Computer-implemented method for determining planning data for surgical procedure of subject

Abstract

A computer-implemented method for determining planning data for a surgical procedure of a subject in a urinary procedure includes providing, by a processor, a predictive model, the prediction model is trained to predict planning data based on at least one of historical patient data, historical task data, historical planning data, current patient data and current task data (S100); obtaining, by the processor, at least one of current patient data and current task data (S200); inputting the at least one of the current patient data and the current task data into the predictive model in order to determine planning data (S300); a user recommendation is provided by the processor based on the determined planning data with respect to a plan for a urinary flow plan, such as a PCNL flow (S400).

Description

Technical Field

[0001] The present invention relates to a computer-implemented method for determining planning data for a surgical procedure on an object, an apparatus and a computer program for determining planning data for a surgical procedure on an object. Background Art

[0002] Medical procedures such as surgical procedures (e.g., percutaneous nephrolithotomy (PCNL)) are difficult to plan due to the multiple factors that influence them. However, the quality and efficiency of such procedures depend on reliable planning data.

[0003] It is now apparent that there is also a need to provide a method for determining planning data.

[0004] These and other objects are solved by the subject matter of the independent claims, which become apparent after reading the following description. The invention provides a method, a device and a computer program. The dependent claims relate to preferred embodiments of the invention. Summary of the Invention

[0005] In view of the above, it is an object of the present invention to provide a method allowing an improved determination of planning data for a surgical procedure on a subject, in particular a human being.

[0006] The inventors have recognized that some medical procedures, such as urological procedures (e.g., percutaneous nephrolithotomy (PCNL), a minimally invasive procedure that destroys and removes kidney stones from the kidney through a small puncture wound through the skin), have high complication rates. Therefore, planning a PCNL procedure can be challenging in terms of duration, required equipment, level of expertise, and personnel (including those from other departments, including those who should be on call).

[0007] These and other objects are solved by the subject matter of the independent claims, which become apparent after reading the following description. The invention provides a method, a device and a computer program. The dependent claims relate to preferred embodiments of the invention.

[0008] Planning surgical procedures is challenging given clinical complexity (e.g., the various patient factors to consider) and clinical workflow complexity. For example, in PCNL, patient obesity, the proximity of the calyceal selected for initial puncture to at-risk organs and arteries, stone size and the associated risk of ureteral obstruction by stone fragments, and stone composition all factor into the duration of the procedure and the choice of cutting device. This makes each case unique relative to the location of the calyceal and at least one stone. In some cases, the entire procedure must be reorganized due to insufficient planning data or new factors that were not accounted for.

[0009] In one aspect of the present disclosure, a computer-implemented method for determining planning data for a surgical procedure for a subject in a urology procedure comprises:

[0010] A processor provides a prediction model, wherein the prediction model is trained to predict planning data based on at least one of historical patient data, historical task data, historical planning data, current patient data, and current task data (S100);

[0011] The processor obtains at least one of current patient data and current task data (S200);

[0012] Inputting the at least one of the current patient data and the current task data into the prediction model to determine planning data (S300);

[0013] A user recommendation regarding planning of a urology procedure plan, such as a PCNL procedure, is provided by the processor based on the determined planning data ( S400 ).

[0014] The term "subject" in the present context should be understood broadly and may include any person and any animal.

[0015] The term "planning data" as used herein should be understood broadly and may relate to any data necessary for or useful in planning a medical procedure (urological surgery). Planning data may include at least one of staffing of personnel, instrument availability, room availability, room preparation, patient data, planning of the area of the PCNL needle into the subject, and / or prior approach plans.

[0016] The term "surgical procedure" as used herein should be understood broadly and may relate to any surgical procedure performed on a human or animal. The surgical procedure may be, for example, a urological procedure (eg, a PCNL procedure).

[0017] As used herein, the term "predictive model" should be understood broadly and may relate to any predictive model based on a machine learning algorithm or an artificial intelligence algorithm that is configured to be trained to predict planning data based on historical patient data, historical task data, historical planning data, current patient data, and current task data. The predictive model may be trained using training data to predict planning data.

[0018] The machine learning model according to the present invention may preferably include at least one of the following: decision tree, naive Bayes classification, nearest neighbor, neural network, convolutional neural network (CNN), generative adversarial network (GAN), multi-conditional GAN, support vector machine, linear regression, logistic regression, random forest and / or gradient boosting algorithm.

[0019] As used herein, the term "historical patient data" should be understood in a broad sense and may relate to any patient data that can be obtained from a historical surgical procedure. It should be understood that a historical surgical procedure refers to the surgical procedures of other patients and the associated patient data, which can be used to compare and plan the surgical procedures of the current patient. Therefore, historical patient data can be received from real surgical procedures performed in the past. A lot of patient-specific data can be extracted from imaging data (e.g., computed tomography (CT) images, ultrasound (US) images, fused US / MRI (magnetic resonance imaging) images) and patient metadata (e.g., age, body circumference for obesity, body mass index (BMI), etc.), which can come from connected enterprise imaging systems, radiology information systems, hospital information systems, electronic medical record systems, electronic medical systems, and any combination of these systems. In some embodiments, this information can come from the same hospital. In some embodiments, this information can come from different hospitals. A user can request the required historical patient data from at least one system (e.g., electronic medical record (EHR) system) via a suitable user interface.

[0020] As used herein, the term "historical task data" should be understood broadly and may refer to any task data that can be obtained from historical surgical procedures. This means that historical task data can be received from actual surgical procedures performed in the past. Historical task data may include data related to the clinical procedures / manipulations performed on the patient, such as the target to be treated, clinical procedure codes, etc.

[0021] As used herein, the term "historical planning data" should be understood broadly and may relate to any planning data that can be obtained from historical surgical procedures. Historical planning data may include initial planning data and implemented planning data. This means that historical planning data may be received from actual surgical procedures performed in the past. It should be noted that historical planning data is particularly derived from historical task data. Therefore, historical task data may be at least part of the historical planning data.

[0022] As used herein, the term "current patient data" should be understood broadly and may relate to any patient data that may be used for a future surgical procedure on a subject to be treated. Current patient data may include age, BMI, etc. This data may include historical data for the same patient. The difference between "historical patient data" and "current patient data" is that historical patient data may include data from other patients, whereas current patient data may also include historical information from the current patient.

[0023] The term "current task data" as used herein should be understood broadly and may relate to any task data that may be used for a future surgical procedure on an object to be treated. The current task data may include the target to be treated, the treatment modality, and the location of the target in the object.

[0024] The term "current" as used herein refers to the present or future as opposed to history which refers to the past.

[0025] As used herein, the term "providing" should be understood broadly and may refer to any manner of presenting data. "Providing" may include presenting planning data and / or user recommendations on a screen, on a graphical user interface (GUI) on a dashboard, in a data file, in a table, etc. In some embodiments, the user can add additional entries (e.g., a patient's daily weight) via a GUI, for example.

[0026] In other words, the present invention may involve using a trained model to predict planning data for a current surgical procedure, wherein the model is trained using data about past surgical procedures on other patients and procedures related to the current patient. In some embodiments, the trained model can search for similar patients based on similarity of parameters (e.g., comparing CT images of the patients). When the model interpolates between the data points, the combination of these data points can optimally predict the characteristics of the currently planned surgical procedure. For example, a prediction can be provided in the sense of "a urologist faced with a similar task used the following instruments, staff, and took 125 minutes to complete this surgical procedure." The prediction can be accompanied by charts, images, graphs, tables, etc. This can be advantageous in terms of efficiency, quality, reduced complexity, and reduced errors.

[0027] It will be clear to those skilled in the art that the method is not performed during a surgical procedure and does not interact with the object to be treated.

[0028] Percutaneous nephrolithotomy (PCNL) is a minimally invasive procedure that destroys and removes kidney stones from the kidney through a small puncture wound (up to about 1 cm) through the skin. As an alternative to shock wave lithotripsy or ureteroscopy, PCNL is suitable for larger stones or stones with complex shapes (staghorn calculi) and is typically performed in an operating room by a urologist or a combined team of urologists and radiologists under general or spinal anesthesia.

[0029] PCNL is typically planned on image data available from a diagnostic contrast CT scan and performed with the patient in a prone position. Through a skin incision and under imaging guidance by X-ray fluoroscopy or ultrasound, a hollow nephrolithotomy needle is advanced to the calyx considered suitable for reaching the stone on the CT image, and a guide wire is inserted. Several expanded sheaths are then inserted over the guide wire until the opening is large enough to allow the nephroscope to pass through and remove the stone. Larger stones may need to be cut first using laser or cutter technology.

[0030] Placement of the guidewire for stone access is considered a more delicate part of the procedure and is often performed by the interventional radiologist, sometimes even in a separate pre-procedure, rather than by the urologist, because the approach path is close to sensitive structures, e.g., vasculature, pleura, spleen, colon, liver, which may be accidentally punctured while calyceal access is being established due to inaccurate needle placement and insertion.

[0031] Although PCNL is generally well tolerated, complications are relatively frequent and include ureterolithiasis (stone fragments sliding into the ureter), failure to completely remove the stone, perforation of the urine collecting system, vascular injury, and colon perforation or pleural injury. Staghorn kidney stones are challenging conditions that require careful preoperative evaluation and close follow-up to avoid stone recurrence.

[0032] Due to its relatively high complication rate, planning a PCNL procedure can be challenging in terms of duration, required instruments, expertise, and personnel (including those who should be on call). Factors that need to be considered are patient- or stone-related factors, including obesity, proximity of the calyceal selected for initial puncture to organs and arteries at risk, size of the stone(s), the associated risk of ureteral obstruction by stone fragments, and stone composition, all of which factor into the duration of the procedure and the choice of cutting equipment. This makes each case unique with respect to the calyceal anatomy and the location of the stone(s).

[0033] For example, obese patients present several technical challenges, including anesthesia, patient positioning, imaging for access, longer skin-to-collecting system distance, and nephrostomy tube migration.

[0034] The present invention proposes the use of AI-based algorithms (e.g., machine learning algorithms) to provide optimal predictions of the planned procedure characteristics of interest, such as expected duration, potential complications, staff requirements, and instrumentation requirements. The present invention can provide a user-oriented system that allows input of all details that cannot be captured automatically. The present invention can present the predictions to a physician (e.g., a urologist) through a user interface.

[0035] The present invention can provide a database that records information including, for example, stone characteristics (e.g., number, size(s), location(s), composition(s)), procedure characteristics (e.g., planned and achieved approach paths, path criticality (distance to organs and structures at risk), duration of use, complications, instruments used, staff presence (including experience level), procedure success (no stones)), patient characteristics (e.g., age, comorbidities, BMI).

[0036] According to an embodiment, the user recommendation may include a recommendation regarding at least one of workflow adjustment, risk stratification, possible side effects, and procedure recommendation.

[0037] According to an embodiment, the procedure recommendation may be derived from at least one of a planned and realized approach path, distance to the organ, structures at risk, recommended procedure duration, recommended instruments, staff availability.

[0038] According to an embodiment, the method may further include providing, by the processor, a database comprising historical patient data, historical task data, and historical planning data. This may be advantageous because it may improve the reliability of the prediction model. As used herein, the term "database" should be broadly understood to refer to any digital database configured to store data in an organized manner.

[0039] According to an embodiment, the method may further comprise storing the current patient data, the current task data, the determined planning data, and the implemented planning data in the database. This may be advantageous because it may result in a higher quality prediction of the trained model. The prediction model may be trained continuously. The database may receive the current patient data, the current task data, the determined planning data, and the implemented task data automatically or manually via a data interface. This interface may be in the form of a manually received HMI (i.e., a user interface) configured to receive input from, for example, a staff member performing the surgical procedure.

[0040] According to an embodiment, a method is provided, further comprising: receiving historical patient data, historical task data, and / or historical planning data via a user interface. This may be advantageous because some specific data, such as staff observations, cannot be automatically received. The user interface may be a tablet, a PC, an HMI, or the like.

[0041] According to an embodiment, a method is provided, comprising selecting a collection of historical patient data, historical task data, and historical planning data using the predictive model and the obtained current patient data and current task data, and providing the selected collection. This can be advantageous because the collection details how the person who will be performing the surgical procedure has performed similar situations in the past. Thus, the data selected from the database can be used as a supplementary aid to the determined planning data.

[0042] According to an embodiment, the patient data may include at least one of age, comorbidities, and BMI, and / or the historical planning data may include at least determined planning data and implemented planning data. The term "implemented planning data" may relate to planning data implemented in order to perform a surgical procedure, for example, an implemented approach route to a kidney stone. The implemented approach route to the kidney stone may deviate from a previously determined approach route. This may also be advantageous for practitioners because it helps them to classify the determined planning data with respect to reliability in some cases.

[0043] According to an embodiment, the patient data may include a CT image of the subject, and / or at least part of the patient data may be extracted from the CT image. This may be advantageous because the CT image and the corresponding information may provide more detailed information.

[0044] According to an embodiment, the task data may include at least information about a target to be treated located in the subject and / or information about a standard approach path to the target located in the subject. A standard approach path may include, for example, a specific calyx where a kidney stone is located. This information may be presented in a modified CT image that depicts the approach path and location of the kidney stone. This may reduce the complexity of the upcoming surgical procedure.

[0045] According to an embodiment, the planning data may include at least one of the instruments required for the surgical procedure, the duration of the surgical procedure, the staff required for the surgical procedure, the risk of adverse events, and an approach path to a target to be treated in the subject. This may be advantageous in terms of planning quality.

[0046] According to an embodiment, the planning data may include contact information for the attending physician. This may be useful for receiving further information if questions remain. Contact information may include an address, phone number, email address, etc. Contact information for cases similar to the current case may be retrieved from a database, where the case is described or defined by the planning data, patient data, and task data.

[0047] According to an embodiment, the method may include determining an uncertainty measure for the determined planning data. The term "uncertainty measure" as used herein should be understood broadly and may relate to the uncertainty of the determined planning data. The uncertainty measure may also be determined by using a prediction model. In other words, the uncertainty measure is a further result of the prediction model, which reveals the quality of the prediction result. The uncertainty measure may be presented in the same way as the presentation of the planning data as described above. The uncertainty measure may be on a scale between 0 and 1 or 0% to 100%, etc. The uncertainty measure may also be used to determine rare cases. This may be useful information for staff entrusted with specific cases.

[0048] According to an embodiment, a case report of the surgical procedure can be generated and provided for further processing based on the determined uncertainty measure. For example, when the uncertainty measure is below a threshold (e.g., 80%), a case report is generated as indicating a rare case for the current surgical procedure. The case report can be used for further analysis of planning data, patient data, and task data. This may help to better understand and be aware of difficult situations. The case report can include any available data for a particular surgical procedure. Since the prediction model can be trained to also provide uncertainty measures for these predictions, not only can these predictions be shown, but these predictions can also be used to identify rare cases (i.e., cases whose characteristics can only be predicted with great uncertainty). This situation can be indicated to the user with a recommendation that a case report will be published. These case reports can also be used to train young professionals.

[0049] According to an embodiment, the trained prediction model is trained continuously. In other words, the current patient data, the determined planning data, and the current task data can be used to train the prediction model. This may be advantageous in terms of prediction quality and reliability.

[0050] Another aspect relates to an apparatus for determining planning data for a surgical procedure on a subject, the apparatus comprising a module for performing the steps of the above-described method. The module may include a processor, a computer unit, and a workstation having corresponding interfaces for providing the predictive model and obtaining patient data, task data, and determining planning data. The module may include a screen, a dashboard, a tablet, etc. for providing and presenting the determined planning data.

[0051] A final aspect relates to a computer program comprising instructions which, when executed by a computer, cause the computer to perform the above method; and / or a computer-readable medium comprising instructions which, when executed by a computer, cause the computer to perform the above method.

[0052] The computer program may be stored on a computer unit, which may also be part of an embodiment. The computer unit may be configured to execute or cause the execution of the steps of the above-described method. Furthermore, the computer unit may be configured to operate components of the above-described device. The computing unit can be configured to operate automatically and / or execute user commands. The computer program can be loaded into the working memory of a data processor. Thus, the data processor can be equipped to perform the method according to one of the aforementioned embodiments. This exemplary embodiment of the present invention covers both computer programs that use the present invention from the outset and computer programs that convert existing programs into programs that use the present invention through an update. Furthermore, the computer program may be able to provide all necessary steps to implement the flow of the exemplary embodiment of the method described above. According to another exemplary embodiment of the present invention, a computer-readable medium, such as a CD-ROM, USB stick, etc., is provided, wherein the computer-readable medium has the computer program described in the previous sections stored thereon. The computer program can be stored and / or distributed on a suitable medium (e.g., an optical storage medium or solid-state medium provided with or as part of other hardware), but may also be distributed in other forms (e.g., via the Internet or other wired or wireless telecommunications systems). However, the computer program may also be presented over a network like the World Wide Web and can be downloaded from such a network into the working memory of a data processor. According to a further exemplary embodiment of the present invention, a medium for making a computer program available for downloading is provided, said computer program being arranged to perform a method according to one of the aforementioned embodiments of the invention.

[0053] It will be appreciated that user recommendations regarding planning of a urology procedure (e.g., a PCNL procedure) can be displayed directly to the user of the imaging device (e.g., a urology station, an ultrasound device) or transmitted to a DICOM station. In some embodiments, the recommendations are incorporated into the EHR system. In some embodiments, the displayed recommendations also include a recommendation regarding the need for consultation with an additional specialist (e.g., an interventional radiologist).

[0054] Note that the above embodiments can be combined with each other, regardless of the aspects involved. Thus, the method can be combined with structural features of devices and / or systems of other aspects, and similarly, these devices and systems can be combined with features of each other and can also be combined with features described above with respect to the method.

[0055] These and other aspects of the invention will be apparent from and elucidated with reference to the embodiments described hereinafter. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Exemplary embodiments of the present invention will be described in the following drawings.

[0057] Figure 1 is a schematic diagram of a method for determining planning data for a surgical procedure for a subject in a urology procedure;

[0058] Figure 2 A schematic diagram of an apparatus for determining planning data for a surgical procedure for a subject in a urology procedure is shown; and

[0059] Figure 3 Shown are exemplary images selected with the aid of a method for determining planning data for a surgical procedure for a subject in a urological procedure.

[0060] List of reference numerals:

[0061] S100 provides trained prediction models

[0062] S200 obtains current patient data and current task data

[0063] S300 Confirm planning data

[0064] S400 provides user recommendations

[0065] 10 Equipment

[0066] 100, 101, 102, 103 Diagnostic CT images

[0067] 104, 105 approach path

[0068] 106, 107 Different Views DETAILED DESCRIPTION

[0069] Figure 1 is a schematic diagram of a method for determining planning data for a surgical procedure.

[0070] Step 100 includes providing, by a processor, a prediction model that is trained to predict planning data based on historical patient data, historical task data, historical planning data, current patient data, and current task data. In this example, the prediction model may be a machine learning model. The surgical procedure may be percutaneous nephrolithotomy (PCNL), a procedure for removing stones from the kidney via skin puncture. In this example, the subject may be a human.

[0071] The prediction model is a model trained using historical patient data, historical task data, and historical planning data. The historical planning data may include at least determined planning data and realized planning data.

[0072] The planning data may include one of: instruments required for the surgical procedure, duration of the surgical procedure, staff required for the surgical procedure, risk of adverse events, and an approach path to a target to be treated located in the subject.

[0073] The (historical and current) task data may include at least information about a target to be treated located in the subject and / or information about a standard approach path to the target located in the subject. In this example, the task data may include the location, size, and / or form of one or more stones in the kidney. The standard approach path may include one or more calyces suitable for reaching the stone. The standard approach path may be depicted on the CT image.

[0074] The patient data may include at least one of the following: age, comorbidities, and body mass index (BMI). The patient data may include a CT image of the subject and / or at least a portion of the patient data extracted from the CT image. The (historical and current) planning data may include contact information for the attending physician. The processor may be part of a computer unit, a workstation, or a virtual machine. The processor may be a single entity or distributed across multiple entities.

[0075] Step 200 includes obtaining, by the processor, current patient data and current task data. The current patient data and current task data may be obtained automatically via an interface between the processor and a database system (e.g., a hospital database system). The current patient data and current task data may be obtained manually from user input via an HMI, wherein a staff member (e.g., a physician) enters the current patient data and current task data.

[0076] In this example, the current patient data may include: the patient has a BMI of 26.6 and is 36 years old. The current task data may include: the presence of a stone having an approximately circular shape in a position relative to a reference point that must be removed. The current patient data may include: the corresponding diagnostic CT image.

[0077] Step 300 includes inputting at least one of current patient data and current task data into a prediction model in order to determine planning data.The prediction model receives the current patient data and the current task data as input and provides planning data as output.

[0078] The determined planning data includes interpolations between data points / sets used to train the predictive model. The determined planning data may then, for example, include: specific tools are required, an approach path through a specific calyx is proposed, a urologist is required, and the procedure will last 85 minutes with no complications expected.

[0079] Step 400 includes providing, by a processor, user recommendations regarding planning of a urology procedure (e.g., a PCNL procedure) based on the determined planning data. The user recommendations may be depicted in a graph, report, table, adjusted CT image, etc. The user recommendations may be presented on a screen or dashboard.

[0080] Additionally, the user recommendation may include a recommendation regarding at least one of: workflow adjustments, risk stratification, possible side effects, and procedure recommendations.

[0081] Additionally, the procedure recommendation may be derived from at least one of: planned and achieved approach paths, distance to organs, structures at risk, recommended procedure duration, recommended instruments, staff availability.

[0082] The method may further include receiving historical patient data, historical task data, and / or historical planning data through a user interface. The method may further include selecting a set of historical patient data, historical task data, and historical planning data by utilizing the prediction model and the obtained current patient data and current task data, and providing the selected set.

[0083] The method may further include determining an uncertainty measure for the determined planning data. Based on the determined uncertainty measure, a case report of the surgical procedure may be generated and provided for further processing. The prediction model may be continuously trained using any further received datasets for the currently performed surgical procedure.

[0084] Figure 2 A schematic diagram of a device for determining planning data for a surgical procedure on a subject in a urology procedure is shown.

[0085] Device 100 includes a processor. The processor may be part of a computer unit. The computer unit may be configured to execute or cause the steps of the method to be executed. In this example, the processor is part of a workstation located in a hospital. Alternatively, the processor may be located in a cloud application and accessed by an interface and corresponding communication unit. The device may also include a presentation module (e.g., a screen or dashboard). The device may also include one or more interfaces for data exchange.

[0086] Figure 3 Shown are exemplary images selected with the aid of a method for determining planning data for a surgical procedure for a subject in a urological procedure.

[0087] Figure 3 Four diagnostic CT images 100, 101, 102, and 103 are included. Image 100 shows a diagnostic CT image of an object to be treated in two views 106, 107, each of which includes an approach path 104 and 105. Diagnostic CT images 101, 102, and 103 are diagnostic CT images from historical cases stored in a database.

[0088] The diagnostic CT images 101, 102 and 103 each also show two views and a corresponding approach path.The diagnostic CT images 101, 102 and 103 are selected from a plurality of cases stored in a database by the above method.

[0089] The selected diagnostic CT images 101, 102, and 103 show the greatest similarity with the current task data and the current patient data. In particular, diagnostic CT image 103 shows the best match of the three images 101, 102, and 103. This set of CT images is presented to the user (e.g., a doctor) who must perform the surgical procedure during the planning phase.

Claims

1. A computer-implemented method for determining planning data for a surgical procedure for a subject in a urology procedure, comprising: A processor provides a prediction model, wherein the prediction model is trained to predict planning data based on at least one of historical patient data, historical task data, historical planning data, current patient data, and current task data (S100); The processor obtains at least one of current patient data and current task data (S200); Inputting the at least one of the current patient data and the current task data into the prediction model to determine planning data (S300); A user recommendation regarding planning of a urology procedure plan, such as a PCNL procedure, is provided by the processor based on the determined planning data ( S400 ).

2. The method according to claim 1, wherein The user recommendation may include a recommendation regarding at least one of workflow adjustment, risk stratification, possible side effects, and procedure recommendation.

3. The method according to claim 2, wherein: The procedure recommendation is derived from at least one of a planned and realized approach path, distance to the organ, structures at risk, recommended procedure duration, recommended instruments, and staff availability.

4. The method according to any one of the preceding claims, further comprising: Historical patient data, historical procedure data, and / or historical planning data are received via a user interface.

5. The method according to claim 2, further comprising: A set of historical patient data, historical task data, and historical planning data is selected by utilizing the prediction model and the obtained current patient data and current task data, and the selected set is provided.

6. The method according to any one of the preceding claims, wherein The patient data includes at least one of age, comorbidities, and BMI, and / or, wherein the historical planning data includes at least determined planning data and realized planning data.

7. The method according to any one of the preceding claims, wherein The patient data comprises a CT image (100) of the subject, and / or wherein at least part of the patient data is extracted from the CT image (100).

8. A method according to any one of the preceding claims, wherein The task data comprises at least information about a target to be handled located in the object and / or information about a standard approach path (104, 105) to the target located in the object.

9. The method according to any one of the preceding claims, wherein The planning data includes at least one of instruments required for the surgical procedure, duration of the surgical procedure, staff required for the surgical procedure, risk of adverse events, and an approach path to a target to be treated located in the subject.

10. The method according to any one of the preceding claims, wherein The planning data further includes contact information of an attending physician, and wherein the planning data includes the contact information of the attending physician.

11. The method according to any one of the preceding claims, further comprising: An uncertainty measure is determined for the determined planning data.

12. The method according to claim 11, wherein A case report of the surgical procedure is selected and provided for further processing based on the determined uncertainty measure.

13. The method according to any one of the preceding claims, wherein The trained prediction model is trained continuously.

14. A device (10) for determining planning data for a surgical procedure on a subject, comprising means for performing the steps of the method according to claim 1.

15. A computer program comprising instructions which, when executed by a computer, cause the computer to perform the method according to any one of claims 1 to 13; and / or a computer-readable medium comprising instructions which, when executed by a computer, cause the computer to perform the method according to any one of claims 1 to 13.

Images