CN112545480B

CN112545480B - System and method for detecting benign lesions of prostate and seminal vesicle on MRI

Info

Publication number: CN112545480B
Application number: CN201910920341.3A
Authority: CN
Inventors: 岳新; 贺长征; 张虽虽; 王霄英
Original assignee: Beijing Smarttree Medical Technology Co Ltd
Current assignee: Beijing Smarttree Medical Technology Co Ltd
Priority date: 2019-09-26
Filing date: 2019-09-26
Publication date: 2022-11-25
Anticipated expiration: 2039-09-26
Also published as: CN112545480A

Abstract

The invention provides a system for detecting benign lesions of prostate and seminal vesicle on MRI, which comprises an AI scheduling module, a data acquisition module and a data processing module, wherein when a patient finishes scanning an examination item, namely the examination of prostate mpMRI, the AI scheduling module extracts header file information of a DICOM image and searches a T2WI sequence and a T1WI sequence; the prostate segmentation and dissection module segments the T2WI image into a prostate gland, a peripheral zone, a migration zone, a central zone, an anterior fibrous stroma zone, a urethra and bilateral seminal vesicle glands; each benign lesion auxiliary diagnosis module receives the sequence image and the segmentation data matched with the benign lesion auxiliary diagnosis module and respectively outputs diagnosis data of prostatic hyperplasia, prostatitis, prostatic cyst, prostatic hemorrhage, seminal vesiculitis, seminal vesicle atrophy and seminal vesicle hemorrhage; the structured reporting module automatically generates diagnostic results based on the diagnostic data. The invention also discloses a method for detecting benign lesions of prostate and seminal vesicle on MRI. The invention can automatically fill the diagnosis data of benign prostate lesion in the report, thereby improving the working efficiency of doctors.

Description

System and method for detecting benign lesions of prostate and seminal vesicle on MRI

Technical Field

The present invention relates to the field of medical information, and more particularly, to a system and method for detecting benign lesions of the prostate and seminal vesicle on MRI.

Background

Chronic benign disease of the prostate is a common condition in elderly male patients, including: benign Prostatic Hyperplasia (BPH), prostatitis, etc. can cause symptoms such as dysuria and frequent micturition, and affect the quality of life of patients. More importantly, BPH and prostatitis often occur simultaneously with malignant tumors (prostate cancer), and during image examination, the characteristics of several different common diseases are crossed, which causes difficulty in prostate cancer detection. Generally, BPH interferes with the diagnosis of Transitional Zone (TZ) prostate cancer, resulting in missed diagnosis, and prostatitis interferes with the diagnosis of Peripheral Zone (PZ) prostate cancer, resulting in misdiagnosis. Therefore, in the case of multiparametric Magnetic Resonance Imaging (mpMRI) of the prostate, benign changes such as BPH and prostatitis should be detected as much as possible to explain the symptoms of the patient and to distinguish them from clinically significant carcinoma (prostate). In the prior art, data of benign lesions of prostate and seminal vesicle are not automatically filled in a report, and the benign lesions are mostly distributed diffusely on an image, so that the benign lesions often influence the detection of prostate cancer, and the working efficiency and the diagnosis effect of doctors are reduced.

Disclosure of Invention

In view of the above, the main objective of the present invention is to provide a system and a method for detecting benign prostate and seminal vesicle lesions on MRI, which can solve the problem that the work efficiency and diagnosis effect of doctors are reduced because the diagnosis data of benign prostate and seminal vesicle lesions is not automatically filled in the report in the prior art.

In order to achieve the purpose, the technical scheme of the invention is realized as follows:

in one aspect, the present invention provides a system for detecting benign lesions of prostate and seminal vesicle on MRI, comprising an AI scheduling module, a prostate dissection module, a plurality of benign lesion auxiliary diagnosis modules and a structured report module, wherein the AI scheduling module is respectively connected to the prostate dissection module, each benign lesion auxiliary diagnosis module and the structured report module, and is configured to extract header file information of a DICOM image when a patient has scanned an examination item, which is prostate multi-parameter magnetic resonance imaging (mpMRI), and search for a T2WI sequence image and a T1WI sequence image based on the header file information, and send the T2WI sequence image and the T1WI sequence image to the prostate dissection module and each benign lesion auxiliary diagnosis module; the prostate segmentation and dissection module is connected with the AI scheduling module and is used for receiving the T2WI sequence image and the T1WI sequence image, dissecting and segmenting the T2WI sequence image to obtain T2WI prostate gland segmentation data, prostate peripheral zone segmentation data, prostate transition zone segmentation data, prostate central zone segmentation data, pre-prostate fiber matrix zone segmentation data, urethra segmentation data and T2WI bilateral seminal vesicle segmentation data, dissecting and segmenting the T1WI sequence image to obtain T1WI prostate gland segmentation data and T1WI bilateral seminal vesicle segmentation data; all the segmentation data are sent to an AI scheduling module; at the moment, the AI scheduling module is also used for judging the segmentation data matched with the benign lesion auxiliary diagnosis module and sending the segmentation data to the benign lesion auxiliary diagnosis module; each benign lesion auxiliary diagnosis module is respectively connected with the AI scheduling module and is used for receiving the sequence images and the segmentation data matched with the benign lesion auxiliary diagnosis module, analyzing and processing the sequence images, outputting the diagnosis data of the lesions based on the segmentation data and feeding back the diagnosis data to the AI scheduling module; wherein, the diagnosis data are texts, numerical values and key images; and the structural report module is connected with the AI scheduling module and used for automatically generating a diagnosis result based on the diagnosis data, outputting diagnosis data of prostatic hyperplasia, diagnosis data of prostatitis, diagnosis data of prostate cyst, diagnosis data of prostatic hemorrhage, diagnosis data of seminal vesiculitis, diagnosis data of seminal vesicle atrophy and diagnosis data of seminal vesicle hemorrhage, and displaying key images corresponding to found benign lesions at corresponding positions of the structural report interface.

Preferably, when the benign lesion auxiliary diagnosis module is a prostate hyperplasia and prostatitis detection module, the AI scheduling module sends the T2WI sequence image, the prostate transition zone segmentation data and the prostate peripheral zone segmentation data to the prostate hyperplasia and prostatitis detection module, the prostate hyperplasia and prostatitis detection module removes pixels in an area outside a circumscribed rectangle where the T2WI sequence image is located to obtain a processed image, performs data standardization on the processed image and inputs the processed image to a classification network, outputs the number of classified layers of prostate hyperplasia in the T2WI sequence image based on the prostate transition zone segmentation data, outputs the number of classified layers of prostatitis in the T2WI sequence image based on the prostate peripheral zone segmentation data, and generates prostate hyperplasia data and/or prostatitis data when the number of classified layers of prostate hyperplasia and the number of classified layers of prostatitis are greater than or equal to a first preset threshold, and simultaneously outputs a layer of image with the largest area as a first key image and a second key image; when the classification layer number of the prostatic hyperplasia and the classification layer number of the prostatitis are smaller than a first preset threshold value, generating data without the prostatic hyperplasia and/or data without the prostatitis; and sending the data with prostate hyperplasia and/or the data with prostatitis, the first key image, the second key image and the data without prostate hyperplasia and/or the data without prostatitis to the AI scheduling module.

Preferably, when the benign lesion auxiliary diagnosis module is a seminal vesicle atrophy and seminal vesiculitis detection module, the AI scheduling module sends the T2WI sequence image and the T2WI double-sided seminal vesicle gland segmentation data to the seminal vesicle atrophy and seminal vesiculitis detection module, the seminal vesicle atrophy and seminal vesiculitis detection module removes pixels of an area outside a circumscribed rectangle where the T2WI sequence image is located to obtain a processed image, performs data standardization processing on the processed image and inputs the processed image to a classification network, outputs the number of classification layers of seminal vesicle atrophy and the number of classification layers of seminal vesiculitis in the T2WI sequence image based on the T2WI double-sided seminal vesicle gland segmentation data, and when the number of classification layers of seminal vesicle atrophy is greater than or equal to a second preset threshold, generates seminal vesicle atrophy data and simultaneously outputs a layer image with the largest area as a third key image; when the classification layer number of the seminal vesiculitis is greater than or equal to a first preset threshold value, generating seminal vesiculitis data, and meanwhile, outputting a layer of image with the largest area as a fourth key image; when the classification layer number of seminal vesicle atrophy is smaller than a second preset threshold value, generating seminal vesicle atrophy data, and when the classification layer number of seminal vesicle inflammation is smaller than a first preset threshold value, generating seminal vesicle inflammation data; and sending the seminal vesicle atrophy data and/or the seminal vesiculitis data, the third key image, the fourth key image and the seminal vesicle atrophy-free data and/or the seminal vesiculitis-free data to the AI scheduling module.

Preferably, when the benign lesion auxiliary diagnosis module is a prostate bleeding and seminal vesicle bleeding detection module, the AI scheduling module sends the T1WI sequence image, the T1WI prostate gland segmentation data, and the T1WI bilateral seminal vesicle gland segmentation data to the prostate bleeding and seminal vesicle bleeding detection module, the prostate bleeding and seminal vesicle bleeding detection module removes pixels of an area outside the circumscribed rectangle where the T1WI sequence image is located to obtain a processed image, performs data standardization on the processed image, inputs the processed image to the classification network, outputs the number of classification layers of prostate bleeding in the T1WI sequence image based on the T1WI prostate gland segmentation data, and generates prostate bleeding data when the number of classification layers of prostate bleeding is greater than or equal to a first preset threshold, and simultaneously outputs a layer image with the largest area as a fifth key image; outputting the classification layer number of seminal vesicle hemorrhage in the T1WI sequence image based on the T1WI bilateral seminal vesicle gland segmentation data, generating seminal vesicle hemorrhage data when the classification layer number of seminal vesicle hemorrhage is larger than or equal to a second preset threshold value, and outputting a layer image with the largest area as a sixth key image; when the classification layer number of the prostate hemorrhage is smaller than a first preset threshold value, generating non-prostate-hemorrhage data, and when the classification layer number of the seminal vesicle hemorrhage is smaller than a second preset threshold value, generating non-seminal vesicle hemorrhage data; and sending the data of bleeding with prostate and/or the data of bleeding with seminal vesicle, the fifth key image, the sixth key image and the data of bleeding without prostate and/or the data of bleeding without seminal vesicle to the AI scheduling module.

Preferably, when the benign lesion auxiliary diagnosis module is a prostate cyst detection module, the AI scheduling module sends a T2WI sequence image and T2WI prostate gland segmentation data to the prostate cyst detection module, the prostate cyst detection module segments at least one suspicious cyst region through a trained UNET segmentation network, and calculates the number of pixels of each cyst region by using a connected domain algorithm, when the number of pixels is greater than or equal to a third preset threshold, the prostate cyst data is generated, and a layer of image with the largest cyst area is output as a seventh key image; and when the number of the pixels is smaller than a third preset threshold value, generating data without the front-row cyst, and sending the data with the front-row cyst and the data without the front-row cyst to the AI scheduling module.

In another aspect, the present invention also provides a method for detecting benign lesions of prostate and seminal vesicle on MRI, comprising: when the patient is scanned and the examination item is the examination of the prostate multi-parameter magnetic resonance imaging (mpMRI), the AI scheduling module extracts the header file information of the DICOM image, searches the T2WI sequence image and the T1WI sequence image based on the header file information, and sends the T2WI sequence image and the T1WI sequence image to the prostate segmentation and dissection module and each benign lesion auxiliary diagnosis module; the prostate segmentation and dissection module receives the T2WI sequence image and the T1WI sequence image, performs dissection segmentation on the T2WI sequence image to segment out T2WI prostate gland segmentation data, prostate peripheral zone segmentation data, prostate transition zone segmentation data, prostate central zone segmentation data, pre-prostate fiber matrix zone segmentation data, urethra segmentation data and T2WI bilateral seminal vesicle segmentation data, performs dissection segmentation on the T1WI sequence image to segment out T1WI prostate gland segmentation data and T1WI bilateral seminal vesicle gland segmentation data; all the segmentation data are sent to an AI scheduling module; at the moment, the AI scheduling module is also used for judging the segmentation data matched with the benign lesion auxiliary diagnosis module and sending the segmentation data to the benign lesion auxiliary diagnosis module; each benign lesion auxiliary diagnosis module receives the sequence image and the segmentation data matched with the benign lesion auxiliary diagnosis module, analyzes and processes the sequence image, outputs diagnosis data of a lesion based on the segmentation data, and feeds the diagnosis data back to the AI scheduling module; wherein, the diagnosis data are texts, numerical values and key images; the structured report module automatically generates a diagnosis result based on the diagnosis data, outputs diagnosis data of prostatic hyperplasia, prostatitis, prostate cyst, bleeding prostate, seminal vesiculitis, seminal vesicular atrophy and seminal vesicular hemorrhage, and displays key images corresponding to found benign lesions at corresponding positions of the structured report interface.

Preferably, the method further comprises: when the benign lesion auxiliary diagnosis module is a prostate hyperplasia and prostatitis detection module, the AI scheduling module sends the T2WI sequence image, the prostate transition zone segmentation data and the prostate peripheral zone segmentation data to the prostate hyperplasia and prostatitis detection module, the prostate hyperplasia and prostatitis detection module removes pixels in an area outside an external rectangle where the T2WI sequence image is located to obtain a processed image, the processed image is subjected to data standardization processing and then is input to a classification network, the classification layer number of the prostate hyperplasia in the T2WI sequence image is output based on the prostate transition zone segmentation data, the classification layer number of the prostatitis in the T2WI sequence image is output based on the prostate peripheral zone segmentation data, when the classification layer number of the current column hyperplasia and the classification layer number of the prostatitis are greater than or equal to a first preset threshold value, prostate hyperplasia data and/or prostatitis data are generated, and the layer image with the largest area is output as a first key image and a second key image; when the classification layer number of the prostatic hyperplasia and the classification layer number of the prostatitis are smaller than a first preset threshold value, generating data without the prostatic hyperplasia and/or data without the prostatitis; and sending the data with prostate hyperplasia and/or the data with prostatitis, the first key image, the second key image and the data without prostate hyperplasia and/or the data without prostatitis to the AI scheduling module.

Preferably, the method further comprises: when the benign lesion auxiliary diagnosis module is a seminal vesicle atrophy and seminal vesiculitis detection module, the AI scheduling module sends the T2WI sequence image and the T2WI bilateral seminal vesicle gland segmentation data to the seminal vesicle atrophy and seminal vesiculitis detection module, the seminal vesicle atrophy and seminal vesiculitis detection module removes pixels of an area outside a circumscribed rectangle where the T2WI sequence image is located to obtain a processed image, performs data standardization processing on the processed image and inputs the processed image into a classification network, outputs the classification layer number of seminal vesicle atrophy and the classification layer number of seminal vesiculitis in the T2WI sequence image based on the T2WI bilateral seminal vesicle gland segmentation data, and generates seminal vesicle atrophy data when the classification layer number of seminal vesicle atrophy is larger than or equal to a second preset threshold value and simultaneously outputs a layer image with the largest area as a third key image; when the classification layer number of the seminal vesiculitis is greater than or equal to a first preset threshold value, generating seminal vesiculitis data, and meanwhile, outputting a layer of image with the largest area as a fourth key image; when the classification layer number of seminal vesicle atrophy is smaller than a second preset threshold value, generating seminal vesicle atrophy data, and when the classification layer number of seminal vesicle inflammation is smaller than a first preset threshold value, generating seminal vesicle inflammation data; and sending the seminal vesicle atrophy data and/or the seminal vesiculitis data, the third key image, the fourth key image and the seminal vesicle atrophy-free data and/or the seminal vesiculitis-free data to the AI scheduling module.

Preferably, the method further comprises: when the benign lesion auxiliary diagnosis module is a prostate bleeding and seminal vesicle bleeding detection module, the AI scheduling module sends the T1WI sequence image, the T1WI prostate gland segmentation data and the T1WI bilateral seminal vesicle gland segmentation data to the prostate bleeding and seminal vesicle bleeding detection module, the prostate bleeding and seminal vesicle bleeding detection module removes pixels of an area outside a circumscribed rectangle where the T1WI sequence image is located to obtain a processed image, performs data standardization processing on the processed image and inputs the processed image into a classification network, outputs the classification layer number of the prostate bleeding in the T1WI sequence image based on the T1WI prostate gland segmentation data, and generates prostate bleeding data and simultaneously outputs a layer image with the largest area as a fifth key image when the classification layer number of the current prostate bleeding is greater than or equal to a first preset threshold value; outputting the classification layer number of seminal vesicle hemorrhage in the T1WI sequence image based on the T1WI bilateral seminal vesicle gland segmentation data, generating seminal vesicle hemorrhage data when the classification layer number of seminal vesicle hemorrhage is larger than or equal to a second preset threshold value, and outputting a layer image with the largest area as a sixth key image; when the classification layer number of the prostate hemorrhage is smaller than a first preset threshold value, generating non-prostate-hemorrhage data, and when the classification layer number of the seminal vesicle hemorrhage is smaller than a second preset threshold value, generating non-seminal vesicle hemorrhage data; and sending the data of bleeding with prostate and/or the data of bleeding with seminal vesicle, the fifth key image, the sixth key image and the data of bleeding without prostate and/or the data of bleeding without seminal vesicle to the AI scheduling module.

Preferably, the method further comprises: when the benign lesion auxiliary diagnosis module is a prostate cyst detection module, the AI scheduling module sends a T2WI sequence image and T2WI prostate gland segmentation data to the prostate cyst detection module, the prostate cyst detection module segments at least one suspicious cyst region through a trained UNET segmentation network, and calculates the number of pixels of each cyst region by using a connected domain algorithm, when the number of pixels is greater than or equal to a third preset threshold value, prostate cyst data are generated, and a layer of image with the largest cyst area is output as a seventh key image; and when the number of the pixels is smaller than a third preset threshold value, generating data without the front-row cyst, and sending the data with the front-row cyst and the data without the front-row cyst to the AI scheduling module.

The invention has the technical effects that:

1. because the invention is provided with the AI scheduling module, the prostate dissection module, the plurality of benign lesion auxiliary diagnosis modules and the structuralized report module, T2WI prostate gland dissection data, prostate peripheral zone dissection data, prostate transition zone dissection data, prostate central zone dissection data, pre-prostatic fiber stroma zone dissection data, urethra dissection data and T2WI bilateral seminal vesicle gland dissection data can be segmented on the T2WI sequence image, T1WI prostate gland dissection data and T1WI bilateral seminal vesicle gland dissection data are segmented on the T1WI sequence image, each prostate benign lesion auxiliary diagnosis module can diagnose benign lesions based on the dissection data of each partition of the prostate, the diagnosis data of the prostatic hyperplasia, the diagnosis data of the prostatitis, the diagnosis data of the prostate cyst, the diagnosis data of the prostate bleeding, the diagnosis data of the seminal vesiculitis, the diagnosis data of the seminal vesicular atrophy and the diagnosis data of the seminal vesicular bleeding are respectively output, the data of the benign lesions are displayed in a report, the interference of various common benign lesions in the diagnosis process of a doctor is avoided, and the benign lesions are clearly reported, so that on one hand, the clinical symptoms can be explained, on the other hand, the differential diagnosis capability of the prostate cancer can be improved, the image background information is provided for the detection of the prostate cancer foci, and the image background information is automatically displayed in the report, thereby saving the diagnosis time of the doctor and improving the working efficiency;

2. because the invention is provided with the prostate hyperplasia and prostatitis detecting module, the diagnosis data of the prostate hyperplasia and/or the prostatitis can be automatically output to the report, the clinical symptoms of the patient can be explained, and the diagnosis data can be distinguished from the clinically significant carcinoma of the prostate;

3. because the seminal vesicle atrophy and seminal vesiculitis detection module is arranged, the diagnosis data of the seminal vesicle atrophy and/or the seminal vesiculitis can be automatically output to the report, and the correct diagnosis of the image examination is also significant, and because the seminal vesicle inflammation and atrophy of the old can cause the reduction of the seminal vesicle gland volume and the unclear display of the internal structure, the diagnosis data of the seminal vesicle atrophy and/or the seminal vesiculitis can be automatically output to the report, and the diagnosis data can be identified from the seminal vesicle invaded by the prostate cancer;

4. because the invention is provided with the prostate bleeding and seminal vesicle bleeding detection module, diagnosis data of prostate bleeding and/or seminal vesicle bleeding can be automatically output to a report, clinical symptoms (such as hemospermia) can be explained, and diagnosis after puncture is facilitated;

5. because the invention is provided with the prostate cyst detection module, the diagnosis data of the prostate cyst can be automatically output to a report and used as a reference standard for positioning when in puncture biopsy.

Drawings

The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the invention without limiting the invention. In the drawings:

figure 1 shows a schematic diagram of a system for detecting benign lesions of prostate and seminal vesicle on MRI according to a first embodiment of the present invention;

FIG. 2 is a diagram of the structural report interface displaying the data of the prostate hyperplasia detected by the prostate hyperplasia and prostatitis detecting module in the system for detecting benign prostatic diseases and seminal vesicle diseases on MRI according to the second embodiment of the present invention;

FIG. 3 is a diagram of the structural report interface displaying the data of prostate hyperplasia and prostatitis detected by the prostate and seminal vesicle detecting module in the system for detecting benign prostatic diseases on MRI according to the second embodiment of the present invention;

fig. 4 is a schematic diagram of the seminal vesicle atrophy and seminal vesicle inflammation detection module with seminal vesicle atrophy data displayed on the structured report interface in the system for detecting benign lesions of prostate and seminal vesicle on MRI according to the third embodiment of the present invention;

FIG. 5 is a schematic diagram of the seminal vesicle atrophy and seminal vesiculitis detection module detected seminal vesiculitis data displayed on a structured report interface in a system for detecting benign lesions of prostate and seminal vesicular glands on MRI according to the third embodiment of the present invention;

fig. 6 is a diagram illustrating the display of the bleeding prostate data on the structured report interface detected by the bleeding prostate and seminal vesicle detection module in the system for detecting benign lesions of prostate and seminal vesicle on MRI according to the fourth embodiment of the present invention;

figure 7 shows a schematic diagram of a structured report interface displaying data of a prostate bleeding and a seminal vesicle bleeding detected by a detection module for prostate bleeding and seminal vesicle bleeding in a system for detecting benign lesions of prostate and seminal vesicle on MRI according to a fourth embodiment of the present invention;

fig. 8 is a diagram illustrating the display of data for a prostate cyst detected by the prostate cyst detection module on a structured report interface in a system for detecting benign lesions of the prostate and seminal vesicle on MRI according to fifth embodiment of the present invention;

FIG. 9 shows a flowchart of a method for detecting benign lesions of the prostate and seminal vesicle on MRI according to a sixth embodiment of the present invention;

FIG. 10 is a diagram of the structural report interface displaying the data of the prostate hyperplasia detected by the prostate hyperplasia and prostatitis detecting module in the method for detecting benign prostate and seminal vesicle diseases on MRI according to the sixth embodiment of the present invention;

FIG. 11 is a diagram of the structural report interface displaying the data of prostatitis detected by the prostate hyperplasia and prostatitis detecting module in the sixth embodiment of the invention;

fig. 12 is a schematic diagram of the seminal vesicle atrophy and seminal vesicle inflammation detection module with seminal vesicle atrophy data displayed on the structured report interface in the method for detecting benign prostatic and seminal vesicle lesions on MRI according to the sixth embodiment of the invention;

figure 13 shows a schematic diagram of the seminal vesicle atrophy and seminal vesiculitis detection module with seminal vesiculitis data displayed on the structured report interface in the method for detecting benign lesions of prostate and seminal vesicle on MRI according to the sixth embodiment of the invention;

fig. 14 is a diagram illustrating the display of the bleeding prostate data on the structured report interface detected by the bleeding prostate and seminal vesicle detection module in the method for detecting benign lesions of prostate and seminal vesicle on MRI according to the sixth embodiment of the present invention;

fig. 15 is a diagram illustrating the display of seminal vesicle hemorrhage data detected by the prostate hemorrhage and seminal vesicle hemorrhage detecting module on the structured report interface in the method for detecting benign prostatic and seminal vesicle lesions on MRI according to the sixth embodiment of the invention;

fig. 16 is a diagram illustrating the data of prostate cyst detected by the prostate cyst detection module displayed on the structured report interface in the method for detecting benign lesions of prostate and seminal vesicle on MRI according to the sixth embodiment of the present invention.

Detailed Description

The present invention will be described in detail below with reference to the accompanying drawings in conjunction with embodiments.

Example one

Figure 1 shows a schematic diagram of a system for detecting benign lesions of prostate and seminal vesicle on MRI according to a first embodiment of the present invention; as shown in fig. 1, the system includes: AI scheduling module 10, prostate anatomy segmentation module 20, a plurality of benign lesion auxiliary diagnostic modules 30, and a structured reporting module 40, wherein,

an AI scheduling module 10, respectively connected to the prostate segmentation and dissection module 20, each benign lesion auxiliary diagnosis module 30 and the structured report module 40, for extracting header file information of the DICOM image when the patient has scanned the examination item as the examination of the prostate multi-parameter magnetic resonance imaging (mpMRI), searching the T2WI sequence image and the T1WI sequence image based on the header file information, and sending the T2WI sequence image and the T1WI sequence image to the prostate segmentation and dissection module 20 and each benign lesion auxiliary diagnosis module 30;

a prostate segmentation and dissection module 20, connected to the AI scheduling module 10, for receiving the T2WI sequence image and the T1WI sequence image, performing segmentation and dissection on the T2WI sequence image to obtain T2WI prostate gland segmentation data, prostate peripheral zone segmentation data, prostate transition zone segmentation data, prostate central zone segmentation data, pre-prostate fibrous stroma zone segmentation data, urethra segmentation data, and T2WI bilateral seminal vesicle segmentation data, performing segmentation and dissection on the T1WI sequence image to obtain T1WI prostate gland segmentation data and T1WI bilateral seminal vesicle segmentation data; and sends all the above-mentioned segmentation data to the AI scheduling module 10; at this time, the process of the present invention,

the AI scheduling module 10 is further configured to determine segmentation data matched with the benign lesion auxiliary diagnosis module 30, and send the segmentation data to the benign lesion auxiliary diagnosis module 30;

the clinician suspects the possibility of prostate cancer at the time of the symptomatic urological visit of the patient, applying for an mpMRI examination, examination item: mpMRI, pelvic floor scan or pan-luminal floor scan; type of patient: before outpatient and inpatient puncture; the scan range is local to the prostate.

The prostate anatomy segmentation module segments the T2WI sequence images and the T1WI sequence images.

Each benign lesion auxiliary diagnosis module 30 is connected to the AI scheduling module 10, and is configured to receive the sequence image and the segmentation data matched therewith, analyze and process the sequence image, output diagnosis data of a lesion based on the segmentation data, and feed back the diagnosis data to the AI scheduling module 10; wherein, the diagnosis data are texts, numerical values and key images;

and the structural report module 40 is connected with the AI scheduling module 10 and is used for automatically generating a diagnosis result based on the diagnosis data, outputting diagnosis data of prostatic hyperplasia, prostatitis, prostatocystis, prostahemorrhage, seminal vesiculitis, seminal vesicle atrophy and seminal vesicle hemorrhage, and displaying key images corresponding to found benign lesions at corresponding positions of the structural report interface.

When there is no prostatic hyperplasia, prostatitis, prostate cyst, prostate hemorrhage, seminal vesiculitis, seminal vesicle atrophy and seminal vesicle hemorrhage, the control of the relevant benign lesion of the structured report interface is not activated, and the key image is not returned; and activating a control of the relevant benign lesion of the structured report interface and returning a key image when prostate hyperplasia, prostatitis, prostate cyst, prostate hemorrhage, seminal vesiculitis, seminal vesicle atrophy and seminal vesicle hemorrhage exist.

The embodiment of the invention is provided with an AI scheduling module, a prostate dissection module, a plurality of benign lesion auxiliary diagnosis modules and a structured report module, which can divide T2WI prostate gland dissection data, prostate peripheral zone dissection data, prostate transition zone dissection data, prostate central zone dissection data, pre-prostate fiber matrix zone dissection data, urethra dissection data and T2WI bilateral seminal vesicle segmentation data on a T2WI sequence image, divide T1WI prostate gland dissection data and T1WI bilateral seminal vesicle gland dissection data on the T1WI sequence image, each prostate benign lesion auxiliary diagnosis module can diagnose benign lesions based on the dissection data of each partition of the prostate, respectively output diagnosis data of prostatic hyperplasia, diagnosis data of prostatitis, diagnosis data of prostate gland hemorrhage, diagnosis data of seminal vesiculitis, diagnosis data of seminal vesicle atrophy and diagnosis data of seminal vesicle hemorrhage, display the benign lesion data in a report, avoid the interference of various common lesions in the lesion diagnosis process of a doctor, can improve the diagnosis efficiency of benign prostate cancer diagnosis, and can improve the diagnosis efficiency of the diagnosis of benign cancer diagnosis.

Example two

When the benign lesion auxiliary diagnosis module 30 is a prostate hyperplasia and prostatitis detection module, the AI scheduling module 10 sends the T2WI sequence image, the prostate transition zone segmentation data and the prostate peripheral zone segmentation data to the prostate hyperplasia and prostatitis detection module, the prostate hyperplasia and prostatitis detection module removes pixels in an area outside a circumscribed rectangle where the T2WI sequence image is located to obtain a processed image, performs data standardization on the processed image and inputs the processed image into a classification network, outputs the classification layer number of prostate hyperplasia in the T2WI sequence image based on the prostate transition zone segmentation data, outputs the classification layer number of prostatitis in the T2WI sequence image based on the prostate peripheral zone segmentation data, and generates prostate hyperplasia data and/or prostatitis data when the classification layer number of prostate hyperplasia and the classification layer number of prostatitis are greater than or equal to a first preset threshold value, and outputs a layer image with the largest area as a first key image and a second key image; when the classification layer number of the prostatic hyperplasia and the classification layer number of the prostatitis are smaller than a first preset threshold value, generating data without the prostatic hyperplasia and/or data without the prostatitis; and sends the data with prostate hyperplasia and/or the data with prostatitis, the first key image, the second key image and the data without prostate hyperplasia and/or the data without prostatitis to the AI scheduling module 10.

The AI scheduling module 10 sends the data with prostate hyperplasia and/or the data with prostatitis, the first key image, the second key image, and the data without prostate hyperplasia and/or the data without prostatitis to the structured reporting module 40.

The first preset threshold is generally 3, when the number of classified layers of prostate hyperplasia and the number of classified layers of prostatitis are greater than or equal to 3, prostate hyperplasia data and/or prostatitis data are generated, and a layer of image with the largest area is output as a first key image and a second key image.

Fig. 2 is a schematic diagram illustrating that the prostate hyperplasia and prostatitis detection module in the system for detecting benign prostate and seminal vesicle lesions on MRI according to the second embodiment of the present invention detects that there is prostate hyperplasia data displayed on the structured report interface, and as shown in fig. 2, the structured report interface marks a square symbol in front of a BPH (benign prostate hyperplasia) control, and automatically activates the control, so as to display the first key image on the structured report interface.

FIG. 3 is a diagram of a structured report interface displaying data of prostate hyperplasia and prostatitis detected by the module in the system for detecting benign prostatic and seminal vesicle lesions on MRI according to the second embodiment of the present invention; as shown in fig. 3, the structured reporting interface places a "√" in front of the prostatitis control, which is automatically activated, and a second key image is displayed in the structured reporting interface.

The embodiment of the invention is provided with the prostate hyperplasia and prostatitis detection module, can automatically output the diagnosis data of the prostate hyperplasia and/or the prostatitis to the report, can explain the clinical symptoms of the patient, and can distinguish the clinical significant carcinoma of the prostate.

EXAMPLE III

When the benign lesion auxiliary diagnosis module 30 is a seminal vesicle atrophy and seminal vesiculitis detection module, the AI scheduling module 10 sends the T2WI sequence image and the T2WI double-sided seminal vesicle gland segmentation data to the seminal vesicle atrophy and seminal vesiculitis detection module, the seminal vesicle atrophy and seminal vesiculitis detection module removes the region pixels outside the circumscribed rectangle where the T2WI sequence image is located to obtain a processed image, and inputs the processed image into a classification network after data standardization processing, outputs the classification layer number of seminal vesicle atrophy and the classification layer number of seminal vesiculitis in the T2WI sequence image based on the T2WI double-sided seminal vesicle gland segmentation data, when the classification layer number of seminal vesicle atrophy is greater than or equal to a second preset threshold value, the seminal vesicle atrophy data is generated, and simultaneously outputs a layer image with the largest area as a third key image; when the classification layer number of the seminal vesiculitis is greater than or equal to a first preset threshold value, generating seminal vesiculitis data, and meanwhile, outputting a layer of image with the largest area as a fourth key image; when the classification layer number of seminal vesicle atrophy is smaller than a second preset threshold value, generating seminal vesicle atrophy data, and when the classification layer number of seminal vesicle inflammation is smaller than a first preset threshold value, generating seminal vesicle inflammation data; and sends the seminal vesicle atrophy data and/or seminal vesiculitis data, the third key image, the fourth key image and seminal vesicle atrophy-free data and/or seminal vesiculitis-free data to the AI scheduling module 10.

AI dispatch module 10 sends seminal vesicle atrophy data and/or seminal vesiculitis data, third key image, fourth key image, and seminal vesicle atrophy-free data and/or seminal vesiculitis-free data to structured reporting module 40.

The second preset threshold value is generally 1, when the classification layer number of the seminal vesicle atrophy is more than or equal to 1, seminal vesicle atrophy data are generated, and meanwhile, a layer of image with the largest area is output to serve as a third key image; the first preset threshold is generally 3, when the classification layer number of the seminal vesiculitis is greater than or equal to 3, seminal vesiculitis data is generated, and meanwhile, a layer of image with the largest area is output as a fourth key image.

Fig. 4 shows a schematic diagram of the seminal vesicle atrophy and seminal vesiculitis detection module in the system for detecting benign lesions of prostate and seminal vesicle on MRI according to the third embodiment of the present invention, which detects the display of seminal vesicle atrophy data on the structured report interface, as shown in fig. 4, the structured report interface marks a square in front of the seminal vesicle atrophy control, automatically activates the control, and displays the third key image on the structured report interface.

FIG. 5 is a schematic diagram of the seminal vesicle atrophy and seminal vesiculitis detection module detected seminal vesiculitis data displayed on a structured report interface in a system for detecting benign lesions of prostate and seminal vesicular glands on MRI according to the third embodiment of the present invention; as shown in fig. 5, the structured reporting interface places a "√" in front of a seminal vesiculitis control, which is automatically activated, and a fourth key image is displayed in the structured reporting interface.

The embodiment of the invention is provided with the seminal vesicle atrophy and seminal vesiculitis detection module, can automatically output seminal vesicle atrophy andor seminal vesiculitis diagnosis data to a report, has significance for image examination to make correct diagnosis, and can distinguish prostate cancer invading seminal vesicle because seminal vesicle atrophy andor seminal vesiculitis diagnosis data of old people are automatically output to the report because seminal vesicle inflammation and atrophy can cause the reduction of seminal vesicle gland volume and unclear internal structure display.

Example four

When the benign lesion auxiliary diagnosis module 30 is a prostate bleeding and seminal vesicle bleeding detection module, the AI scheduling module 10 sends the T1WI sequence image, the T1WI prostate gland segmentation data and the T1WI bilateral seminal vesicle gland segmentation data to the prostate bleeding and seminal vesicle bleeding detection module, the prostate bleeding and seminal vesicle bleeding detection module removes pixels in an area outside an external rectangle where the T1WI sequence image is located to obtain a processed image, performs data standardization on the processed image, inputs the processed image to a classification network, outputs a classification layer number of the prostate bleeding in the T1WI sequence image based on the T1WI prostate gland segmentation data, and generates the prostate bleeding data when the classification layer number of the current prostate bleeding is greater than or equal to a first preset threshold value, and simultaneously outputs a layer image with the largest area as a fifth key image; outputting the number of classification layers of seminal vesicle bleeding in the T1WI sequence image based on the T1WI bilateral seminal vesicle gland segmentation data, generating seminal vesicle bleeding data when the number of classification layers of seminal vesicle bleeding is larger than or equal to a second preset threshold value, and outputting a layer of image with the largest area as a sixth key image; when the classification layer number of the prostate hemorrhage is smaller than a first preset threshold value, generating non-prostate-hemorrhage data, and when the classification layer number of the seminal vesicle hemorrhage is smaller than a second preset threshold value, generating non-seminal vesicle hemorrhage data; and sends the data of bleeding with prostate and/or the data of bleeding with seminal vesicle, the fifth key image, the sixth key image, and the data of bleeding without prostate and/or the data of bleeding without seminal vesicle to the AI scheduling module 10.

AI scheduling module 10 sends the prostate bleeding data and/or seminal vesicle bleeding data, the fifth key image, the sixth key image, and prostate bleeding free data and/or seminal vesicle bleeding free data to structured reporting module 40.

Wherein, the first preset threshold is generally 3, when the classification layer number of the prostate bleeding is more than or equal to 3, the prostate bleeding data is generated, and meanwhile, the layer image with the largest area is output as a fifth key image; the second preset threshold is generally 1, when the number of classification layers of seminal vesicle hemorrhage is greater than or equal to 1, seminal vesicle hemorrhage data is generated, and meanwhile, a layer of image with the largest area is output as a sixth key image.

Figure 6 shows a schematic diagram of a structured report interface displaying the data of the detected prostate bleeding and seminal vesicle bleeding in the system for detecting benign lesions of prostate and seminal vesicle on MRI according to the fourth embodiment of the invention; as shown in fig. 6, the structured reporting interface marks "√" in front of the prostate bleeding control, which is automatically activated, a fifth key image is displayed in the structured reporting interface.

Fig. 7 is a diagram illustrating the display of the data of the bleeding from the prostate and the seminal vesicle detected by the bleeding from the prostate and seminal vesicle detection module on the structured report interface in the system for detecting benign lesions of prostate and seminal vesicle on MRI according to the fourth embodiment of the present invention; as shown in fig. 7, the structured reporting interface places a "√" in front of a seminal vesicle hemorrhage control, which is automatically activated, and a sixth key image is displayed in the structured reporting interface.

The embodiment of the invention is provided with a prostate bleeding and seminal vesicle bleeding detection module, can automatically output diagnosis data of prostate bleeding and/or seminal vesicle bleeding to a report, can explain clinical symptoms (such as hemospermia), is beneficial to diagnosis after puncture, generally considers that the probability of prostate cancer in an obvious and continuous bleeding area is very low, is an important indirect symptom, and excludes the probability of prostate cancer in the area.

EXAMPLE five

When the benign lesion auxiliary diagnosis module 30 is a prostate cyst detection module, the AI scheduling module 10 sends a T2WI sequence image and T2WI prostate gland segmentation data to the prostate cyst detection module, the prostate cyst detection module segments at least one suspicious cyst region through a trained UNET segmentation network, and calculates the number of pixels of each cyst region by using a connected domain algorithm, when the number of pixels is greater than or equal to a third preset threshold, prostate cyst data is generated, and a layer of image with the largest cyst area is output as a seventh key image; when the number of pixels is smaller than a third preset threshold, no front cyst data is generated, and the data with and without front cysts are sent to the AI scheduler module 10.

The AI dispatch module 10 sends the data with prostate cyst, the seventh key image, and the data without prostate cyst to the structured report module 40.

The third preset threshold is generally 30, when the number of pixels in the front row cyst area is greater than or equal to 30, the front row cyst data is generated, and the layer image with the largest area is output as the seventh key image.

Fig. 8 is a diagram illustrating the fact that the prostate cyst detection module detects that there is data about prostate cysts displayed on the structured report interface in the system for detecting benign lesions of prostate and seminal vesicle on MRI according to the fifth embodiment of the present invention, and as shown in fig. 8, the structured report interface marks a square symbol in front of the prostate cyst control, automatically activates the control, and displays the seventh key image on the structured report interface.

The embodiment of the invention is provided with a prostate cyst detection module, and diagnosis data of prostate cyst can be automatically output to a report and used as a reference standard for positioning when in needle biopsy.

Example six

Figure 9 shows a flow chart of a method for detecting benign lesions of the prostate and seminal vesicle on MRI according to a sixth embodiment of the invention; as shown in fig. 9, the method includes the steps of:

step S601, when the patient finishes the scanning of the examination item, namely the examination of the prostate multi-parameter magnetic resonance imaging (mpMRI), the AI scheduling module extracts the header file information of the DICOM image, searches the T2WI sequence image and the T1WI sequence image based on the header file information, and sends the T2WI sequence image and the T1WI sequence image to the prostate segmentation anatomical module and each benign lesion auxiliary diagnostic module;

step S602, a prostate segmentation and dissection module receives the T2WI sequence image and the T1WI sequence image, dissects and dissects the T2WI sequence image to segment T2WI prostate gland segmentation data, prostate peripheral zone segmentation data, prostate transition zone segmentation data, prostate central zone segmentation data, pre-prostatic fibrous stroma zone segmentation data, urethra segmentation data and T2WI bilateral seminal vesicle gland segmentation data, dissects and dissects the T1WI sequence image to segment T1WI prostate gland segmentation data and T1WI bilateral seminal vesicle gland segmentation data; all the segmentation data are sent to an AI scheduling module; at the moment, the AI scheduling module is also used for judging the segmentation data matched with the benign lesion auxiliary diagnosis module and sending the segmentation data to the benign lesion auxiliary diagnosis module;

the clinician suspects the possibility of prostate cancer at the time of the symptomatic urological visit of the patient, applying for an mpMRI examination, examination item: mpMRI, pelvic floor scan or pan-luminal floor scan; patient type: before outpatient and inpatient puncture; the scan range is local to the prostate.

The prostate anatomy segmentation module segments a T2WI sequence image and a T1WI sequence image.

Step S603, each benign lesion auxiliary diagnosis module receives the sequence image and the segmentation data matched with the benign lesion auxiliary diagnosis module, analyzes and processes the sequence image, outputs diagnosis data of a lesion based on the segmentation data, and feeds back the diagnosis data to the AI scheduling module; wherein, the diagnosis data are texts, numerical values and key images;

step S604, the structured report module automatically generates a diagnosis result based on the diagnosis data, outputs diagnosis data of prostatic hyperplasia, prostatitis, cyst of prostate, bleeding of prostate, seminal vesiculitis, seminal vesicular atrophy and seminal vesicular hemorrhage, and displays key images corresponding to found benign lesions at corresponding positions of the structured report interface;

when no prostatic hyperplasia, prostatitis, prostatic cyst, prostate hemorrhage, seminal vesiculitis, seminal vesicle atrophy and seminal vesicle hemorrhage exist, the control of the relevant benign lesion of the structured report interface is not activated, and the key image is not returned; when prostate hyperplasia, prostatitis, prostate cyst, prostate hemorrhage, seminal vesiculitis, seminal vesicle atrophy and seminal vesicle hemorrhage exist, the control of the relevant benign lesion of the structured report interface is activated, and the key image is returned.

Wherein, the method also comprises: when the benign lesion auxiliary diagnosis module is a prostate hyperplasia and prostatitis detection module, the AI scheduling module sends the T2WI sequence image, the prostate transition zone segmentation data and the prostate peripheral zone segmentation data to the prostate hyperplasia and prostatitis detection module, the prostate hyperplasia and prostatitis detection module removes pixels of an area outside a circumscribed rectangle where the T2WI sequence image is located to obtain a processed image, the processed image is subjected to data standardization processing and then is input to a classification network, the classification layer number of the prostate hyperplasia in the T2WI sequence image is output based on the prostate transition zone segmentation data, classification of the prostatitis layer number in the T2WI sequence image is output based on the prostate peripheral zone segmentation data, when the classification layer number of the current prostate hyperplasia and the classification layer number of the prostatitis are larger than or equal to a first preset threshold value, prostate hyperplasia data and/or prostatitis data are generated, and a layer of image with the largest area is output as a first key image and a second key image; when the classification layer number of the prostatic hyperplasia and the classification layer number of the prostatitis are smaller than a first preset threshold value, generating data without the prostatic hyperplasia and/or data without the prostatitis; and sending the data with prostate hyperplasia and/or the data with prostatitis, the first key image, the second key image and the data without prostate hyperplasia and/or the data without prostatitis to the AI scheduling module.

The AI scheduling module sends the data with prostate hyperplasia and/or the data with prostatitis, the first key image, the second key image and the data without prostate hyperplasia and/or the data without prostatitis to the structured report module.

Fig. 10 is a schematic diagram illustrating that the prostate hyperplasia and prostatitis detection module detects that there is prostate hyperplasia data displayed on the structured report interface in the method for detecting benign lesions of prostate and seminal vesicle on MRI according to the sixth embodiment of the present invention, and as shown in fig. 10, the structured report interface marks a square symbol in front of a BPH (benign prostatic hyperplasia) control, and automatically activates the control, so as to display the first key image on the structured report interface.

FIG. 11 is a diagram of the structural report interface displaying the data of prostatitis detected by the prostate hyperplasia and prostatitis detecting module in the sixth embodiment of the invention; as shown in fig. 11, the structured reporting interface was "check" in front of the prostatitis control, which was automatically activated, displaying the second key image on the structured reporting interface. Wherein, the method also comprises: and a navigation map generating unit in the structured report module automatically marks the position of a cancer focus on the navigation map based on the processed data, and displays the marked navigation map at a corresponding position of the structured report interface.

Wherein, the method also comprises: when the benign lesion auxiliary diagnosis module is a seminal vesicle atrophy and seminal vesiculitis detection module, the AI scheduling module sends the T2WI sequence image and the T2WI double-side seminal vesicle gland segmentation data to the seminal vesicle atrophy and seminal vesiculitis detection module, the seminal vesicle atrophy and seminal vesiculitis detection module removes pixels in an area outside a circumscribed rectangle where the T2WI sequence image is located to obtain a processed image, performs data standardization on the processed image and inputs the processed image into a classification network, outputs the classification layer number of seminal vesicle atrophy and the classification layer number of seminal vesiculitis in the T2WI sequence image based on the T2WI double-side seminal vesicle gland segmentation data, and generates seminal vesicle atrophy data when the classification of the seminal vesicle atrophy layer number is greater than or equal to a second preset threshold value and simultaneously outputs a layer image with the largest area as a third key image; when the number of classification layers of the seminal vesiculitis is larger than or equal to a first preset threshold value, seminal vesiculitis data are generated, and meanwhile, a layer of image with the largest area is output to serve as a fourth key image; when the classification layer number of seminal vesicle atrophy is smaller than a second preset threshold value, generating seminal vesicle atrophy data, and when the classification layer number of seminal vesicle inflammation is smaller than a first preset threshold value, generating seminal vesicle inflammation data; and sending the seminal vesicle atrophy data and/or the seminal vesiculitis data, the third key image, the fourth key image and the seminal vesicle atrophy-free data and/or the seminal vesiculitis-free data to the AI scheduling module.

The AI scheduling module sends the seminal vesicle atrophy data and/or the seminal vesiculitis data, the third key image, the fourth key image and the seminal vesicle atrophy-free data and/or the seminal vesiculitis-free data to the structured reporting module.

Fig. 12 is a schematic diagram of the seminal vesicle atrophy and seminal vesicle inflammation detection module for detecting seminal vesicle atrophy data displayed on the structured report interface in the method for detecting benign prostatic and seminal vesicle lesions on MRI according to the third embodiment of the invention, as shown in fig. 12, the structured report interface marks a square symbol in front of the seminal vesicle atrophy control, automatically activates the control, and displays the third key image on the structured report interface.

Fig. 13 is a schematic diagram of seminal vesicle atrophy and seminal vesiculitis detection data displayed on a structured report interface by the seminal vesicle atrophy and seminal vesiculitis detection module in the method for detecting benign prostatic and seminal vesicle lesions on MRI according to the third embodiment of the invention; as shown in fig. 13, the structured reporting interface "v" in front of the seminal vesiculitis control, which is automatically activated, and a fourth key image is displayed on the structured reporting interface.

Wherein, the method also comprises: when the benign lesion auxiliary diagnosis module is a prostate bleeding and seminal vesicle bleeding detection module, the AI scheduling module sends the T1WI sequence image, the T1WI prostate gland segmentation data and the T1WI bilateral seminal vesicle gland segmentation data to the prostate bleeding and seminal vesicle bleeding detection module, the prostate bleeding and seminal vesicle bleeding detection module removes pixels of an area outside a circumscribed rectangle where the T1WI sequence image is located to obtain a processed image, performs data standardization processing on the processed image and inputs the processed image into a classification network, outputs the classification layer number of the prostate bleeding in the T1WI sequence image based on the T1WI prostate gland segmentation data, and generates prostate bleeding data and simultaneously outputs a layer image with the largest area as a fifth key image when the classification layer number of the current prostate bleeding is greater than or equal to a first preset threshold value; outputting the classification layer number of seminal vesicle hemorrhage in the T1WI sequence image based on the T1WI bilateral seminal vesicle gland segmentation data, generating seminal vesicle hemorrhage data when the classification layer number of seminal vesicle hemorrhage is larger than or equal to a second preset threshold value, and outputting a layer image with the largest area as a sixth key image; when the classification layer number of the prostate hemorrhage is smaller than a first preset threshold value, generating non-prostate-hemorrhage data, and when the classification layer number of the seminal vesicle hemorrhage is smaller than a second preset threshold value, generating non-seminal vesicle hemorrhage data; and sending the data of the prostate bleeding and/or the data of the seminal vesicle bleeding, the fifth key image, the sixth key image and the data of the prostate bleeding and/or the data of the seminal vesicle bleeding to the AI scheduling module.

The AI scheduling module sends the data of the prostate bleeding and/or the data of the seminal vesicle bleeding, the fifth key image, the sixth key image and the data of the prostate bleeding and/or the data of the seminal vesicle bleeding to the structured report module.

The first preset threshold is generally 3, when the number of classification layers of prostate bleeding is greater than or equal to 3, prostate bleeding data are generated, and a layer of image with the largest area is output as a fifth key image; the second preset threshold is generally 1, when the number of classification layers of seminal vesicle hemorrhage is greater than or equal to 1, seminal vesicle hemorrhage data is generated, and meanwhile, a layer of image with the largest area is output as a sixth key image.

Fig. 14 is a diagram illustrating the display of the bleeding prostate data on the structured report interface detected by the bleeding prostate and seminal vesicle detection module in the method for detecting benign lesions of prostate and seminal vesicle on MRI according to the fourth embodiment of the present invention; as shown in fig. 14, the structured reporting interface places a "√" front of a prostate bleeding control, which is automatically activated, and a fifth key image is displayed in the structured reporting interface.

Fig. 15 is a diagram illustrating the display of the seminal vesicle hemorrhage data detected by the prostate hemorrhage and seminal vesicle hemorrhage detecting module on the structured report interface in the method for detecting benign prostatic and seminal vesicle lesions on MRI according to the fourth embodiment of the invention; as shown in fig. 15, the structured reporting interface places a "√" in front of a seminal vesicle hemorrhage control, which is automatically activated, and a sixth key image is displayed in the structured reporting interface.

Wherein, the method also comprises: when the benign lesion auxiliary diagnosis module is a prostate cyst detection module, the AI scheduling module sends a T2WI sequence image and T2WI prostate gland segmentation data to the prostate cyst detection module, the prostate cyst detection module segments at least one suspicious cyst region through a trained UNET segmentation network, and calculates the number of pixels of each cyst region by using a connected domain algorithm, when the number of pixels is greater than or equal to a third preset threshold, prostate cyst data are generated, and a layer of image with the largest cyst area is output as a seventh key image; and when the number of the pixels is smaller than a third preset threshold value, generating data without the front cyst, and sending the data with the front cyst and the data without the front cyst to the AI scheduling module.

The AI scheduling module sends the data with the prostate cyst, the seventh key image and the data without the prostate cyst to the structured report module.

Fig. 16 is a diagram illustrating the fact that the prostate cyst detection module detects that there is data about prostate cysts displayed on the structured report interface in the system for detecting benign lesions of prostate and seminal vesicle on MRI according to the fifth embodiment of the present invention, and as shown in fig. 16, the structured report interface marks a square symbol in front of the prostate cyst control, automatically activates the control, and displays the seventh key image on the structured report interface.

An AI scheduling module, a prostate dissection module, a plurality of benign lesions auxiliary diagnosis modules and a structured report module in the embodiment of the invention can segment T2WI prostate gland dissection data, prostate peripheral zone dissection data, prostate transition zone dissection data, prostate central zone dissection data, pre-prostate fiber matrix zone dissection data, urethra dissection data and T2WI bilateral seminal vesicle gland dissection data on a T2WI sequence image, segment T1WI prostate gland dissection data and T1WI bilateral seminal vesicle gland dissection data on a T1WI sequence image, each prostate benign lesion auxiliary diagnosis module can diagnose benign lesions based on the dissection data of each partition of the prostate, respectively output diagnosis data of prostate hyperplasia, diagnosis data of prostatitis, diagnosis data of prostate cyst, diagnosis data of prostate hemorrhage, diagnosis data of seminal vesiculitis, diagnosis data of seminal cyst atrophy and diagnosis data of seminal vesicle hemorrhage, display the benign lesion data in a report, avoid lesions from being interfered by various common lesions in the diagnosis process of a doctor, can explain benign lesions, can improve the diagnosis efficiency of a benign diagnosis report, and provide an automatic diagnosis report for prostate cancer diagnosis, and provide a clinical diagnosis efficiency for a prostate cancer diagnosis and a doctor; because the module for detecting the prostatic hyperplasia and the prostatitis in the embodiment of the invention can automatically output the diagnosis data of the prostatic hyperplasia and/or the prostatitis to the report, can explain the clinical symptoms of patients and can distinguish the clinical significant cancer of the prostate; because the seminal vesicle atrophy and seminal vesiculitis detection module in the embodiment of the invention can automatically output seminal vesicle atrophy and/or seminal vesiculitis diagnosis data to a report, and has significance for making correct diagnosis for image examination, and because seminal vesicle inflammation and atrophy of old people can cause the reduction of seminal vesicle gland volume and unclear internal structure display, because the seminal vesicle atrophy and/or seminal vesiculitis diagnosis data are automatically output to the report, the seminal vesicle atrophy and seminal vesiculitis invasion can be identified with prostatic cancer; because the prostate bleeding and seminal vesicle bleeding detection module in the embodiment of the invention can automatically output diagnosis data of prostate bleeding and/or seminal vesicle bleeding to a report, can explain clinical symptoms (such as hemospermia), is helpful for diagnosis after puncture, generally considers that the probability of prostate cancer in an obvious and continuous bleeding area is very low, is an important indirect symptom, and excludes the probability of prostate cancer in the area; due to the prostate cyst detection module in the embodiment of the invention, the diagnosis data of the prostate cyst can be automatically output to the report and used as the reference standard for positioning when performing the needle biopsy.

From the above description, it can be seen that the above embodiments of the present invention achieve the following technical effects: because the embodiment of the invention is provided with the AI scheduling module, the prostate dissection module, the plurality of benign lesion auxiliary diagnosis modules and the structured report module, T2WI prostate gland dissection data, prostate peripheral zone dissection data, prostate transition zone dissection data, prostate central zone dissection data, pre-prostate fiber matrix zone dissection data, urethra dissection data and T2WI bilateral seminal vesicle segmentation data can be segmented on a T2WI sequence image, T1WI prostate gland dissection data and T1WI bilateral seminal vesicle gland dissection data are segmented on the T1WI sequence image, each prostate benign lesion auxiliary diagnosis module can diagnose benign lesions based on the dissection data of each partition of the prostate, and respectively output diagnosis data of prostatic hyperplasia, diagnosis data of prostatitis, diagnosis data of prostate gland hemorrhage, diagnosis data of seminal vesiculitis, diagnosis data of seminal vesicle atrophy and diagnosis data of seminal vesicle hemorrhage, the benign lesion data are displayed in a lesion report, thereby avoiding the interference of various common doctors in the diagnosis process, on the one hand, the diagnosis data can be used for identifying benign lesions, and the diagnosis of prostate cancer can be displayed in a clear and the background report can be provided for the diagnosis of benign prostate cancer, and the diagnosis of the doctor can be used for improving the diagnosis efficiency of the diagnosis of prostate cancer; because the embodiment of the invention is provided with the prostate hyperplasia and prostatitis detection module, the diagnosis data of the prostate hyperplasia and/or the prostatitis can be automatically output to a report, the clinical symptoms of a patient can be explained, and the diagnosis data can be distinguished from clinically significant carcinoma of the prostate; because the seminal vesicle atrophy and seminal vesiculitis detection module is arranged in the embodiment of the invention, seminal vesicle atrophy and/or seminal vesiculitis diagnosis data can be automatically output to a report, and the correct diagnosis of image examination is also significant, and because seminal vesicle inflammation and atrophy of old people can cause the reduction of seminal vesicle gland volume and the unclear display of internal structure, the seminal vesicle atrophy and/or seminal vesiculitis diagnosis data can be automatically output to the report, and the seminal vesicle atrophy and seminal vesiculitis diagnosis data can be identified from prostate cancer invading seminal vesicle; because the embodiment of the invention is provided with the prostate bleeding and seminal vesicle bleeding detection module, diagnosis data of prostate bleeding and/or seminal vesicle bleeding can be automatically output to a report, clinical symptoms (such as hemospermia) can be explained, diagnosis after puncture is facilitated, the probability of prostate cancer in an area with obvious and continuous bleeding is generally considered to be very low, the prostate cancer is an important indirect symptom, and the probability of prostate cancer in the area is eliminated; because the embodiment of the invention is provided with the prostate cyst detection module, the diagnosis data of the prostate cyst can be automatically output to a report and used as a reference standard for positioning when the needle biopsy is carried out.

It will be apparent to those skilled in the art that the modules or steps of the present invention described above may be implemented by a general purpose computing device, they may be centralized on a single computing device or distributed across a network of multiple computing devices, and they may alternatively be implemented by program code executable by a computing device, such that they may be stored in a storage device and executed by a computing device, or fabricated separately as individual integrated circuit modules, or fabricated as a single integrated circuit module from multiple modules or steps. Thus, the present invention is not limited to any specific combination of hardware and software.

The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A system for detecting benign lesions of prostate and seminal vesicle on MRI, comprising an AI scheduling module, a prostate dissection module, a plurality of benign lesion auxiliary diagnosis modules and a structured report module, wherein,

the AI scheduling module is respectively connected with the prostate segmentation and dissection module, each benign lesion auxiliary diagnosis module and the structured report module, and is used for extracting header file information of a DICOM image when a patient finishes scanning an examination item, namely examination of multi-parameter magnetic resonance imaging of the prostate, searching a T2WI sequence image and a T1WI sequence image based on the header file information, and sending the T2WI sequence image and the T1WI sequence image to the prostate segmentation and dissection module and each benign lesion auxiliary diagnosis module;

the prostate segmentation and dissection module is connected with the AI scheduling module and is used for receiving the T2WI sequence image and the T1WI sequence image, carrying out anatomical segmentation on the T2WI sequence image, segmenting T2WI prostate gland segmentation data, prostate peripheral zone segmentation data, prostate transition zone segmentation data, prostate central zone segmentation data, pre-prostatic fibrous stroma zone segmentation data, urethra segmentation data and T2WI bilateral seminal vesicle gland segmentation data, carrying out anatomical segmentation on the T1WI sequence image, and segmenting T1WI prostate gland segmentation data and T1WI bilateral seminal vesicle gland segmentation data; sending all the segmentation data to the AI scheduling module; at this time, the AI scheduling module is further configured to determine segmentation data matched with the benign lesion auxiliary diagnosis module, and send the segmentation data to the benign lesion auxiliary diagnosis module; the benign lesion auxiliary diagnosis module comprises a prostatic hyperplasia and prostatitis detection module, a seminal vesicle atrophy and seminal vesiculitis detection module, a prostate bleeding and seminal vesicle bleeding detection module and a prostate cyst detection module;

each benign lesion auxiliary diagnosis module is respectively connected with the AI scheduling module and is used for receiving the sequence images and the segmentation data matched with the benign lesion auxiliary diagnosis module, analyzing and processing the sequence images, outputting the diagnosis data of the lesions based on the segmentation data and feeding the diagnosis data back to the AI scheduling module; wherein the diagnostic data are text, numerical values and key images;

the structural report module is connected with the AI scheduling module and used for automatically generating a diagnosis result based on the diagnosis data, outputting diagnosis data of prostatic hyperplasia, prostatitis, prostatocystis, prostahemorrhage, seminal vesiculitis, seminal vesicle atrophy and seminal vesicle hemorrhage, and displaying key images corresponding to found benign lesions at corresponding positions of a structural report interface.

2. The system for detecting benign prostatic disorders and seminal vesicle gland on MRI according to claim 1, wherein when the benign auxiliary diagnosis module is a prostate hyperplasia and prostatitis detection module, the AI scheduling module sends the T2WI sequence image, the prostate transition zone segmentation data and the prostate peripheral zone segmentation data to the prostate hyperplasia and prostatitis detection module, the prostate hyperplasia and prostatitis detection module removes pixels in the region outside the circumscribed rectangle where the T2WI sequence image is located to obtain a processed image, and performs data normalization processing on the processed image and inputs the processed image into a classification network, outputs the classified number of prostate hyperplasia layers in the T2WI sequence image based on the prostate transition zone segmentation data, outputs the classified number of prostatitis layers in the T2WI sequence image based on the prostate peripheral zone segmentation data, and generates prostate hyperplasia data and/or data with the largest number of prostatitis layers as a first key image and a second key image when the classified number of prostate hyperplasia layers and the classified number of prostatitis layers are greater than or equal to a first preset threshold; when the classified layer number of the prostatic hyperplasia and the classified layer number of the prostatitis are smaller than the first preset threshold value, generating data without prostatic hyperplasia and/or data without prostatitis; and sending the data with prostate hyperplasia and/or the data with prostatitis, the first key image, the second key image and the data without prostate hyperplasia and/or the data without prostatitis to the AI scheduling module.

3. The system of claim 1, wherein when the benign lesion auxiliary diagnosis module is a seminal vesicle atrophy and seminal vesiculitis detection module, the AI scheduling module sends the T2WI sequence images and the T2WI bilateral seminal vesicle gland segmentation data to the seminal vesicle atrophy and seminal vesiculitis detection module, the seminal vesicle atrophy and seminal vesiculitis detection module removes pixels of an area outside a circumscribed rectangle where the T2WI sequence images are located to obtain processed images, and inputs the processed images into a classification network after data normalization processing, and outputs the number of classification layers of seminal vesicle atrophy and classification of seminal vesiculitis in the T2WI sequence images based on the T2WI bilateral seminal vesicle gland segmentation data, and when the number of classification layers of seminal vesicle atrophy is greater than or equal to a second preset threshold, seminal vesicle data is generated and simultaneously outputs a layer image with the largest area as a third key image; when the classification layer number of the seminal vesiculitis is greater than or equal to a first preset threshold value, generating seminal vesiculitis data, and outputting a layer of image with the largest area as a fourth key image; when the classification layer number of the seminal vesicle atrophy is smaller than a second preset threshold value, generating seminal vesicle atrophy data, and when the classification layer number of the seminal vesicle inflammation is smaller than a first preset threshold value, generating the seminal vesicle inflammation data; and sending the seminal vesicle atrophy data and/or the seminal vesiculitis data, the third key image, the fourth key image and the seminal vesicle atrophy data and/or the seminal vesiculitis absence data to the AI scheduling module.

4. The system for detecting benign prostatic and seminal vesicle lesions on MRI as claimed in claim 1, wherein when the benign lesion auxiliary diagnosis module is a prostate hemorrhage and seminal vesicle hemorrhage detection module, the AI scheduling module sends the T1WI sequence images, the T1WI prostate gland segmentation data and the T1WI bilateral seminal vesicle gland segmentation data to the prostate hemorrhage and seminal vesicle hemorrhage detection module, the prostate hemorrhage and seminal vesicle hemorrhage detection module removes pixels of the region outside the circumscribed rectangle where the T1WI sequence images are located to obtain processed images, and inputs the processed images into a classification network after data normalization processing, and outputs classification of prostate hemorrhage in the T1WI sequence images based on the T1WI prostate gland segmentation data, when the number of layers of classification of prostate hemorrhage is greater than or equal to a first preset threshold, prostate hemorrhage data is generated, and simultaneously outputs a layer image with the largest area as a fifth key image; outputting the number of classification layers of seminal vesicle bleeding in the T1WI sequence image based on the T1WI bilateral seminal vesicle gland segmentation data, generating seminal vesicle bleeding data when the number of classification layers of seminal vesicle bleeding is larger than or equal to a second preset threshold value, and outputting a layer of image with the largest area as a sixth key image; when the classification layer number of the prostate hemorrhage is smaller than a first preset threshold value, generating non-prostate-hemorrhage data, and when the classification layer number of the seminal vesicle hemorrhage is smaller than a second preset threshold value, generating non-seminal vesicle hemorrhage data; and sending the data of the bleeding with prostate and/or the bleeding with seminal vesicle, the fifth key image, the sixth key image and the data of the bleeding without prostate and/or the bleeding without seminal vesicle to the AI scheduling module.

5. The system for detecting benign lesions of prostate and seminal vesicle on MRI as claimed in claim 1, wherein when the benign lesion auxiliary diagnosis module is a prostate cyst detection module, the AI scheduling module sends the T2WI sequence images and the T2WI prostate gland segmentation data to the prostate cyst detection module, the prostate cyst detection module segments at least one suspicious cyst region through a trained UNET segmentation network, and calculates the number of pixels of each cyst region by using a connected domain algorithm, when the number of pixels is greater than or equal to a third preset threshold, prostate cyst data is generated, and a layer of image with the largest cyst area is output as a seventh key image; and when the number of the pixels is smaller than a third preset threshold value, generating data without the prostate cyst, and sending the data with the prostate cyst and the data without the prostate cyst to the AI scheduling module.