-specific cage, plate, or disk and the second patient-specific cage, plate, or disk are configured to provide a target correction to the patient's spine when implanted in the patient;
generating fabrication data for manufacturing the first patient-specific cage, plate, or disk and the second patient-specific cage, plate, or disk; and
after the first patient-specific cage, plate, or disk and the second patient-specific cage, plate, or disk are implanted in the patient,
determining post-surgery measurements for a post-operative correction of the patient based on post-surgery anatomical images of the patient,
determining an outcome comparison by comparing measurements of the target correction to the post-surgery measurements,
determining an outcome score associated with the target correction based on the outcome comparison, and
retraining the at least one trained machine learning model based on the outcome score.
23 . The system of claim 22 , wherein the first patient-specific cage, plate, or disk includes a first patient-specific cage, and wherein the second patient-specific cage, plate, or disk includes a second patient-specific cage.
24 . The system of claim 22 , wherein the first patient-specific cage, plate, or disk includes a patient-specific cage, and wherein the second patient-specific cage, plate, or disk includes a patient-specific plate.
25 . The system of claim 22 , wherein the first patient-specific cage, plate, or disk includes a first patient-specific disk, and wherein the second patient-specific cage, plate, or disk includes a second patient-specific disk.
26 . The system of claim 22 , wherein the process further comprises:
generating a virtual model of the first patient-specific cage, plate, or disk and the second patient-specific cage, plate, or disk;
receiving user feedback on the first patient-specific cage, plate, or disk and/or the second patient-specific cage, plate, or disk; and
before generating the fabrication data, adjusting one or more characteristics of the first patient-specific cage, plate, or disk and/or the second patient-specific cage, plate, or disk based at least in part on the user feedback.
27 . A computer-implemented method for designing a personalized spinal implant system, the method comprising:
sending diagnostic data of a patient for analysis by a remote computer using at least one machine learning model, the diagnostic data including image data of at least a first vertebral level and a second vertebral level;
receiving, from the remote computer, first design parameters for a first patient-specific implant configured to be implanted at the first vertebral level and second design parameters for a second patient-specific implant configured to be implanted at a second vertebral level, wherein the first patient-specific implant and the second patient-specific implant are configured to provide a target correction to a spine of the patient when implanted in the patient;
providing one or more user-inputted adjustments to the first design parameters and/or the second design parameters;
receiving, from the remote computer, updated first design parameters for the first patient-specific implant and/or updated second design parameters for the second patient-specific implant based at least in part on the one or more user-inputted adjustments; and
after the first patient-specific implant and the second patient-specific implant are implanted in the patient,
sending post-operative correction data for comparing, by the remote computer, the target correction to a post-operative correction of the patient, wherein the remote computer is programmed to
determine post-surgery measurements for the post-operative correction of the patient based on post-surgery anatomical images of the patient,
determine an outcome comparison by comparing measurements of the target correction to the post-surgery measurements;
determine an outcome score associated with the target correction based on the outcome comparison, and
retrain at least one trained machine learning model based on the outcome score.
28 . The computer-implemented method of claim 27 , further comprising receiving fabrication data for manufacturing the first patient-specific implant and the second patient-specific implant.
29 . The computer-implemented method of claim 28 , further comprising using the fabrication data to manufacture the first patient-specific implant and the second patient-specific implant.
30 . A computer-implemented method for designing a personalized spinal implant system, the method comprising:
sending diagnostic data of a patient for analysis by a remote computer using at least one trained machine learning model, the diagnostic data including image data of at least a first vertebral level and a second vertebral level;
receiving, from the remote computer, first design parameters for a first patient-specific implant configured to be implanted at the first vertebral level and second design parameters for a second patient-specific implant configured to be implanted at the second vertebral level, wherein the first patient-specific implant and the second patient-specific implant are configured to provide a target correction to a spine of the patient when implanted in the patient, wherein the first patient-specific implant is a first patient-specific cage, plate, or disk, and wherein the second patient-specific implant is a second patient-specific cage, plate or disk;
providing one or more user-inputted adjustments to the first design parameters and/or the second design parameters;
receiving, from the remote computer, updated first design parameters for the first patient-specific implant and/or updated second design parameters for the second patient-specific implant based at least in part on the one or more user-inputted adjustments, wherein the first patient-specific implant and/or the second patient-specific implant are designed using the updated first design parameters and/or the updated second design parameters;
after the first patient-specific implant and the second patient-specific implant are implanted in the patient,
sending post-operative correction data for comparing, by the remote computer, the target correction to a post-operative correction of the patient, wherein the remote computer is programmed to determine a position comparison by comparing respective positions of the first patient-specific implant and the second patient-specific implant in post-surgery images to target positions of the first patient-specific implant and the second patient-specific implant associated with the target correction,
determine an outcome score associated with the target correction based on the position comparison, and retrain the at least one trained machine learning model based on the outcome score.
31 . A system comprising: one or more processors; and
a memory storing instructions that, when executed by the one or more processors, cause the system to perform operations comprising:
sending diagnostic data of a patient for analysis by a remote computer system using at least one trained machine learning model, the diagnostic data including image data of at least a first vertebral level and a second vertebral level;
receiving, from the remote computer system, first design parameters for a first patient-specific implant configured to be implanted at the first vertebral level and second design parameters for a second patient-specific implant configured to be implanted at the second vertebral level, wherein the first patient-specific implant and the second patient-specific implant are configured to provide a target correction to a spine of the patient when implanted in the patient;
providing one or more user-inputted adjustments to the first design parameters and/or the second design parameters;
receiving, from the remote computer system, updated first design parameters for the first patient-specific implant and/or updated second design parameters for the second patient-specific implant based at least in part on the one or more user-inputted adjustments, wherein the first patient-specific implant and/or the second patient-specific implant are designed using the updated first design parameters and/or the updated second design parameters; and
after the first patient-specific implant and the second patient-specific implant are implanted in the patient, sending post-operative correction data for comparing, by the remote computer system, the target correction to a post-operative correction of the patient, wherein the remote computer system is programmed to determine post-surgery measurements for the post-operative correction of the patient based on post-surgery anatomical images of the patient,
determine an outcome comparison by comparing measurements of the target correction to the post-surgery measurements, determine an outcome score associated with the target correction based on the outcome comparison, and retrain the at least one trained machine learning model based on the outcome score.
32 . The system of claim 31 wherein the first patient-specific implant is a first patient-specific cage, plate, or disk, and wherein the second patient-specific implant is a second patient-specific cage, plate or disk.
33 . The system of claim 31 , wherein the operations further comprise generating robotic instructions executable to control a robotic surgery system such that the robotic surgery system implants the first patient-specific implant and the second patient-specific implant in the patient.
34 . The system of claim 31 , wherein the operations further comprise:
generating a preoperative surgical plan for controlling a robotic apparatus based on a virtual model of a surgery target area of the patient and a configuration of the first patient-specific implant, wherein the preoperative surgical plan includes at least one implant application parameter, and
controlling the robotic apparatus based on the at least one implant application parameter to position a robotic arm of the robotic apparatus for delivery of the first patient-specific implant.
35 . The system of claim 34 , wherein the at least one implant application parameter includes an insertion point or an insertion angle.
36 . A method of manufacturing patient-specific medical devices, comprising:
receiving, from a remote computer system having at least one trained machine learning model, manufacturing data, the manufacturing data including first design parameters for a first patient-specific implant configured to be implanted at a first vertebral level and second design parameters for a second patient-specific implant configured to be implanted at a second vertebral level
manufacturing the first patient-specific implant and the second patient-specific implant based on the manufacturing data, wherein
the first patient-specific implant includes a first patient-specific cage, plate, or disk,
the second patient-specific implant includes a second patient-specific cage, plate, or disk, and
the first patient-specific implant and the second patient-specific implant are configured to provide a target correction to a spine of a patient when implanted in the patient,
the remote computer system is programmed to
determine post-surgery measurements for a post-operative correction of the patient based on post-surgery anatomical images of the patient;
determine an outcome comparison by comparing measurements of the target correction to the post-surgery measurements,
determine an outcome score associated with the target correction based on the outcome comparison, and
retrain the at least one trained machine learning model based on the outcome score.
37 . The method of claim 36 wherein receiving the manufacturing data includes receiving, via a network, the manufacturing data from a remote computer.
38 . The method of claim 36 wherein receiving the manufacturing data includes receiving, via a network, one or more CAD models of the first patient-specific implant and the second patient-specific implant.
39 . The method of claim 36 wherein receiving the manufacturing data includes receiving, via a network, patient information, the patient information including data associated with the first design parameters and the second design parameters.
40 . The method of claim 36 , further comprising:
before manufacturing the first patient-specific implant and the second patient-specific implant, generating a virtual model of the first patient-specific implant and the second patient-specific implant using the manufacturing data.
41 . The method of claim 36 further comprising generating, based on the manufacturing data, instructions executable by a manufacturing system to control the manufacturing system to fabricate the first patient-specific implant and the second patient-specific implant.
42 . The method of claim 36 wherein the manufacturing data includes instructions executable by a manufacturing system to control the manufacturing system to fabricate the first patient-specific implant and the second patient-specific implant.
43 . A non-transitory computer-readable storage medium storing instructions that, when executed by a computing system, cause the computing system to perform operations comprising:
analyzing, using at least one trained machine learning model, diagnostic data of a patient to identify a section of a spine of the patient for surgical correction, the section of the patient's spine including at least a first vertebral level and a second vertebral level;
designing, based on the identified section of the patient's spine and/or the diagnostic data, a first patient-specific implant configured to be implanted at the first vertebral level and a second patient-specific implant configured to be implanted at the second vertebral level, wherein the first patient-specific implant includes a first patient-specific cage, plate, or disk, the second patient-specific implant includes a second patient-specific cage, plate, or disk, and the first patient-specific implant and the second patient-specific implant are configured to provide a target correction to the patient's spine when implanted in the patient;
generating fabrication data for manufacturing the first patient-specific implant and the second patient-specific implant;
after the first patient-specific implant and the second patient-specific implant are implanted in the patient,
determining a position comparison by comparing respective positions of the first patient-specific implant and the second patient-specific implant in post-surgery images to target positions of the first patient-specific implant and the second patient-specific implant associated with the target correction;
determining an outcome score associated with the target correction based on the position comparison; and
retraining the at least one trained machine learning model based on the outcome score.