Method and apparatus for predicting blood loss in patient undergoing surgery

Abstract

Provided are a method and an apparatus for predicting blood loss in a patient undergoing surgery. The method for predicting blood loss comprises the steps of: detecting a gauze image from a first image of a surgical site of a patient; setting a reference blood color of the patient on the basis of the first image; inputting the reference blood color and the gauze image to a pre-trained blood loss estimation model to estimate the ratio of blood and body fluid absorbed in the gauze; and predicting the amount of lost blood absorbed in the gauze on the basis of the estimated ratio of blood and body fluid.

Description

Method and device for predicting the amount of bleeding in a patient undergoing surgery

[0001] The present invention relates to a method and device for predicting the amount of bleeding of a patient, which can predict the amount of bleeding of a patient in real time from an image of the patient undergoing surgery.

[0002] This study is related to the 2022 Artificial Intelligence Leading Innovation Research Center Support Project for Medical Video AI Innovation Research and Development (Project No. 0670-20220033) supported by the Seoul National University AI Research Institute with funding from the Ministry of Science and ICT (government), the 2023 Basic Medical Joint Research Project for Laparoscopy / Robot Real-time Bleeding Amount Estimation and Development of Artificial Intelligence (AI) System (Project No. 16-2023-0010) supported by Seoul National University Bundang Hospital, and the 2023 Ultra-Generation Research Lab Support Project for Convergence Electronics for Next-Generation Medical Device Development (Project No. 19-2023-0020) supported by Seoul National University Hospital.

[0003] To maintain optimal patient condition during surgery, it is essential to maintain homeostasis by maintaining a constant and stable blood loss throughout the operation. Furthermore, predicting the amount of blood lost during surgery and determining the appropriate amount of intravenous fluids or transfusions is crucial.

[0004] However, in the past, medical staff would visually observe blood or body fluid collected in gauze or blood collection materials used during surgery, such as suction tubes, to predict the amount of bleeding from the patient, which lowered the accuracy of predicting the amount of bleeding.

[0005] Because of this, incorrect predictions of the amount of blood loss for patients undergoing surgery may lead to inaccurate amounts of fluid or blood transfusion being administered to the patient, which may cause hypotension, cardiovascular depression, decreased oxygen transport capacity, decreased tissue perfusion, and edema, which may lead to problems such as postoperative infection or complications.

[0006] The present invention aims to provide a method and device for predicting the amount of bleeding of a patient, which can predict the amount of bleeding of a patient in real time from an image of the patient undergoing surgery.

[0007] A method for predicting the amount of bleeding in a patient undergoing surgery according to an embodiment of the present invention comprises the steps of: acquiring a first image of a surgical site of a patient; detecting a gauze image of one or more gauzes of the surgical site based on the first image; setting a reference blood color of the patient based on the first image; inputting the reference blood color and the gauze image into a pre-learned bleeding amount estimation model to estimate a ratio of blood and body fluid absorbed in the one or more gauzes; and predicting the amount of bleeding absorbed in the one or more gauzes based on the blood and body fluid ratio.

[0008] The step of estimating the blood and body fluid ratio includes the step of dividing the gauze image into a plurality of regions; and the step of estimating the blood and body fluid ratio absorbed in each region from the color of each of the plurality of regions based on the reference blood color.

[0009] The step of estimating the absorption ratio of the blood and body fluid includes the step of determining a plurality of color change sections for the reference blood color based on the amount of saturation change according to the mixing ratio of the blood and the body fluid based on the reference blood color; and the step of estimating the blood and body fluid ratio absorbed in each of the regions by comparing the color of each of the plurality of regions with each of the plurality of color change sections.

[0010] The step of predicting the amount of bleeding absorbed by the one or more gauzes includes the step of estimating the expected amount of absorption by the one or more gauzes based on the gauze information about the one or more gauzes acquired from the gauze image; and the step of predicting the amount of bleeding absorbed by the one or more gauzes from the expected amount of absorption based on the blood and body fluid ratio for each of the plurality of regions.

[0011] Here, the gauze information includes at least one of the number of gauzes, the total area, and whether or not it is folded.

[0012] The step of detecting the gauze image includes the steps of: dividing the first image into a plurality of regions; converting the grayscale levels of the remaining regions adjacent to the first region into binary values ​​based on the grayscale levels of the first region among the plurality of regions; and detecting the gauze image from the first image based on the converted binary values.

[0013] The step of setting the reference blood color further includes the step of updating the reference blood color based on the first image at preset intervals.

[0014] In addition, the method for predicting the amount of bleeding of the present embodiment further includes a step of calculating the amount of blood and body fluid from the amount of bleeding absorbed based on the blood and body fluid ratio.

[0015] In addition, the method for predicting the amount of bleeding of the present embodiment further includes the steps of: acquiring a second image of the surgical site of the patient at the same time as the first image is acquired, and acquiring the amount of bleeding collected during the surgery from the second image; and calculating the final amount of bleeding of the patient by adding the predicted amount of absorbed bleeding and the collected amount of bleeding.

[0016] In addition, the method for predicting the amount of bleeding of the present embodiment further includes the steps of determining a plurality of color change sections for the reference blood color based on the amount of saturation change according to the mixing ratio of the blood and the body fluid based on the reference blood color; the step of estimating the blood and body fluid ratio of the collected bleeding amount by comparing the color of the collected bleeding amount with each of the plurality of color change sections; and the step of calculating the blood amount and body fluid amount in the final bleeding amount based on the blood and body fluid ratio absorbed by the one or more gauzes and the blood and body fluid ratio of the collected bleeding amount.

[0017] A device for predicting the amount of bleeding of a patient undergoing surgery according to an embodiment of the present invention includes: a memory storing a bleeding amount prediction program; and a processor executing the bleeding amount prediction program to obtain a first image of a surgical site of a patient, detect a gauze image of at least one gauze of the surgical site based on the first image, set a reference blood color of the patient based on the first image, input the reference blood color and the gauze image into a pre-learned bleeding amount estimation model to estimate a blood and body fluid ratio absorbed by the at least one gauze, and predict the amount of bleeding absorbed by the at least one gauze based on the blood and body fluid ratio.

[0018] The processor divides the gauze image into a plurality of regions, and estimates the blood and body fluid ratio absorbed in each region from the color of each of the plurality of regions based on the reference blood color.

[0019] The processor determines a plurality of color change sections for the reference blood color based on a saturation change amount according to a mixing ratio of the blood and the body fluid based on the reference blood color, and estimates the ratio of the blood and body fluid absorbed in each of the regions by comparing the color of each of the plurality of regions with each of the plurality of color change sections.

[0020] The processor estimates an expected absorption amount by the one or more gauzes based on gauze information about the one or more gauzes obtained from the gauze image, and predicts the amount of bleeding absorbed by the one or more gauzes from the expected absorption amount based on the blood and body fluid ratio for each of the plurality of regions.

[0021] Here, the gauze information includes at least one of the number of gauzes, the total area, and whether or not it is folded.

[0022] The processor divides the first image into a plurality of regions, converts the grayscale levels of the remaining regions adjacent to the first region into binary values ​​based on the grayscale levels of the first region among the plurality of regions, and detects the gauze image from the first image based on the converted binary values.

[0023] The processor updates the reference blood color based on the first image at preset intervals.

[0024] The processor calculates the blood volume and body fluid volume from the absorbed bleeding amount based on the blood and body fluid ratio.

[0025] The processor acquires a second image of the surgical site of the patient at the same time as the first image is acquired, acquires the amount of bleeding collected during the surgery from the second image, and calculates the final amount of bleeding of the patient by adding the predicted amount of absorbed bleeding and the collected amount of bleeding.

[0026] The processor determines a plurality of color change sections for the reference blood color based on a saturation change amount according to a mixing ratio of the blood and the body fluid based on the reference blood color, estimates a blood and body fluid ratio of the collected bleeding amount by comparing the color of the collected bleeding amount with each of the plurality of color change sections, and calculates a blood volume and a body fluid volume in the final bleeding amount based on the blood and body fluid ratio absorbed by the one or more gauzes and the blood and body fluid ratio of the collected bleeding amount.

[0027] The present invention uses a learned neural network model to predict the amount of bleeding of a patient absorbed by gauze from a first image of the patient's surgical site, and combines this with the collected amount of bleeding obtained from a second image of the patient's surgical site to accurately calculate the final amount of bleeding discharged by the patient during surgery.

[0028] In addition, the present invention can increase the accuracy of predicting the amount of bleeding absorbed into the gauze by setting a reference blood color for the patient's blood from bleeding occurring in real time at the patient's surgical site and using this to estimate the ratio of blood and body fluid absorbed into the gauze.

[0029] In addition, the present invention can accurately estimate the ratio of blood and body fluid absorbed into the gauze and the ratio of blood and body fluid collected based on the established reference blood color, thereby calculating the amount of blood and body fluid from the final amount of bleeding of the patient.

[0030] Accordingly, the present invention enables medical staff performing surgery on a patient to quickly and accurately determine the patient's bleeding condition, thereby increasing the safety of the surgery and preventing secondary accidents caused by the surgery.

[0031] FIG. 1 is a drawing showing a bleeding amount prediction device according to an embodiment of the present invention.

[0032] Figure 2 is a diagram conceptually illustrating the function of the bleeding amount prediction program of Figure 1.

[0033] Figure 3 is a diagram showing a method for learning the bleeding amount estimation model of Figure 2.

[0034] Figure 4 is a drawing showing a method for predicting bleeding amount according to an embodiment of the present invention.

[0035] Figures 5 to 7 are drawings specifically showing a method for predicting the amount of bleeding absorbed by the gauze of Figure 4.

[0036] Figures 8 and 9 are examples of a method for predicting the amount of bleeding of the present invention.

[0037] A method and device for predicting the amount of bleeding of a patient undergoing surgery are provided. The method for predicting the amount of bleeding includes the steps of detecting a gauze image from a first image of a surgical site of a patient, setting a reference blood color of the patient based on the first image, inputting the reference blood color and the gauze image into a pre-learned bleeding amount estimation model to estimate the ratio of blood and body fluid absorbed into the gauze, and predicting the amount of bleeding absorbed into the gauze based on the estimated blood and body fluid ratio.

[0038] The advantages and features of the present invention, and the methods for achieving them, will become clearer with reference to the embodiments described in detail below together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below and may be implemented in various different forms. These embodiments are provided solely to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the invention, and the present invention is defined solely by the scope of the claims.

[0039] When describing embodiments of the present invention, detailed descriptions of known functions or configurations will be omitted if they are deemed to unnecessarily obscure the gist of the invention. Furthermore, the terms described below are defined in light of their functions in the embodiments of the present invention and may vary depending on the intent or custom of the user or operator. Therefore, their definitions should be based on the overall content of this specification.

[0040] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings.

[0041] FIG. 1 is a drawing showing a bleeding amount prediction device according to an embodiment of the present invention.

[0042] Referring to FIG. 1, the bleeding amount prediction device (100) of the present embodiment receives images of a patient undergoing surgery from a plurality of cameras, for example, a first camera (200) and a second camera (300), and can predict the bleeding amount of the patient following surgery based on the received images.

[0043] Here, the amount of blood loss may refer to blood and body fluids discharged from a patient during surgery, for example, the amount of blood loss may refer to the sum of blood and body fluids discharged from a patient during surgery.

[0044] Typically, bodily fluids include blood. However, for convenience of explanation, in this embodiment, bodily fluids are defined as any liquid other than blood discharged from a patient undergoing surgery. These bodily fluids may include fluids within the human body, such as tissue fluid and lymph.

[0045] Accordingly, the bleeding amount prediction device (100) of the present embodiment can predict the amount of bleeding discharged from a patient undergoing surgery based on images received from the first camera (200) and the second camera (300), and calculate the amount of blood and body fluid from the predicted amount of bleeding.

[0046] This bleeding amount prediction device (100) may include an input / output unit (110), a processor (120), and a memory (130).

[0047] The input / output unit (110) can receive images of a patient undergoing surgery from each of the first camera (200) and the second camera (300).

[0048] Here, the first camera (200) can capture the surgical site of the patient and output a first image. The second camera (300) can capture the surgical site of the patient, for example, the interior of the operating room, and output a second image.

[0049] In addition, the input / output unit (110) can receive a bleeding amount prediction result for a patient undergoing surgery from a processor (120) to be described later and output it to an external device, for example, one or more monitors (not shown) that display the surgical situation.

[0050] The processor (120) receives the first image and the second image from the input / output unit (110), and can predict the amount of bleeding of a patient undergoing surgery from the first image and the second image using the bleeding amount prediction program (140) stored in the memory (130).

[0051] Additionally, the processor (120) can calculate the blood volume and body fluid volume, respectively, based on the ratio of blood and body fluid within the predicted bleeding volume.

[0052] The memory (130) can store a bleeding amount prediction program (140) and information necessary for its execution. The bleeding amount prediction program (140) can be software including commands that can predict the bleeding amount of a patient from the first and second images received by the input / output unit (110), i.e., the surgical site image and the surgical site image of the patient undergoing surgery, and calculate the blood volume and body fluid volume from the predicted bleeding amount.

[0053] Accordingly, the processor (120) executes a bleeding amount prediction program (140) stored in the memory (130), and uses this to predict the bleeding amount of a patient undergoing surgery from the first and second images received in the input / output unit (110), and calculates the amount of blood and body fluid based on the ratio of blood and body fluid within the bleeding amount.

[0054] Figure 2 is a diagram conceptually illustrating the function of the bleeding amount prediction program of Figure 1.

[0055] Referring to FIG. 2, the bleeding amount prediction program (140) of the present embodiment may include a reference color setting unit (141), a gauze image detection unit (143), a bleeding amount estimation model (145), and a final bleeding amount calculation unit (147).

[0056] The reference color setting unit (141), gauze image detection unit (143), bleeding amount estimation model (145), and final bleeding amount calculation unit (147) illustrated in FIG. 2 are divided to easily explain the function of the bleeding amount prediction program (140) of this embodiment, and the present invention is not limited thereto.

[0057] For example, according to an embodiment of the present invention, the reference color setting unit (141), the gauze image detection unit (143), the bleeding amount estimation model (145), and the final bleeding amount calculation unit (147) may have their functions merged or separated, and may be implemented as a series of commands included in a single program.

[0058] The reference color setting unit (141) can set a reference color, for example, a reference blood color, for the blood of a patient undergoing surgery based on the first image provided from the input / output unit (110). The reference color setting unit (141) can output the set reference blood color to a bleeding amount estimation model (145) to be described later.

[0059] Here, one or more gauze may be inserted into the surgical site of the patient by a doctor or the like to absorb bleeding resulting from the surgery. Accordingly, the first image provided from the input / output unit (110) may include an image of one or more gauzes, for example, gauze that has absorbed bleeding occurring at the surgical site.

[0060] However, the gauze in the first image may have absorbed not only the patient's blood due to bleeding at the surgical site, but also the patient's body fluids discharged from the surgical site. Therefore, the reference color setting unit (141) of the present embodiment can set the patient's reference blood color from the color of blood resulting from real-time bleeding at the surgical site where no gauze is present in the first image.

[0061] Additionally, the reference color setting unit (141) can repeat setting the reference blood color for the patient in the first image at preset intervals.

[0062] The gauze image detection unit (143) can detect a gauze image for one or more gauzes present in the first image based on the first image.

[0063] As described above, the first image may include one or more gauzes for absorbing bleeding. Accordingly, the gauze image detection unit (143) may detect gauze images of one or more gauzes included in the first image.

[0064] For example, the gauze image detection unit (143) can divide the first image, i.e., the first image including one or more gauzes, into a plurality of regions. Here, each of the plurality of regions can have the same area.

[0065] Next, the gauze image detection unit (143) can extract the grayscale level of each region and compare the grayscale level of a first region among a plurality of regions with the grayscale levels of a plurality of other regions adjacent to the first region in at least four directions.

[0066] Here, the gauze image detection unit (143) can extract the average of the grayscale levels of each of the plurality of pixels in each area as the grayscale level of the corresponding area.

[0067] In addition, the gauze image detection unit (143) can convert the grayscale levels of each of a plurality of other areas adjacent to the first area into binary values, for example, binary codes, based on the comparison results of the grayscale levels described above.

[0068] For example, if the grayscale level of one of the multiple different regions is higher than the grayscale level of the first region, that is, if one region is brighter than the first region, the gauze image detection unit (143) can convert the grayscale level of one region into a first binary value, for example, 1.

[0069] Additionally, if the grayscale level of one area is lower than the grayscale level of the first area, i.e., if one area is darker than the first area, the gauze image detection unit (143) can convert the grayscale level of one area into a second binary value, for example, 0.

[0070] Accordingly, the gauze image detection unit (143) can detect a gauze image by extracting an area having a first binary value from among multiple areas of the first image and cropping a gauze portion from the extracted area.

[0071] For example, gauze used in a patient's surgery may be mostly white. Therefore, among the multiple regions of the first image, regions containing gauze may have a higher grayscale level than regions not containing gauze, such as regions containing organs in the surgical site.

[0072] Accordingly, the gauze image detection unit (143) can accurately detect the gauze image in the first image by extracting a relatively brighter area among the multiple areas of the first image, that is, an area having the first binary value, as an area where gauze exists.

[0073] The bleeding amount estimation model (145) can estimate the ratio of blood and body fluid absorbed into gauze among the bleeding amount discharged from a patient undergoing surgery based on a preset reference blood color and a previously detected gauze image.

[0074] In addition, the bleeding amount estimation model (145) can predict the bleeding amount absorbed into the gauze from the blood and body fluid ratio estimated based on the gauze information obtained from the gauze image.

[0075] This bleeding amount estimation model (145) may include one or more pre-learned neural network models.

[0076] Figure 3 is a diagram showing a method for learning the bleeding amount estimation model of Figure 2.

[0077] Referring to FIG. 3, the bleeding amount estimation model (145) of the present embodiment may include one or more neural network models, for example, a first estimation unit (151), a second estimation unit (153), and a prediction unit (155).

[0078] The first estimation unit (151) can be trained to estimate the ratio of blood and body fluid among the blood absorbed by the gauze from the gauze image when a gauze image, for example, an image of gauze that has absorbed bleeding and a reference blood color are input.

[0079] For example, the first estimation unit (151) can set the reference blood color to the highest level of saturation, obtain the saturation change amount according to the mixing ratio of blood and body fluid based on this, and determine multiple color change sections for the reference blood color based on the saturation change amount.

[0080] Here, the saturation change can be obtained by changing the mixing ratio of blood and body fluid according to the preset displacement amount.

[0081] For example, since the reference blood color is set to the highest level of saturation, the reference blood color may have a blood-to-body fluid ratio of 100:0. The first estimation unit (151) can obtain the saturation change amount for the reference blood color by gradually changing the blood-to-body fluid ratio from 100:0 to 0:100 according to a preset displacement amount, for example, a displacement amount of 10. Accordingly, the first estimation unit (151) can determine 10 color change sections for the reference blood color based on the saturation change amount.

[0082] Here, the 10 color change intervals may include a highest saturation color with a blood to body fluid ratio of 100:0, for example, a reference blood color, a lowest saturation color with a blood to body fluid ratio of 0:100, and colors having 8 saturation values ​​between the highest saturation color and the lowest saturation color.

[0083] Additionally, the first estimation unit (151) can divide the gauze image into multiple regions. Here, the multiple regions can have the same area.

[0084] The first estimation unit (151) can compare the color of each of a plurality of areas of the gauze image with each of a plurality of predetermined color change sections to match the color change section corresponding to each area.

[0085] Accordingly, the first estimation unit (151) estimates the ratio of blood and body fluid absorbed in each of the multiple regions of the gauze image, and from this, the blood and body fluid ratio for the entire gauze image can be estimated.

[0086] Meanwhile, the first estimation unit (151) can input correct data, such as the actual ratio of blood and body fluid, as label data, and compare the output estimation result with the label data to determine an estimation loss, such as a first estimation loss value.

[0087] Accordingly, the first estimation unit (151) can further input the determined first estimated loss value and repeatedly perform the aforementioned learning, i.e., learning to estimate the blood and body fluid ratio absorbed from the gauze image, so that the first estimated loss value is minimized.

[0088] The second estimation unit (153) can be trained to estimate the expected absorption amount of the gauze from the gauze image when the gauze image is input.

[0089] In general, to absorb bleeding that occurs during surgery, multiple gauzes are placed in the surgical site, or one or more gauzes are folded at least once and placed in the surgical site.

[0090] Accordingly, the second estimation unit (153) can obtain gauze information from the gauze image and determine the status of one or more gauzes in the gauze image.

[0091] Here, the gauze information may include the number of gauzes included in the gauze image, the total area of ​​the gauze, and whether one or more gauzes are folded.

[0092] For example, if the gauze image includes multiple gauzes or multiple gauzes are overlapped, the expected absorption amount may be greater compared to when the gauze image includes one gauze.

[0093] Additionally, if the total area of ​​one gauze included in the gauze image is smaller than a preset gauze area, for example, the area when the gauze is fully unfolded, the gauze is folded, so the absorption time by the gauze increases and the expected absorption amount may decrease.

[0094] Accordingly, the second estimation unit (153) can determine the gauze condition in the gauze image from the gauze information and estimate the expected absorption amount of the gauze accordingly.

[0095] In addition, the second estimation unit (153) can input correct data, for example, the actual absorption amount of the gauze, as label data, and compare the output estimation result with the label data to determine an estimation loss, for example, a second estimation loss value.

[0096] Accordingly, the second estimation unit (153) can further input the determined second estimated loss value and repeatedly perform learning to estimate the expected absorption amount of the gauze from the aforementioned gauze image so that the second estimated loss value is minimized.

[0097] The prediction unit (155) can be trained to predict the amount of bleeding absorbed by one or more gauzes included in the gauze image when the blood and body fluid ratio of the gauze image output from the first estimation unit (151) described above and the expected absorption amount of the gauze output from the second estimation unit (153) are input.

[0098] In addition, the prediction unit (155) can input correct data, such as the actual absorption bleeding amount of the gauze, as label data, and compare the output estimation result with the label data to determine the prediction loss value.

[0099] Accordingly, the prediction unit (155) can repeatedly perform learning to receive additional inputs of the determined predicted loss values ​​and predict the amount of bleeding absorbed by one or more gauzes described above so that the predicted loss values ​​are minimized.

[0100] Referring again to FIG. 2, the final bleeding amount calculation unit (147) can calculate the final bleeding amount discharged from the patient undergoing surgery based on the absorbed bleeding amount of one or more gauzes in the gauze image output from the bleeding amount estimation model (145) and the collected bleeding amount obtained from the second image provided through the input / output unit (110).

[0101] Here, the second image may be received by the input / output unit (110) at the same time as the first image. This second image may include an image of a blood collection unit placed within the surgical site, for example, a collection container (not shown) in which blood discharged from a patient through suction is collected.

[0102] The final bleeding amount calculation unit (147) can obtain the bleeding amount collected from the patient from the collection tube image included in the second image. At this time, the final bleeding amount calculation unit (147) can obtain the bleeding amount collected from the second image while predicting the bleeding amount absorbed by the gauze at the surgical site of the patient from the aforementioned bleeding amount estimation model (145).

[0103] Accordingly, the final bleeding amount calculation unit (147) can calculate the final bleeding amount of a patient undergoing surgery by adding the bleeding amount absorbed by the gauze and the bleeding amount collected in the collection container.

[0104] In addition, the final bleeding amount calculation unit (147) can calculate the ratio of blood and body fluid from the collected bleeding amount acquired from the second image based on the preset reference blood color of the patient. The calculation of the blood and body fluid ratio by the final bleeding amount calculation unit (147) can be substantially the same as estimating the ratio of blood and body fluid absorbed into the gauze based on the saturation change amount of the reference blood color in the first estimation unit (151) of the aforementioned bleeding amount estimation model (145).

[0105] The final bleeding amount calculation unit (147) can calculate the blood amount and body fluid amount from the patient's final bleeding amount based on the blood and body fluid ratio absorbed by the gauze estimated in the bleeding amount estimation model (145) and the blood and body fluid ratio calculated from the collected bleeding amount.

[0106] Accordingly, the processor (120) can receive the patient's final bleeding amount, the blood volume and the body fluid volume for the final bleeding amount from the final bleeding amount calculation unit (147), and output the same to one or more monitors that display the surgical situation through the input / output unit (110).

[0107] In this way, the bleeding amount prediction device (100) of the present embodiment can accurately calculate the final bleeding amount discharged by the patient during surgery by predicting the bleeding amount of the patient absorbed into the gauze from the first image of the patient's surgical site using the learned neural network model and combining this with the collected bleeding amount obtained from the second image of the patient's surgical site.

[0108] The bleeding amount prediction device (100) of this embodiment sets a reference blood color for the patient's blood from bleeding occurring in real time at the patient's surgical site, and uses this to estimate the ratio of blood and body fluid absorbed into the gauze, thereby increasing the accuracy of prediction of the bleeding amount absorbed into the gauze.

[0109] In addition, the bleeding amount prediction device (100) of the present embodiment can accurately estimate the ratio of blood and body fluid absorbed into the gauze and the ratio of blood and body fluid collected based on the set reference blood color, thereby calculating the blood amount and body fluid amount from the patient's final bleeding amount.

[0110] Accordingly, the present invention enables medical staff performing surgery on a patient to quickly and accurately determine the patient's bleeding condition, thereby increasing the stability of the surgery and preventing secondary accidents caused by the surgery.

[0111] FIG. 4 is a drawing showing a method for predicting the amount of bleeding according to an embodiment of the present invention, FIGS. 5 to 7 are drawings specifically showing a method for predicting the amount of bleeding absorbed by the gauze of FIG. 4, and FIGS. 8 and 9 are exemplary drawings of the method for predicting the amount of bleeding of the present invention.

[0112] Referring to FIG. 4, the bleeding amount prediction device (100) of the present embodiment can receive a first image of the surgical site of a patient undergoing surgery and a second image of the surgical site from the first camera (200) (S10).

[0113] Accordingly, the processor (120) of the bleeding amount prediction device (100) executes the bleeding amount prediction program (140) stored in the memory (130), and uses this to predict the bleeding amount of a patient undergoing surgery, calculate the final bleeding amount, and calculate the blood amount and body fluid amount from the calculated final bleeding amount.

[0114] For example, as illustrated in FIG. 8, the reference color setting unit (141) can detect bleeding occurring in real time in a surgical site where no gauze exists in the first image, and set a reference blood color for the patient's blood according to the brightness, hue, or saturation of the color corresponding to the detected bleeding using a preset saturation table or color table (S20).

[0115] Next, the bleeding amount estimation model (145) estimates the ratio of blood and body fluid absorbed into one or more gauzes at the surgical site of the patient from the preset reference blood color and the detected gauze image, and based on this, can predict the bleeding amount absorbed into one or more gauzes (S30).

[0116] As described above, the bleeding amount estimation model (145) may include one or more pre-trained neural network models, such as a first estimation unit (151), a second estimation unit (153), and a prediction unit (155). Accordingly, the bleeding amount estimation model (145) may predict the amount of bleeding absorbed by one or more gauzes at the surgical site of a patient undergoing surgery through the first estimation unit (151), the second estimation unit (153), and the prediction unit (155).

[0117] Referring to FIGS. 5 and 6, the gauze image detection unit (143) can detect a gauze image including one or more gauzes from an area where the gauze exists in the first image (S110).

[0118] For example, as illustrated in FIG. 9, the gauze image detection unit (143) can divide the first image into multiple regions (S111).

[0119] Next, the gauze image detection unit (143) can extract the grayscale level of each of the plurality of regions and compare the grayscale level of the first region among the plurality of regions with the grayscale level of the region adjacent to the first region (S112).

[0120] Here, the grayscale level of each of the multiple regions can be extracted as the average of the grayscale levels of each of the multiple pixels of each region.

[0121] Additionally, the region adjacent to the first region among the plurality of regions may be another region adjacent to the first region in at least four directions among the up / down / left / right / diagonal directions.

[0122] As a result of the comparison of the aforementioned grayscale levels, if the grayscale level of the first region is smaller than the grayscale level of the adjacent region (Y), the gauze image detection unit (143) can convert the grayscale level of the adjacent region into a preset first binary value, for example, a value of 1 (S113).

[0123] On the other hand, if the grayscale level of the first region is greater than the grayscale level of the adjacent region (N), the gauze image detection unit (143) can convert the grayscale level of the adjacent region into a preset second binary value, for example, a value of 0 (S114).

[0124] Accordingly, the gauze image detection unit (143) can extract an area having a first binary value among multiple areas of the first image and detect a gauze image by cropping the gauze portion from the extracted area (S115).

[0125] Referring to FIGS. 5 and 7, the first estimation unit (151) of the bleeding amount estimation model (145) can estimate the ratio of blood and body fluid absorbed into one or more gauzes from a preset reference blood color and a detected gauze image (S120).

[0126] For example, the first estimation unit (151) can set the reference blood color to the highest level of saturation, obtain the saturation change amount according to the mixing ratio of blood and body fluid based on this, and determine multiple color change sections for the reference blood color based on the saturation change amount (S121).

[0127] Next, the first estimation unit (151) divides the gauze image into multiple regions, and compares the color of each divided region with each of multiple predetermined color change regions to match the color change region corresponding to each region (S123).

[0128] Accordingly, the first estimation unit (151) estimates the ratio of blood and body fluid absorbed in each of the multiple areas of the gauze image, and from this, can estimate the ratio of blood and body fluid absorbed in one or more gauzes (S125).

[0129] Referring again to FIG. 5, the second estimation unit (153) of the bleeding amount estimation model (145) can estimate the expected absorption amount of the gauze from the gauze image (S130).

[0130] For example, the second estimation unit (153) can obtain gauze information for one or more gauzes from the gauze image and determine the condition of the gauze. Then, the second estimation unit (153) can estimate the expected absorption amount for one or more gauzes in the gauze image based on the determined gauze condition.

[0131] Next, the prediction unit (155) of the bleeding amount estimation model (145) can predict and output the bleeding amount absorbed by one or more gauzes from the estimated blood and body fluid ratio absorbed by the gauze and the expected absorption amount (S140).

[0132] Referring again to FIG. 4, the final bleeding amount calculation unit (147) can obtain the collected bleeding amount from the image of the collection tube that collects the bleeding discharged by the patient in the second image provided through the input / output unit (110) (S40).

[0133] As previously described, the input / output unit (110) receives the first and second images at the same time. Accordingly, the final bleeding amount calculation unit (147) can obtain the bleeding amount collected from the patient from the second image while the aforementioned bleeding amount estimation model (145) predicts the bleeding amount absorbed by the gauze at the patient's surgical site from the first image.

[0134] Next, the final bleeding amount calculation unit (147) can calculate the final bleeding amount discharged from the patient undergoing surgery based on the bleeding amount absorbed by the gauze predicted by the bleeding amount estimation model (145) and the collected bleeding amount acquired from the second image (S50).

[0135] Next, the final bleeding amount calculation unit (147) calculates the ratio of blood and body fluid from the collected bleeding amount obtained from the second image based on the preset reference blood color of the patient, and based on the ratio of blood and body fluid absorbed into the gauze estimated by the bleeding amount estimation model (145) and the ratio of blood and body fluid calculated from the collected bleeding amount, the blood amount and body fluid amount can be calculated from the final bleeding amount of the patient, respectively (S60).

[0136] Accordingly, the processor (120) can receive the patient's final bleeding amount, the blood volume and the body fluid volume for the final bleeding amount from the final bleeding amount calculation unit (147), and output the same to one or more monitors that display the surgical situation through the input / output unit (110).

[0137] The combinations of each block of the block diagram and each step of the flowchart of the present invention described above may be implemented by a computer program composed of a plurality of instructions. Such a computer program may be installed in an encoding processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program executed by the encoding processor of the computer or other programmable data processing device generates means for performing the functions described in each block of the block diagram or each step of the flowchart. In addition, the computer program may be stored in a recording medium such as a computer-available or computer-readable memory that can be directed to a computer or other programmable data processing device to implement functions in a specific manner, so that the program stored in the computer-available or computer-readable memory can produce an article of manufacture that includes means for performing the functions described in each block of the block diagram or each step of the flowchart. The computer program may also be installed in a computer or other programmable data processing device, so that a series of operation steps are performed on the computer or other programmable data processing device to create a computer-executable process, so that the program that executes the computer or other programmable data processing device can provide steps for performing the functions described in each block of the block diagram and each step of the flowchart.

[0138] Additionally, each block or step may represent a module, segment, or portion of code that includes one or more executable instructions for performing a specific logical function(s). It should also be noted that in some alternative embodiments, the functions described in the blocks or steps may occur out of order. For example, two blocks or steps depicted in succession may actually be performed substantially concurrently, or the blocks or steps may sometimes be performed in reverse order, depending on the functionality involved.

[0139] The above description is merely an illustrative illustration of the technical idea of ​​the present invention, and those skilled in the art will appreciate that various modifications and variations can be made without departing from the essential quality of the present invention. Therefore, the embodiments disclosed in the present invention are intended to illustrate, rather than limit, the technical idea of ​​the present invention, and the scope of the technical idea of ​​the present invention is not limited by these embodiments. The scope of protection of the present invention should be interpreted by the following claims, and all technical ideas within a scope equivalent thereto should be interpreted as being included in the scope of the rights of the present invention.

[0140] By monitoring the amount of bleeding in a patient undergoing surgery in real time, the present invention can be applied to real-time surgical environments requiring accurate blood loss assessment. Therefore, the present invention is expected to be applicable to UI / UX that integrates data curation, clinical support functions, and robotic surgery environments, leveraging the results of deep learning-based blood loss estimation.

Claims

1. A step of acquiring a first image of the patient's surgical site; A step of detecting a gauze image for one or more gauzes of the surgical site based on the first image; A step of setting a reference blood color of a patient based on the first image; A step of estimating the blood and body fluid ratio absorbed in one or more gauzes by inputting the reference blood color and the gauze image into the learned bleeding amount estimation model; and A method for predicting the amount of blood loss of a patient undergoing surgery, comprising the step of predicting the amount of blood loss absorbed by the one or more gauzes based on the blood and body fluid ratio.

2. In paragraph 1, The step of estimating the blood and body fluid ratio is as follows: a step of dividing the above gauze image into a plurality of regions; and A method for predicting the amount of bleeding of a patient undergoing surgery, comprising the step of estimating the ratio of blood and body fluid absorbed in each region from the color of each of the plurality of regions based on the reference blood color.

3. In paragraph 2, The step of estimating the absorption rate of the blood and body fluids is as follows: A step of determining a plurality of color change sections for the reference blood color based on the amount of saturation change according to the mixing ratio of the blood and the body fluid based on the reference blood color; and A method for predicting the amount of bleeding of a patient undergoing surgery, comprising the step of comparing the color of each of the plurality of regions with each of the plurality of color change sections to estimate the ratio of blood and body fluid absorbed in each region.

4. In paragraph 2, The step of predicting the amount of bleeding absorbed by one or more of the above gauzes is: A step of estimating an expected absorption amount by said one or more gauzes based on gauze information about said one or more gauzes obtained from said gauze image; and A method for predicting the amount of bleeding of a patient undergoing surgery, comprising the step of predicting the amount of bleeding absorbed by the one or more gauzes from the expected absorption amount based on the blood and body fluid ratios for each of the plurality of regions.

5. In paragraph 4, The above gauze information is, A method for predicting the amount of blood loss in a patient undergoing surgery, comprising at least one of the number of gauze, total area, and fold status.

6. In paragraph 1, The step of detecting the above gauze image is: A step of dividing the first image into multiple regions; A step of converting the grayscale level of the remaining areas adjacent to the first area into binary values based on the grayscale level of the first area among the plurality of areas; and A method for estimating the amount of bleeding of a patient undergoing surgery, comprising the step of detecting the gauze image from the first image based on the converted binary value.

7. In paragraph 1, The step of setting the above standard blood color is: A method for estimating the amount of bleeding of a patient undergoing surgery, further comprising the step of updating the reference blood color based on the first image at preset intervals.

8. In paragraph 1, A method for predicting the amount of blood loss of a patient undergoing surgery, further comprising a step of calculating the amount of blood and body fluid from the amount of blood absorbed based on the blood and body fluid ratio.

9. In paragraph 1, A step of acquiring a second image of the patient's surgical site at the same time as the first image is acquired, and acquiring the amount of bleeding collected during the surgery from the second image; and A method for predicting the amount of blood loss of a patient undergoing surgery, further comprising the step of calculating the final amount of blood loss of the patient by adding the predicted amount of absorbed blood loss and the collected amount of blood loss.

10. In paragraph 9, A step of determining a plurality of color change sections for the reference blood color based on the amount of change in saturation according to the mixing ratio of the blood and the body fluid based on the reference blood color; A step of estimating the blood and body fluid ratio of the collected bleeding amount by comparing the color of the collected bleeding amount with each of the plurality of color change sections; and A method for predicting the amount of blood loss of a patient undergoing surgery, further comprising the step of calculating the amount of blood and the amount of body fluid in the final amount of blood loss based on the ratio of blood and body fluid absorbed in the one or more gauzes and the ratio of blood and body fluid in the collected amount of blood loss.

11. Memory where the bleeding amount prediction program is stored; and A device for predicting the amount of bleeding of a patient undergoing surgery, comprising a processor for executing the above-mentioned bleeding amount prediction program to obtain a first image of the surgical site of the patient, detecting a gauze image for one or more gauzes of the surgical site based on the first image, setting a reference blood color of the patient based on the first image, inputting the reference blood color and the gauze image into a previously learned bleeding amount estimation model to estimate a ratio of blood and body fluid absorbed in the one or more gauzes, and predicting the amount of bleeding absorbed in the one or more gauzes based on the blood and body fluid ratio.

12. In paragraph 11, The above processor, A device for predicting the amount of blood loss of a patient undergoing surgery, which divides the gauze image into a plurality of regions and estimates the blood and body fluid ratio absorbed in each region from the color of each of the plurality of regions based on the reference blood color.

13. In paragraph 12, The above processor, A device for predicting the amount of bleeding of a patient undergoing surgery, the device determining a plurality of color change sections for the reference blood color based on the amount of change in saturation according to the mixing ratio of the blood and the body fluid based on the reference blood color, and comparing the color of each of the plurality of sections with each of the plurality of color change sections to estimate the ratio of blood and body fluid absorbed in each of the sections.

14. In paragraph 12, The above processor, Estimating an expected absorption amount by said one or more gauzes based on gauze information obtained from said gauze image, and predicting the amount of bleeding absorbed by said one or more gauzes from the expected absorption amount based on the blood and body fluid ratio for each of said plurality of regions. The above gauze information is a device for predicting the amount of blood loss of a patient undergoing surgery, including at least one of the number of gauzes, total area, and whether or not they are folded.

15. In paragraph 11, The above processor, A device for predicting the amount of bleeding of a patient undergoing surgery, which divides the first image into a plurality of regions, converts the grayscale levels of the remaining regions adjacent to the first region into binary values based on the grayscale levels of the first region among the plurality of regions, and detects the gauze image from the first image based on the converted binary values.

16. In paragraph 11, The above processor, A device for predicting the amount of bleeding of a patient undergoing surgery, which updates the reference blood color based on the first image at preset intervals.

17. In paragraph 11, The above processor, A device for predicting the amount of blood loss of a patient undergoing surgery, which calculates the amount of blood and body fluid from the amount of blood loss absorbed based on the blood and body fluid ratio.

18. In paragraph 11, The above processor, A device for predicting the amount of bleeding of a patient undergoing surgery, which acquires a second image of the surgical site of the patient at the same time as the first image is acquired, acquires the amount of bleeding collected during the surgery from the second image, and calculates the final amount of bleeding of the patient by adding the predicted amount of absorbed bleeding and the collected amount of bleeding.

19. In Article 18, The above processor, A device for predicting the amount of blood loss of a patient undergoing surgery, the device determining a plurality of color change sections for the reference blood color based on the amount of change in saturation according to the mixing ratio of the blood and the body fluid based on the reference blood color, comparing the color of the collected amount of bleeding with each of the plurality of color change sections to estimate the blood and body fluid ratios of the collected amount of bleeding, and calculating the amount of blood and body fluid in the final amount of bleeding based on the blood and body fluid ratios absorbed by the one or more gauzes and the blood and body fluid ratios of the collected amount of bleeding.

20. A computer-readable recording medium storing a computer program, The above computer program, A step of acquiring a first image of the patient's surgical site; A step of detecting a gauze image for one or more gauzes of the surgical site based on the first image; A step of setting a reference blood color of a patient based on the first image; A step of estimating the blood and body fluid ratio absorbed in one or more gauzes by inputting the reference blood color and the gauze image into the learned bleeding amount estimation model; and A computer-readable recording medium comprising instructions for causing a processor to perform a method for predicting the amount of bleeding of a patient undergoing surgery, the method comprising the step of predicting the amount of bleeding absorbed by the one or more gauzes based on the blood and body fluid ratio.

Images