Method for detecting and displaying femoral fracture

Abstract

The present invention relates to a method for detecting and displaying a femoral fracture and, in particular, to a method for detecting and displaying a femoral fracture, which is capable of quickly and accurately detecting and displaying a femoral fracture. The method for detecting and displaying a femoral fracture of the present invention comprises: an image-capturing step for acquiring an original image by capturing a high resolution image of the body including the lower body of a patient, the image showing bones; a detection step for detecting a femur part from the original image; an extraction step for extracting a plurality of small images from the original image of the detected femur part; an examination step for examining, by means of a trained deep learning model, whether a femoral fracture exists in the small images; a coordinate detection step for detecting coordinate values of a part having the femoral fracture; a coordinate merging step for merging the detected coordinate values; and a display step for displaying the position corresponding to the merged coordinate values on the original image. The resolutions of the original image and the small images are the same, and the coordinate values are location values of the part having the femoral fracture with respect to a preset reference point on the original image.

Description

Method for detecting and labeling femoral fractures

[0001] The present invention relates to a method for detecting and displaying a femoral fracture, and more particularly, to a method for detecting and displaying a femoral fracture that can quickly and accurately detect and display a femoral fracture.

[0002] A fracture is a complete or incomplete loss of continuity of a bone, end plate, or joint surface, and is usually caused by an external force.

[0003] At this time, fractures are often accompanied by damage to soft tissues or organs surrounding the bone.

[0004] Depending on where the fracture occurs, fractures can be broadly divided into limb fractures, spinal fractures, and other fractures such as rib, skull, and orbital fractures.

[0005] Meanwhile, the diagnosis of a fracture can be made through X-rays.

[0006] However, diagnosis of fractures based on X-ray images may require additional special tests such as computed tomography or magnetic resonance imaging because the results are ambiguous or it is difficult to accurately confirm the fracture pattern.

[0007] Furthermore, in the case of fracture diagnosis based on medical images, since medical staff directly determine the presence or absence of a fracture by visually examining it, there may be differences in opinion depending on the skills or experience of the medical staff, which may limit the ability to provide reliable diagnosis results.

[0008] Meanwhile, failure to diagnose fractures can lead to worsening of patient outcomes and increased medical costs, so there is a continuous need for the development of new fracture detection systems.

[0009] Atypical femoral fractures are those that occur without trauma or with minor trauma to the femur. Early detection is important, but they are difficult to detect without a purposeful and careful examination by a specialist.

[0010] The present invention is intended to solve the above-mentioned problems, and its purpose is to provide a method for detecting and displaying a femoral fracture that can automatically, quickly and accurately detect a femoral fracture occurring in the femur and display it externally.

[0011] In order to achieve the above object, the present invention provides a method for detecting and displaying a femur fracture, comprising: a shooting step of capturing a body including the lower body of a patient as a high-resolution image in which bones are displayed to obtain an original image; a detection step of detecting a femur portion from the original image; an extraction step of extracting the original image of the detected femur portion into a plurality of small-sized mini-images; an inspection step of checking whether a femur fracture exists in the mini-image using a pre-trained deep learning model; a coordinate detection step of detecting a coordinate value of a site where a femur fracture exists; a coordinate merging step of merging the detected coordinate values; and a display step of displaying a location corresponding to the merged coordinate value on the original image; wherein the resolutions of the original image and the mini-image are the same, and the coordinate value is a location value of a site where a femur fracture exists based on a preset reference point of the original image.

[0012] In the above extraction step, a plurality of small images are extracted so that some parts overlap with each other.

[0013] In the above detection step, if the original image includes even a portion of the femur, the femur area is detected.

[0014] According to the method for detecting and displaying femoral fractures of the present invention as described above, the following effects are achieved.

[0015] It can automatically, quickly, and accurately detect femoral fractures occurring in the femur and display them externally, allowing for early detection and prompt treatment.

[0016] In particular, the present invention detects a femoral fracture site through a plurality of small images including the femur rather than detecting a femoral fracture in the entire high-resolution original image, so that the femoral fracture site can be detected more quickly and early.

[0017] Figure 1 is a flowchart of a femoral fracture detection and display method according to an embodiment of the present invention;

[0018] Figure 2 is a drawing for explaining the extraction step in the femoral fracture detection and display method according to an embodiment of the present invention.

[0019] FIG. 3 is a drawing for explaining a display step in a femoral fracture detection and display method according to an embodiment of the present invention.

[0020] The method for detecting and displaying a femoral fracture of the present invention comprises, as illustrated in FIG. 1, a photographing step (S1), a detection step (S2), an extraction step (S3), an inspection step (S4), a coordinate detection step (S5), a coordinate merging step (S6), and a display step (S7).

[0021] Each step of the present invention is performed using a computer, medical equipment, etc.

[0022] The above shooting step (S1) is a step of obtaining an original image by shooting a high-resolution image of the patient's body, including the lower body, in which bones are displayed.

[0023] At this time, in the above shooting step (S1), shooting is performed using imaging equipment such as X-ray, CT, or MRI.

[0024] The above detection step (S2) is a step of automatically detecting the femur area in the original image using a computer, etc.

[0025] In the above detection step (S2), the shape of the femur can be displayed more accurately by adding color to the femur in the original image.

[0026] In the above detection step (S2), if the original image includes even a portion of the femur, the femur area is detected.

[0027] Conventional deep learning-based methods have examined only femur X-ray images.

[0028] However, the present invention is directed to all medical images that include even a portion of the femur.

[0029] That is, the present invention includes at least a portion of the femur, such as a femur X-ray, a hip joint X-ray, a pelvis X-ray, a knee X-ray, a leg length X-ray, etc., and performs the detection step (S2).

[0030] Therefore, in the process of examining other areas, it is also possible to determine whether there is a femoral fracture in the femur, as described later.

[0031] The above extraction step (S3) is a step of extracting the original image of the detected femur area into a number of small-sized miniature images.

[0032] At this time, the resolution of the original image and the small image are the same, and the only difference is their size.

[0033] Figure 2 illustrates the process of extracting a small image from an original image through an extraction step (S3).

[0034] In the above extraction step (S3), a plurality of small images are extracted so that some parts overlap with each other.

[0035] For example, when a femur is extracted into three small images, such as an upper image, a middle image, and a lower image, the lower part of the upper image and the upper part of the middle image are extracted to overlap each other, and the lower part of the middle image and the upper part of the lower image are extracted to overlap each other.

[0036] The above inspection step (S4) is a step of checking whether a femoral fracture exists in a plurality of the above small images using a pre-trained deep learning model.

[0037] That is, in the above inspection step (S4), the presence of various femoral fractures in a plurality of the above small images is inspected using a pre-learned deep learning model.

[0038] In the above inspection step (S4), since the inspection is performed using a small image rather than the original image, the presence or absence of a femoral fracture can be quickly determined.

[0039] The above coordinate detection step (S5) is a step for detecting the coordinate values ​​of the area where the femoral fracture is present, if a femoral fracture exists in the above inspection step (S4).

[0040] At this time, the above coordinate value means the location value of the area where the femoral fracture is located based on the preset reference point of the original image.

[0041] That is, the coordinate values ​​of the area where the femoral fracture is detected in the above small image are the location values ​​of the area where the femoral fracture is located based on a reference point set in advance in the above original image.

[0042] The above coordinate merging step (S6) is a step of merging the coordinate values ​​detected in the above coordinate detection step (S5).

[0043] Since multiple small images are extracted in an overlapping manner in the above extraction step (S3), when a femoral fracture occurs in an overlapping area of ​​the small images, the coordinate values ​​of the area where the femoral fracture occurred are detected in each of the different small images.

[0044] At this time, in the coordinate merging step (S6), if the coordinate values ​​detected in the coordinate detection step (S5) are the same, they are merged into one coordinate value.

[0045] The above display step (S7) displays the location corresponding to the coordinate value merged by the above coordinate merging step (S6) on the original image.

[0046] Figure 3 shows the coordinate values ​​merged by the coordinate merging step (S6) in the display step (S7), i.e., the area where the femoral fracture occurred, indicated by a blue box in the original image.

[0047] This allows for quick and accurate detection of the site of femoral fracture and displaying it on a high-resolution original image, enabling quick and accurate follow-up measures.

[0048] In particular, the present invention detects a femoral fracture site through a plurality of small images including the femur rather than detecting a femoral fracture in the entire high-resolution original image, so that the femoral fracture site can be detected more quickly and early.

[0049] The present invention can be applied not only to femoral fractures but also to methods for detecting and indicating fractures of other bones.

[0050] The method for detecting and displaying a femoral fracture according to the present invention is not limited to the above-described embodiment, and can be implemented in various ways within the scope permitted by the technical concept of the present invention.

[0051] The present invention has industrial applicability because it can be applied to a method for detecting and displaying femoral fractures.

Claims

1. A shooting step for obtaining an original image by shooting a high-resolution image of the patient's body including the lower body showing the bones; In the above original image, the femur area is detected and the detection step; An extraction step for extracting the original image of the detected femur area into a number of small-sized miniature images; An inspection step for checking whether a femoral fracture exists in the above small image using a pre-learned deep learning model; A coordinate detection step for detecting the coordinate values ​​of the area where a femoral fracture is present; A coordinate merging step for merging the detected coordinate values; It is made by including a display step of displaying the location corresponding to the merged coordinate value on the original image; The resolution of the original image and the small image above are the same, A method for detecting and displaying a femoral fracture, characterized in that the above coordinate values ​​are location values ​​of a site where a femoral fracture is present based on a preset reference point of the original image.

2. In claim 1, A method for detecting and displaying a femoral fracture, characterized in that a plurality of small images extracted in the above extraction step are extracted so that some parts overlap each other.

3. In claim 1, A method for detecting and displaying a femur fracture, characterized in that in the above detection step, the femur area is detected if the original image includes even a portion of the femur.

Images