TWM667262U - System of brain imaging detection - Google Patents

Abstract

The present invention provides a system of brain imaging detection, which includes a medical imaging module for dividing a brain image into a left brain medical image and a right brain medical image and obtaining the grayscale values of multiple pixels thereof, ​​respectively; an operation module for mirror-flipping one of the left and right brain medical images; a subtraction module for performing a grayscale value subtraction with the mirror-flipped medical image and the other medical image to obtain a plurality of grayscale differences of the pixels; and a comparison module for determining whether the grayscale difference of each pixel exceeds a grayscale difference threshold and whether a number of the pixels having the grayscale different exceeding the grayscale difference threshold is greater than or equal to a pixel number threshold, and when the number of the pixels having the grayscale difference exceeding the grayscale difference threshold is greater than or equal to the pixel number threshold, determining that the brain image is abnormal.

Description

Brain imaging detection system

The invention relates to a detection system, and more particularly to a brain image detection system.

Magnetic resonance imaging (MRI) is characterized by high image resolution and can clearly show subtle changes in brain structure and tissue, so it is crucial for diagnosing brain diseases. Compared with other imaging examinations such as CT (Computed Tomography) or X-ray, MRI is not only a non-invasive examination method, but also has no radiation problems. It uses magnetic fields and harmless radio waves to capture brain tissue and produce high-resolution images, so it is safer for patients. In addition, the radiation-free nature of MRI is also suitable for long-term follow-up of patients, especially for patients who need repeated examinations, such as epilepsy patients or multiple sclerosis patients. Therefore, MRI can not only be used for diagnosis, but also for improving monitoring of treatment effects, accelerating the assessment of disease progression, etc. For example, for patients with brain tumors, MRI can be used to detect the size, location, and blood supply of the tumor, and track changes after treatment.

Generally speaking, after a patient has undergone an MRI scan, a medical professional, such as a radiologist, will then manually interpret the MRI images to assess whether the structures and tissues of various brain regions are normal and to check whether there are any abnormalities in the brain images. Based on the interpretation of MRI images, doctors can diagnose various brain diseases, including but not limited to tumors, strokes, cerebral hemorrhages, multiple sclerosis, epilepsy, etc. Furthermore, based on the abnormal features in the MRI images, the type, location, and degree of deterioration of the disease can be further determined. However, the large amount of imaging data generated by MRI requires radiologists to spend a lot of time to make an accurate diagnosis, which may cause patients to wait too long for the doctor's diagnosis report, thereby delaying the progress of treatment.

Taking the neuroradiology department of a general hospital as an example, each patient who undergoes a brain MRI examination will receive an average of about 500 medical images. Assuming there are 50 patients per day, there will be nearly 25,000 MRI images per day that require manual medical diagnosis by radiologists. This results in patients having to wait an average of one to two weeks to receive a diagnosis report, and such a long diagnosis time may miss the best time for treatment.

In recent years, with the development of artificial intelligence in medical imaging, most of the industry has turned to artificial intelligence solutions to determine whether brain images are normal or abnormal (or abnormal). However, although the training of artificial intelligence models for images is relatively fast, it still requires long-term learning through the characteristics of the disease to obtain the lesion as the basis for identification judgment. To date, the most common artificial intelligence models, such as the models to date, mostly use the forward list method to make judgments and identifications. This type of discrimination model is prone to uncertainty or identification doubts for lesions in special locations. In addition, since a single DICOM (Digital Imaging and Communications in Medicine) image contains a lot of detailed data and is very complex, it greatly affects the system analysis and calculation time. In other words, no matter how fast the medical image artificial intelligence model is used for interpretation, it will take tens of seconds to several minutes.

In summary, in order to improve diagnostic efficiency, it is an important topic to provide a more efficient clinical tool to help radiologists screen images so that they can more quickly identify normal and abnormal tissues in brain medical images, thereby saving doctors' time in reviewing images, allowing them to have more time to focus on cases that require more in-depth research, and allowing patients to receive better quality medical care.

Therefore, the purpose of the present invention is to provide a brain image detection system that can quickly screen out abnormal brain images, increase the speed of analyzing a large number of images, and thus quickly prioritize attention to brain images with structural abnormalities.

To achieve the above-mentioned purpose, the present invention provides a brain image detection system, including a medical image module, a computing module, a subtraction module and a comparison module, wherein the medical image module is electrically connected or communication-connected to the computing module, the subtraction module and the comparison module. The medical image module divides the brain image into a left-side brain medical image and a right-side brain medical image and obtains multiple grayscale values of multiple pixels thereof respectively. The computing module performs a mirror image flipping on one of the left-side and right-side brain medical images. The subtraction module performs a grayscale value subtraction on the flipped image from the other image to obtain multiple grayscale difference values of multiple pixels. The comparison module determines whether the grayscale difference of each pixel exceeds the grayscale difference threshold, and determines whether the number of pixels whose grayscale difference exceeds the grayscale difference threshold is greater than or equal to the pixel number threshold. When the number of pixels whose grayscale difference exceeds the grayscale difference threshold is greater than or equal to the pixel number threshold, the brain image is determined to be abnormal.

In one embodiment, the brain image is a two-dimensional brain image.

In one embodiment, when the brain image is a three-dimensional brain image, the medical imaging module performs de-cranial processing on the three-dimensional brain image.

In one embodiment, the medical imaging module reduces the dimensionality of the three-dimensional brain image to one or more two-dimensional brain images, and performs black removal processing on the one or more two-dimensional brain images.

In one embodiment, when in at least one two-dimensional brain image, the sum of the number of pixels whose grayscale difference exceeds the grayscale difference threshold is greater than or equal to the pixel number threshold, the brain image is determined to be abnormal.

In one embodiment, the brain image is a DWI brain image, a DTI brain image, a T1 brain image, or a T2 brain image.

In one embodiment, the medical imaging module captures the mirror-flipped image and one or more corresponding pixel regions in another image, and then the subtraction module performs grayscale value subtraction to obtain multiple grayscale difference values of multiple pixels in the one or more pixel regions.

In one embodiment, the subtraction module calculates the absolute values of the grayscale differences; the grayscale difference threshold is 200, and the pixel number threshold is 11.

In the following, the specific embodiment of the present invention will be described according to the drawings. To avoid redundant description, in this embodiment, the symbols of some components with the same functions are represented by the same symbols.

In addition, it should be noted that the brain is one of the most important organs in the human body, and its importance is beyond words. As the core of the central nervous system, the brain is responsible for controlling and coordinating various physiological and psychological activities of the body, as well as for perception and feeling, processing and interpreting sensory information from the outside world and the inside, including vision, hearing, touch, taste and smell. When it comes to brain disease diagnosis, monitoring, trauma assessment, and scientific research, magnetic resonance imaging (MRI) is usually chosen for use. From the perspective of brain structure, the two sides (left and right) of the brain are anatomically symmetrical. There are many structures and regions between the two hemispheres of the brain, such as the frontal lobe, parietal lobe, lateral lobe, etc., which are roughly the same in size and shape. This structural symmetry is the result of genetic and physiological mechanisms during brain development. The present application uses the grayscale difference between specific regions of left and right brain medical images to determine whether brain images are abnormal.

As shown in FIG1 , in the brain image detection method used by the brain image detection system proposed in the present invention, step S1 is to divide the brain image into a left side brain medical image and a right side brain medical image and respectively obtain multiple grayscale values of multiple pixels. Specifically, the patient's three-dimensional brain image is first obtained by a magnetic resonance imaging scanner, and then the three-dimensional brain image is processed by a PACS system (Picture Archiving and Communication System). For example, a three-dimensional brain image can be de-cranialized (i.e., skull removed), and the three-dimensional brain image can be reduced to one or more two-dimensional brain images, wherein the dimensionality reduction can represent reducing the three-dimensional image to a two-dimensional image or performing dimensionality reduction by mathematical operation, and one or more two-dimensional brain images can be de-blackened (i.e., removing redundant black edges), and the two-dimensional brain images formed after de-cranialization and de-blackening can be divided into a left brain medical image and a right brain medical image, and multiple grayscale values of multiple pixels of the left brain medical image and multiple grayscale values of the right brain medical image can be taken respectively. Brain medical images take multiple grayscale values of multiple pixels. Generally speaking, the structures of the left and right brains are symmetrical. When the overall difference value of the multiple grayscale values of the multiple pixels of the left brain medical image and the multiple grayscale values of the multiple pixels of the right brain medical image is lower, it means that the brain image is normal. In other words, when the difference value of the multiple grayscale values of the multiple pixels of the left brain medical image and the multiple grayscale values of the multiple pixels of the right brain medical image in one or more specific areas is higher, it means that the brain image is abnormal.

Please refer to FIG. 1 and FIG. 2A-2E. Step S2 is to perform mirror flipping on one of the left and right brain medical images. In this embodiment, the left brain medical image can be mirror flipped (first flipping mode) to obtain a mirror flipped left brain medical image as shown in FIG. 2C; in other embodiments, the right brain medical image can be mirror flipped (second flipping mode) (not shown). This embodiment adopts the first flipping mode mirroring. Next, step S3 is to subtract the grayscale value of the mirror-flipped image (e.g., the left side of the brain medical image after mirror-flipping) from another image (e.g., the right side of the brain medical image) to obtain multiple grayscale difference values of multiple pixels respectively. Since the structures of the left side of the brain and the right side of the brain are symmetrical, the grayscale value of the corresponding one or more pixel areas in the captured mirror-flipped image is subtracted from that in the other image to obtain multiple grayscale difference values of multiple pixels respectively. Furthermore, the grayscale values of the multiple pixels of the left-side brain medical image are subtracted from the grayscale values of the multiple pixels of the right-side brain medical image in one or more corresponding pixel regions to obtain multiple grayscale difference values of the multiple pixels respectively; for example, the mirror-flipped left-side brain medical image shown in FIG2C can be overlapped with the right-side brain medical image shown in FIG2D, and the multiple grayscale values of the corresponding pixels of the two images are subtracted, and then an image as shown in FIG2E can be obtained, which represents the result of multiple grayscale difference values of multiple pixels. In other embodiments, the grayscale values of the multiple pixels of the right side brain medical image after mirror flipping can be subtracted from the grayscale values of the multiple pixels of the left side brain medical image in one or more corresponding pixel regions to obtain multiple grayscale difference values of the multiple pixels. Each grayscale difference value is calculated by taking its absolute value. For example, the corresponding one or more pixel regions may be located in specific regions such as the frontal lobe, the parietal lobe, and the lateral lobe. Subsequently, the absolute values of the multiple grayscale difference values of the multiple pixels can be used to determine whether the corresponding one or more pixel regions in the mirror flipped image and the other image are abnormal.

As mentioned above, the two-dimensional brain image is divided into left and right brains, the left brain image is mirror-flipped and the grayscale value is subtracted from the right brain image to obtain the absolute value (first flip mode), or the right brain image is mirror-flipped and the grayscale value is subtracted from the left brain image to obtain the absolute value (second flip mode), and the grayscale difference value is between 0 and 255, which means that the larger the grayscale difference value of the two-dimensional brain image, the greater the difference in grayscale values between the left and right brains, and at the same time, it means that one or more corresponding pixel areas may have abnormal image areas.

In addition, please refer to FIG. 1 and FIG. 1A , step S4 is to determine whether the grayscale difference value of each pixel exceeds the grayscale difference threshold, and to determine whether the number of pixels whose grayscale difference value exceeds the grayscale difference threshold is greater than or equal to the pixel number threshold. Further, the grayscale difference threshold for determining each pixel as abnormal may have some errors. During statistical analysis, a grayscale difference threshold for determining each pixel as abnormal must be found to filter out the grayscale difference values that can truly distinguish the abnormalities of each pixel in the brain image. At the same time, the pixel number threshold for the total number of pixels that are judged as abnormal must be determined to filter out the number of abnormal pixels that can truly distinguish the abnormal cases in a specific area of the brain image. While searching for these two values to distinguish between normal and abnormal cases, the factors of the classification scoring metrics (such as: Accuracy, Recall, Precision and F1 score) can be calculated as the judgment basis for searching for the best grayscale difference threshold and pixel number threshold. In other words, the data displayed in the classification scoring metrics table can search for these two values to distinguish between normal and abnormal images, and serve as data judgment factors for searching for the best grayscale difference threshold (GrayScale) and pixel number threshold (Threshold). The classification scoring metrics table is shown below: Classification scoring metrics table GrayScale Pixel number threshold (Threshold) Accuracy Recall Precision F1 score 190 10 0.686957 0.922013 0.664551 0.772392 190 11 0.694928 0.909434 0.67444 0.774505 190 12 0.707246 0.896855 0.688889 0.779235 200 10 0.75942 0.792453 0.790464 0.791457 200 11 0.767391 0.777358 0.811024 0.793834 200 12 0.763043 0.759748 0.816216 0.786971 210 10 0.716667 0.583648 0.885496 0.703563 210 11 0.700725 0.545912 0.893004 0.677596 210 12 0.692754 0.52327 0.902386 0.66242

Furthermore, as shown in FIG. 1 and FIG. 1A, when the number of pixels whose grayscale difference exceeds the grayscale difference threshold is greater than or equal to the pixel number threshold, step S5 is to judge that the brain image is abnormal; in addition, when the number of pixels whose grayscale difference does not exceed the grayscale difference threshold is less than the pixel number threshold, step S6 is to judge that the brain image is normal. Further, when the total number of pixels whose grayscale difference exceeds the grayscale difference threshold in at least one or more two-dimensional brain images is greater than or equal to the pixel number threshold, the brain image is judged to be abnormal, wherein at least one or more two-dimensional brain images can form a three-dimensional brain image. From the classification score index table, we can know that the grayscale difference threshold ranges from 190 to 210, and the pixel number threshold ranges from 10 to 12. Preferably, when the grayscale difference threshold is 200 and the pixel number threshold is 11, the Accuracy is the highest at 0.767391, which is used as the judgment basis for searching for the best grayscale difference threshold and pixel number threshold. In other words, a grayscale difference threshold is found on the difference image generated by superimposing the medical image after mirror flipping and another medical image, and the values above the grayscale difference threshold are defined as abnormal grayscale, and it is summarized how many pixels in the case with respect to the pixel number threshold above this grayscale difference threshold can be called abnormal images. From the above, we can see that when the threshold for defining abnormal grayscale difference and the threshold for the number of pixels with abnormal grayscale difference are 200 and 11 respectively, there will be optimized Accuracy, Recall, Precision and F1 score. Figure 1A shows that the maximum classification benefit can be achieved when the pixel threshold is 11.

Furthermore, as shown in FIG. 1 and FIG. 2 , FIG. 2 is another method flow chart of the brain image detection method used by the novel brain image detection system, including steps S01 to S04 and steps S1 to S6, wherein steps S1 to S6 can refer to the aforementioned embodiments and will not be described in detail here. In this embodiment, step S01 is to capture a three-dimensional brain image, and this application is directed to a detection method for quickly confirming whether the image is abnormal or normal based on the difference in left-right symmetry of the brain. In this embodiment, the brain image can be a DWI (Diffusion weighted imaging) brain image, or a DTI (Diffusion tensor imaging) brain image, or a T1 brain image, or a T2 brain image. Step S02 is to perform de-cranial processing on the three-dimensional brain image. Next, step S03 is to reduce the dimension of the three-dimensional brain image to a two-dimensional brain image (as shown in FIG. 2A ). For example, after the PACS system receives the three-dimensional brain diffusion weighted image (DWI), it cuts the three-dimensional brain diffusion weighted image from top to bottom into a series of two-dimensional brain images. Then, step S04 is to perform black removal processing on the two-dimensional brain image (as shown in FIG. 2B ). For the three-dimensional brain image, three-dimensional skull removal and two-dimensional redundant black edge processing are performed to ensure that the image is focused on the brain itself, so as to extract the two-dimensional slice image after removing the skull and redundant black edges, and extract the image with obvious grayscale difference in the specific pixel area of the left and right brain medical images.

Please refer to FIG. 3 , the novel brain image detection system 300 includes a medical image module 310, a computing module 320, a subtraction module 330 and a comparison module 340, and the medical image module 310 is electrically connected to the computing module 320, the subtraction module 330 and the comparison module 340. For example, the medical image module 310 is a PACS system (Picture Archiving and Communication System, medical image storage and transmission system), the computing module 320 is a tablet computer, a laptop computer or a desktop computer, and the present application is not limited to the type of the computing module 320, the subtraction module 330 is a subtractor, and the comparison module 340 is a comparator.

The medical imaging module 310 divides the brain image into a left-side brain medical image and a right-side brain medical image and obtains multiple grayscale values of multiple pixels. The operation module 320 performs mirror flipping on one of the left-side and right-side brain medical images. The subtraction module 330 performs grayscale subtraction on the flipped image from the other image to obtain multiple grayscale difference values of multiple pixels. The comparison module 340 determines whether the grayscale difference value of each pixel exceeds the grayscale difference threshold value, and determines whether the number of multiple pixels whose grayscale difference value exceeds the grayscale difference threshold value is greater than or equal to the pixel number threshold value. When the number of pixels whose grayscale difference exceeds the grayscale difference threshold is greater than or equal to the pixel number threshold, the brain image is judged to be abnormal; when the number of pixels whose grayscale difference exceeds the grayscale difference threshold is less than the pixel number threshold, the brain image is judged to be normal.

The brain images are two-dimensional brain images.

When the brain image is a three-dimensional brain image, the medical image module 310 performs de-cranial processing on the three-dimensional brain image.

The medical image module 310 reduces the dimensionality of the three-dimensional brain image to one or more two-dimensional brain images, and performs black removal processing on the one or more two-dimensional brain images.

When the sum of the number of pixels whose grayscale difference exceeds the grayscale difference threshold in at least one or more two-dimensional brain images is greater than or equal to the pixel number threshold, the brain image is judged to be abnormal.

The brain image is a DWI brain image, a DTI brain image, a T1 brain image, or a T2 brain image.

The medical imaging module 310 captures the mirror-flipped image and one or more corresponding pixel regions in another image, and then the subtraction module 330 performs grayscale value subtraction to obtain multiple grayscale difference values of multiple pixels in the one or more pixel regions.

The subtraction module 330 calculates the gray level difference by taking its absolute value; the gray level difference threshold is 200, and the pixel number threshold is 11.

In summary, the present invention provides a brain image detection system that can quickly screen out abnormal brain images, effectively improve the speed of analyzing a large number of brain images, and use the calculated classification score index as a judgment basis for searching for the best grayscale difference threshold and pixel number threshold, thereby giving priority to brain images with structural abnormalities. The operation is convenient, fast and accurate.

S01~S04, S1~S6: Steps 300: Brain image detection system 310: Medical imaging module 320: Calculation module 330: Subtraction module 340: Comparison module

FIG. 1 is a flow chart of a brain image detection method used in the novel brain image detection system.

FIG. 1A is a curve diagram of evaluation indicators of the brain image detection method used by the novel brain image detection system.

FIG. 2 is another flow chart of the brain image detection method used by the novel brain image detection system.

FIG. 2A is an example of a two-dimensional image of the present invention.

FIG. 2B is an example of a two-dimensional image with black edges and skull removed according to the present invention.

FIG. 2C is an example of the mirror image flipping of the left side brain medical image of the present invention.

FIG2D is an example of a right side brain medical image of the present invention.

FIG. 2E is an example of the result of overlapping FIG. 2C and FIG. 2D and taking the grayscale difference thereof.

FIG. 3 is a schematic diagram of the novel brain imaging abnormality detection system.

300: Brain imaging detection system

310: Medical imaging module

320: Computation module

330: Subtraction module

340: Comparison module

Claims (9)

A brain image detection system includes: A medical image module, which divides a brain image into a left-side brain medical image and a right-side brain medical image and obtains multiple grayscale values of multiple pixels; An operation module, which performs mirror image flipping on one of the left-side and right-side brain medical images; A subtraction module, which subtracts the grayscale value of the flipped image from the other image to obtain multiple grayscale difference values of the multiple pixels; and A comparison module, the comparison module determines whether the grayscale difference of each pixel exceeds a grayscale difference threshold, and determines whether the number of the multiple pixels whose grayscale difference exceeds the grayscale difference threshold is greater than or equal to a pixel number threshold; Wherein, the medical imaging module is electrically connected or communication-connected to the operation module, the subtraction module and the comparison module; Wherein, when the number of the multiple pixels whose grayscale difference exceeds the grayscale difference threshold is greater than or equal to the pixel number threshold, the brain image is judged to be abnormal. A brain image detection system as described in claim 1, wherein the brain image is a two-dimensional brain image. A brain image detection system as described in claim 2, wherein when, in at least one of the two-dimensional brain images, the sum of the number of pixels having the grayscale difference exceeding the grayscale difference threshold is greater than or equal to the pixel number threshold, the brain image is judged to be abnormal. A brain image detection system as described in claim 1, wherein when the brain image is a three-dimensional brain image, the medical imaging module performs de-cranial processing on the three-dimensional brain image. A brain image detection system as described in claim 4, wherein the medical image module reduces the dimensionality of the three-dimensional brain image to one or more two-dimensional brain images, and performs black removal processing on the one or more two-dimensional brain images. A brain image detection system as described in claim 5, wherein when, in at least one of the two-dimensional brain images, the sum of the number of pixels whose grayscale difference exceeds the grayscale difference threshold is greater than or equal to the pixel number threshold, the brain image is judged to be abnormal. A brain image detection system as described in claim 1, wherein the brain image is a DWI brain image, or a DTI brain image, or a T1 brain image, or a T2 brain image. A brain image detection system as described in claim 1, wherein the medical imaging module captures the mirror-flipped image and one or more corresponding pixel regions in the other image, and then the subtraction module performs grayscale value subtraction to obtain the multiple grayscale difference values of the multiple pixels in the one or more pixel regions. As described in claim 1, the subtraction module calculates the multiple grayscale difference values by taking their absolute values; the grayscale difference threshold is 200, and the pixel number threshold is 11.