TW202034349A - Focus detection apparatus and method thereof - Google Patents

Abstract

A focus detection apparatus and a method thereof are provided. In the method, a medical image at a first location is obtained, and a first focus is identified on the medical image of the first location. Another medical image at a second location is obtained, and a second focus is identified on the medical image of the second location. Then, whether the second focus is the first focus is determined. If the second focus is the first focus, the first and second focus is combined as a single focus to be presented. Accordingly, every medical image can be determined whether is the same focus as the previously identified focus in order, so as to provide more complete information for instant aided detection technology.

Description

Lesion detection device and method

The present invention relates to a medical image detection, and particularly relates to a real-time lesion detection device and method.

Nowadays, Computer Aided Detection (CADe) systems have been widely used clinically to automatically identify tumors, masses or calcifications and other lesions to assist doctors in diagnosis and treatment. Generally speaking, computer-aided detection technology needs to obtain all scanned images of the target part of an object, and then identify the lesion from all the scanned images. With the advancement of technology, some computer-assisted detection systems can even identify lesions in real time during the scanning of objects. However, the existing real-time identification technology can only know whether there is a lesion, and cannot immediately determine whether the newly identified lesion and the previously identified lesion are the same lesion, which affects the doctor's diagnosis and treatment judgment. Therefore, it is necessary to improve the focus detection technology for medical imaging.

In view of this, the present invention provides a lesion detection device and a method thereof, which can sequentially determine whether every other medical image is the same lesion, thereby providing a real-time detection technology with relatively complete data.

The lesion detection method of the present invention includes the following steps. Obtain the first-position medical image. Identify the first lesion in the medical image at the first location. Obtain medical images at the second location, and the first location is different from the second location. Identify the second lesion in the medical image at the second location. Determine whether the second lesion is the first lesion. If the second focus is the first focus, the first focus and the second focus are combined into a single focus to present.

In an embodiment of the present invention, the foregoing determining whether the second lesion is the first lesion includes the following steps. Based on the distance from the second focus to the center of gravity of the focus group of the first focus, the second focus is determined to be the first focus.

In an embodiment of the present invention, the above determining that the second lesion is the first lesion includes the following steps. Obtain the first distance from the second focus to the center of gravity of several focus groups, and the first focus belongs to one of those focus groups, and the center of gravity of each focus group is the position of all the focus of each focus group. The center of gravity. Determine whether the smallest first distance among these first distances is less than the distance threshold. If the minimum first distance is not less than the distance threshold, the second lesion is added to a new lesion group different from those lesion groups. If the minimum first distance is less than the distance threshold, then the second lesion is added to the k-th lesion group corresponding to the minimum first distance, and k is a positive integer. Obtain the second distance from all the lesions included in the k-th lesion group to the center of gravity of the k-th lesion group. Determine whether the largest second distance among those second distances is less than the distance threshold. If the maximum second distance is less than the distance threshold, the grouping ends.

In an embodiment of the present invention, after judging whether the largest second distance among the second distances is less than the distance threshold, the following steps are further included. If the maximum second distance is not less than the distance threshold, the y-th lesion corresponding to the maximum second distance is moved out of the k-th lesion group, and the position center of gravity of the k-th lesion group is updated, where y is a positive integer.

In an embodiment of the present invention, obtaining the medical image of the first position includes the following steps. Scan and record medical images through the scanner. Record the position of the scanner as the position of the medical image.

In an embodiment of the present invention, the above-mentioned lesion detection method further includes the following steps. If the scanner is not in contact, the recording of medical images and the display of the movement track of the scanner will be suspended.

In an embodiment of the present invention, the above-mentioned lesion detection method further includes the following steps. According to the three-dimensional position information corresponding to the medical images of several different positions, the movement track formed by the corresponding position of the medical image of each different position is presented in a three-dimensional manner. A three-dimensional virtual object is formed, and all suspicious lesion positions are displayed on the three-dimensional virtual object.

In an embodiment of the present invention, the above-mentioned lesion detection method further includes the following steps. Based on the location information corresponding to the medical images in several different locations, the virtual part range is generated.

In an embodiment of the present invention, the above-mentioned lesion detection method further includes the following steps. Obtain the information of past lesions and present the information of the past lesions.

The lesion detection device of the embodiment of the present invention includes a memory and a processor. The memory records several modules and medical images. The processor is coupled to the storage, and accesses and loads the modules recorded in the storage. Those modules include the lesion identification module and the same lesion judgment module. The lesion recognition module recognizes that the medical image at the first location has a first lesion, and the medical image at the second location recognizes that there is a second lesion, and the first location is different from the second location. The same lesion judgment module judges whether the second lesion is the first lesion, and if the second lesion is the first lesion, the first lesion and the second lesion are combined into a single lesion for presentation.

Based on the foregoing, the lesion detection device and method of the embodiment of the present invention can sequentially determine whether the currently identified lesion is the same lesion as the previously identified lesion during the scanning process. In addition, the combination of pause function, three-dimensional presentation, virtual location range presentation and past lesion information presentation will provide a rich and complete scanning assistive technology.

In order to make the above-mentioned features and advantages of the present invention more comprehensible, the following specific embodiments are described in detail in conjunction with the accompanying drawings.

FIG. 1 is a block diagram of components of a lesion detection device 1 according to an embodiment of the present invention. Please refer to FIG. 1, the lesion detection device 1 at least includes but is not limited to a storage 110, a display 130 and a processor 150. The lesion detection device 1 may be a computer host, a server, or even a scanner for real-time medical imaging (for example, a probe based on ultrasonic technology, an optical imaging lens, or any mobile probe, etc.).

The storage 110 may be any type of fixed or removable random access memory (RAM), read-only memory (ROM), flash memory (flash memory), traditional hard disk drive, solid state hard drive Disk (solid-state drive) or similar components, and used to record the lesion identification module 111, the same lesion determination module 112, the position processing module 113, the three-dimensional rendering module 114, the location marking module 115, and the past information providing module Group 116 and other software programs, two-dimensional or three-dimensional medical images (for example, automated breast ultrasound (ABUS), tomosynthesis, magnetic resonance imaging (MRI) and other images), lesions , Lesion location related information, lesion group, past lesion information and other data. The aforementioned modules, data, files and information will be described in detail in subsequent embodiments.

The display 130 may be a liquid crystal display (Liquid-Crystal Display, LCD), a light-emitting diode (Light-Emitting Diode, LED), an organic light-emitting diode (Organic Light-Emitting Diode, OLED) and other display technologies.

The processor 150 is connected to the storage 110 and the display 130, and can be a central processing unit (CPU), or other programmable general-purpose or special-purpose microprocessors (Microprocessors), digital signal processors (DSP), Programmable controller, special application integrated circuit (ASIC) or other similar components or a combination of the above components. In the embodiment of the present invention, the processor 150 is used to perform all operations of the lesion detection device 1 and can access and execute the modules recorded in the storage 110 described above.

In order to facilitate the understanding of the operation process of the embodiment of the present invention, a number of embodiments will be given below to describe in detail the focus detection process of the focus detection device 1 in the embodiment of the present invention for medical images. Hereinafter, various components and modules of the lesion detection device 1 will be used to describe the method according to the embodiment of the present invention. Each process of the method can be adjusted accordingly according to the implementation situation, and is not limited to this.

FIG. 2 is a flowchart illustrating a method for detecting lesions according to an embodiment of the present invention. Please refer to Figure 2, after capturing network packets, uploading by users, obtaining medical images through external or built-in storage media (such as pen drives, optical discs, external hard drives, etc.), or even directly through external or built-in scanners ( For example, an ultrasound probe, camera, video camera, etc.) scan a specific object (for example, a human or other animals) to record medical images, and the medical images are stored in the storage 110. It is worth noting that each medical image corresponds to location information (for example, the position of the scanner relative to the object to be scanned (for example, the distance and angle from a specific part, custom coordinates, etc.), or the medical image corresponds to the target Scan a specific area on the object, etc.), the position information will be used as the position of the medical image (may be a certain point or area) below.

And first, the lesion recognition module 111 will obtain the medical image of the first position (step S210), which means that the scanner records the medical image at the first position relative to the object to be scanned, or the medical image is for the object to be scanned The first position. Then, the lesion identification module 111 will identify whether there is a first lesion in the medical image at the first location. The lesion identification module 111 can use such as texture feature analysis, classifier, deep learning technology or any other computer-aided diagnosis technology to identify the lesion, which is not limited in the embodiment of the present invention. If the lesion recognition module 111 recognizes that the first lesion exists in the medical image at the first location (step S220), the location of the first lesion is recorded. For example, FIG. 3A is an example showing that there is a focus FF in the image block FL corresponding to the first position. In addition, assuming that the medical image at the first location is the first analyzed medical image, the lesion identification module 111 will classify the medical image into the first lesion group.

Then, the lesion recognition module 111 obtains the medical image at the second position through the same or similar method as in step S210 (that is, the scanner records and obtains the medical image at the second position relative to the object to be scanned, or the medical image is Regarding the second position on the object to be scanned) (step S230), the medical image at the second position is determined by the same or similar method as step S220 whether there is a second lesion (step S240), and the location of the second lesion is recorded. The second lesion is classified as a lesion that has not yet been grouped (that is, it does not belong to any grouped lesion, and may be temporarily stored in a queue or other temporary storage area). For example, in FIG. 3B, the second position beside the image block FL identified in FIG. 3A corresponds to the image block SL where the lesion SF is detected. It should be noted that this second position is different from the aforementioned first position, and the medical image of this second position is the next image following the medical image of the first position. In other words, the scanner does not generate other scanned medical images from the first position to the second position.

Next, the same lesion determining module 112 determines whether the second lesion is the first lesion identified by the medical image at the first location (step S250). In an embodiment, the same lesion determination module 112 determines the second lesion as the first lesion based on the distance from the second lesion to the center of gravity of the lesion group to which the first lesion belongs. For example, referring to FIG. 4, which is a flowchart of lesion grouping, the same lesion determination module 112 obtains a lesion (for example, a second lesion) that has not been grouped from a queue or other temporary storage area (step S410). With lesions determining module 112 then calculates the position to obtain the centroid of each group of lesions has not been grouped lesions are not yet added to the first through the distance D _i (1≤i≤K, _i are both positive integers with K, K is the lesion The number of groups, for example, the first distance from the ungrouped lesions to the center of gravity of the third lesion group is D ₃ (step S420). The positional center of gravity of each lesion group is the center of gravity of all the lesions contained in each. The same lesion determination module 112 will determine whether the minimum distance D _k ( 1≤k≤K , k is a positive integer) among the first distances D ₁ ~ D _K is less than the distance threshold D _T (ie, min(D ₁ , D ₂ , …, D _K )=D _k <D _T ?) (step S425). If the minimum first distance D _{k is} not less than the distance threshold value D _T , the same lesion judgment module 112 marks the ungrouped lesion as a wounded lesion and adds it to the new lesion group, and K is increased by one (step S430), and End the clustering operation for this lesion (step S480) (if there are other lesions in the queue that have not been clustered, return to step S410). Assuming that the second lesion is classified into a new lesion group, the same lesion judgment module 112 will display the first lesion and the second lesion respectively (that is, the second lesion is not the first lesion).

On the other hand, if the minimum first distance D _{k is} less than the distance threshold D _T , the same lesion determination module 112 adds the ungrouped lesions to the k- th lesion group corresponding to the minimum first distance D _k (step S440 ). Assuming that the first lesion belongs to the kth lesion group, the first and second lesions belong to the same lesion group (that is, the second lesion is the first lesion), and the same lesion judgment module 112 will display the first lesion through the display 130. A lesion and a second lesion are merged into a single lesion and presented (step S260). It should be noted that the combined display here may be the combination of the outlines of the two lesions, and the presentation in the same visual style. For example, referring to FIGS. 3A to 3C at the same time, the same lesion judgment module 112 combines the lesions FF and SF identified in FIGS. 3A and 3B into the lesion PF, and displays the combined lesion PF on the display 130 (as shown in FIG. 3C) Show). It should be noted that the contour of the lesion PF may be adjusted according to the contour of the lesion FF and SF, and the shape of the lesion shown in Figures 3A to 3C is only a convenient example for illustration.

Since medical images can be used to determine whether the lesions are the same as the previously identified lesions one by one, it is applied to real-time scanning technology. During the scanning process of the scanner moving, the lesion detection device 1 can instantly detect multiple lesions belonging to the same lesion. Combined display to improve the efficiency of diagnosis. In addition, even if the medical images are not obtained from the scanner in real time (for example, downloaded through the Internet, input from a pen drive, etc.), the embodiments of the present invention can still be applied.

In addition, in order to improve the accuracy of the same lesion judgment, please refer to Figure 4. If a lesion is added to an existing lesion group (this example is the kth lesion group), the same lesion judgment module 112 will calculate to obtain the kth lesion group. the center of gravity of the focus group all lesions to k other than invasive lesions of the group included in the second group of focus distance _{D k, j (1 ≤j≤Y,} j and Y are both positive integer, Y is the k lesions number of foci in the group, e.g., a second center of gravity from the first group of five lesions k to k lesion lesion is a group of D _{k, 5)} (step S450), and determines that the second distance D _{k ,1} ~ D _k,Y is the largest second distance D _k,y (1 ≤y≤Y , y is a positive integer) less than the distance threshold D _T (ie, max(D _k,1 , D _k,2 , …, D _{k, Y} )=D _{k, y} <D _T ?) (step S455). If the maximum second distance D _{k,y is} less than the distance threshold D _T , the same lesion determination module 112 ends the grouping (step S480) (if there are other lesions in the queue that have not yet been grouped, return to step S410). And if this maximum second distance D _{k, y} is not less than the distance threshold value D _T, the determining module 112 of this same lesion maximum second distance D _{k, y} y lesions corresponding first focus group out of the k (step S460 ) (Y minus one) (In other embodiments, the same lesion determination module 112 may also directly remove all lesions whose second distance is greater than the distance threshold D _T from the kth lesion group), and update the k The position of the center of gravity of the lesion group. The y- th lesion removed in step S460 will be re-assigned as a lesion that has not yet been grouped (step S470), and the same-lesion judging module 112 will recalculate the second distance D _{k of} the k- th lesion group with the updated position center of gravity _{. 1} ~ D _k,Y (return to step S420).

It should be noted that, in addition to the aforementioned determination of the same lesion based on the distance between the center of gravity of the location of the lesion group, in other embodiments, the same lesion determination module 112 may also directly compare the second lesion with another adjacent lesion. Merging, or merging with another lesion with the highest texture feature similarity within a certain distance range, etc., to be adjusted by the user according to his needs.

In addition to the aforementioned function of real-time judgment of the same lesion, the lesion detection device 1 also has other functions. In one embodiment, the lesion detection device 1 is combined with the application of a separate or built-in scanner, and the position processing module 113 can synchronously present the movement trajectory of the scanner through the display 130, and based on image recognition and sensing on the scanner It is determined that the scanner touches or leaves the surface of the object (for example, the surface of the skin, etc.) by means such as infrared rays, switches, proximity sensors, etc. If the position processing module 113 detects that the scanner does not touch the surface of the object, it will suspend the scanner's recording of medical images and the display of the scanner's movement track.

In one embodiment, in order to enhance the visual experience and the accuracy of auxiliary diagnosis, the three-dimensional rendering module 114 obtains the three-dimensional position information corresponding to the medical images of several different positions obtained by the position processing module 113 (for example, in the object to be scanned). The coordinates, relative positions, etc. on the upper part of the screen), and the movement trajectory formed by the corresponding position of the medical image of each different position presented in a three-dimensional manner through the display 130 (ie, the movement trajectory of the scanner during scanning). In addition, the three-dimensional presentation module 114 can also form a three-dimensional virtual object based on the specific part of the object to be measured, and based on the recognition result of the lesion identification module 111, present all suspicious lesion locations on the three-dimensional virtual object through the display 130.

In one embodiment, the part marking module 115 marks the position information of the scanner at a specific position on the object to be scanned, and generates a virtual part range based on the position information corresponding to the medical images in several different positions. For example, please refer to FIG. 5 which is a screen 500 presented by the display 130. Based on the four positioning points L1 obtained by the scanner positioning, the part marking module 115 calculates the virtual part range PA based on the appearance of the target part (breast in this example). It should be noted that this embodiment does not limit the number of positioning points, as long as it is greater than two.

In addition, the existing real-time scanning technology usually only presents the medical image obtained from the current scan. In an embodiment of the present invention, the past information providing module 116 can be obtained from an external source (e.g., network, database (e.g., Medical image archiving and communication system (Picture Archiving and Communication System, PACS), Hospital Information System (Hospital Information System, HIS), Radiology Information System (RIS), local database, etc.), flash drive, etc.) or storage The device 110 obtains the past lesion information of the object to be scanned (for example, the past lesion location, image, image report, pathology report, etc.), and displays the past lesion information on the currently acquired or scanned medical image through the display 130. For example, please refer to FIG. 6 for a screen 600 displayed on the display 130. The screen 600 presents the lesion CF identified by the current lesion identification module 111, the past lesion PF in the past lesion information, and the past recording time information D1. In this way, the diagnostician can easily compare the current and past lesions, effectively improving the accuracy of judgment.

For the integration of lesion information, the processor 150 displays the subject's past lesion information on the interface through the display 130 before/while the user reads the image, and the loaded past lesion information can be automatically compared with the current lesion. In addition, in response to the user's selection of a specific area, the past information providing module 116 can load medical images and related lesion information of the selected area and the patient at different times. If combined with the multi-screen browsing mode (the display 130 displays multiple medical images at the same time), the user can quickly browse multiple recorded medical images. The past information providing module 116 can also provide sequential presentation of different medical images in the scanning direction, for example, browsing from the nipple radially to the 12 o'clock direction. Alternatively, the past information providing module 116 may provide a scanning image plane to view the image, for example, browse and play in a sagittal view.

To sum up, the lesion detection device and method of the embodiment of the present invention can sequentially determine whether to merge the same lesion, and apply it to the real-time scanning technology. The same lesion can be presented in real time with the movement of the scanner. The efficiency of auxiliary diagnosis. In order to provide a rich and complete scanning assistance technology, the embodiment of the present invention further combines the pause function, three-dimensional presentation, virtual part range presentation and past lesion information presentation.

Although the present invention has been disclosed in the above embodiments, it is not intended to limit the present invention. Anyone with ordinary knowledge in the technical field can make some changes and modifications without departing from the spirit and scope of the present invention. The scope of protection of the present invention shall be determined by the scope of the attached patent application.

1: Lesion detection device 110: Storage 111: Lesion recognition module 112: Same lesion judgment module 113: Position Processing Module 114: 3D rendering module 115: part marking module 116: Past information provision module 130: display 150: processor S210~S260, S410~S480: steps FL, SL: image block FF, SF, PF, CF: lesions L1: anchor point PA: Virtual part range 500, 600: screen D1: Past recording time information

FIG. 1 is a block diagram of components of a lesion detection device according to an embodiment of the present invention. Fig. 2 is a flowchart of a lesion detection method according to an embodiment of the present invention. Figures 3A~3C are an example to illustrate the judgment and analysis of the same lesion. Fig. 4 is a flowchart of lesion grouping according to an embodiment of the present invention. Figure 5 is an example illustrating the virtual part range. Figure 6 is an example of past lesion information.

S210~S260: steps

Claims (10)

A method for detecting lesions, including: Obtain a first-position medical image; Identify the presence of a first lesion in the medical image at the first location; Obtaining a medical image of a second location, where the first location is different from the second location; Identify the presence of a second lesion in the medical image at the second location; Determine whether the second lesion is the first lesion; and If the second focus is the first focus, then the first focus and the second focus are combined into a single focus for presentation. For the lesion detection method described in item 1 of the scope of patent application, the step of judging whether the second lesion is the first lesion includes: According to the distance from the second focus to the center of gravity of the focus group of the focus group to which the first focus belongs, the second focus is determined to be the first focus. The lesion detection method according to item 2 of the scope of patent application, wherein the step of determining that the second lesion is the first lesion includes: Obtain the first distance from the second focus to the center of gravity of the multiple focus groups that have not yet been added, where the first focus belongs to one of the focus groups, and the center of gravity of each focus group is The center of gravity of all the lesions contained in each; Judging whether a minimum first distance among the first distances is less than a distance threshold; If the minimum first distance is not less than the distance threshold, adding the second lesion to a new lesion group different from the lesion groups and marking it as a wounding lesion; If the minimum first distance is less than the distance threshold, add the second lesion to the k-th lesion group corresponding to the minimum first distance, and k is a positive integer; Obtain the second distance from all the lesions in the kth lesion group except wound lesions to the center of gravity of the kth lesion group; Determine whether a largest second distance among the second distances is less than the distance threshold; and If the maximum second distance is less than the distance threshold, the grouping is ended. For example, in the lesion detection method described in item 3 of the scope of patent application, after the step of determining whether a largest second distance among the second distances is less than the distance threshold value, it further includes: If the maximum second distance is not less than the distance threshold, the yth lesion corresponding to the maximum second distance is removed from the kth lesion group, and the position center of gravity of the kth lesion group is updated, where y is a positive integer. For the lesion detection method described in item 1 of the scope of patent application, the step of obtaining the medical image of the first position includes: Scan and record the medical image through a scanner; and Record the position of the scanner as the position of the medical image. The lesion detection method described in item 5 of the scope of patent application further includes: If the scanner is not in contact, the recording of the medical image and the display of the movement track of the scanner are suspended. The lesion detection method described in item 1 of the scope of patent application further includes: According to the three-dimensional position information corresponding to the medical images of different positions, the movement track formed by the corresponding positions of the medical images of the different positions presented in a three-dimensional manner; and A three-dimensional virtual object is formed, and all suspicious lesion positions are displayed on the three-dimensional virtual object. The lesion detection method described in item 1 of the scope of patent application further includes: According to the position information corresponding to the medical images in different positions, a virtual part range is generated. The lesion detection method described in item 1 of the scope of patent application further includes: Obtain at least one past lesion information; and Present the at least one past lesion information. A focus detection device includes: A storage for recording multiple modules and at least one medical image; and A processor coupled to the storage, and accesses and loads the modules recorded in the storage, the modules include: A lesion recognition module that recognizes that a first lesion exists in a medical image at a first location, and that a second lesion exists in a medical image at a second location, where the first location is different from the second location; and The same lesion judgment module judges whether the second lesion is the first lesion, and if the second lesion is the first lesion, the first lesion and the second lesion are combined into a single lesion for presentation.

Images