CN116958151A - Method, system and equipment for distinguishing adrenal hyperplasia from fat-free adenoma based on CT image characteristics - Google Patents

Method, system and equipment for distinguishing adrenal hyperplasia from fat-free adenoma based on CT image characteristics Download PDF

Info

Publication number
CN116958151A
CN116958151A CN202311221342.1A CN202311221342A CN116958151A CN 116958151 A CN116958151 A CN 116958151A CN 202311221342 A CN202311221342 A CN 202311221342A CN 116958151 A CN116958151 A CN 116958151A
Authority
CN
China
Prior art keywords
adrenal
value
lesion
volume
hyperplasia
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202311221342.1A
Other languages
Chinese (zh)
Other versions
CN116958151B (en
Inventor
白鑫
孙昊
张古沐阳
金征宇
付锐
许梨梨
张家慧
张晓霄
陈丽
彭倩瑜
郭二嘉
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Peking Union Medical College Hospital Chinese Academy of Medical Sciences
Original Assignee
Peking Union Medical College Hospital Chinese Academy of Medical Sciences
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Peking Union Medical College Hospital Chinese Academy of Medical Sciences filed Critical Peking Union Medical College Hospital Chinese Academy of Medical Sciences
Priority to CN202311221342.1A priority Critical patent/CN116958151B/en
Publication of CN116958151A publication Critical patent/CN116958151A/en
Application granted granted Critical
Publication of CN116958151B publication Critical patent/CN116958151B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10072—Tomographic images
    • G06T2207/10081—Computed x-ray tomography [CT]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20021—Dividing image into blocks, subimages or windows
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20092—Interactive image processing based on input by user
    • G06T2207/20104—Interactive definition of region of interest [ROI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30096—Tumor; Lesion

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Medical Informatics (AREA)
  • Quality & Reliability (AREA)
  • Radiology & Medical Imaging (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Geometry (AREA)
  • Apparatus For Radiation Diagnosis (AREA)

Abstract

The application relates to the field of intelligent medical treatment, in particular to a method, a system and equipment for distinguishing adrenal hyperplasia and fat-free adenoma based on CT image characteristics. Comprises the steps of acquiring CT images of adrenal gland parts; performing feature calculation on the CT image to obtain a feature value, wherein the feature is a density feature and/or a volume feature, the density feature refers to lesion density of an interested region in an axial adrenal lesion image, and the volume feature refers to lesion volume of the interested region in the adrenal lesion image; and inputting the characteristic values into a disease diagnosis model for classification, judging that adrenal gland hyperplasia is generated when the threshold value of the characteristic values is within a preset value, and judging that fat-free adenoma is generated otherwise. The application combines the density and volume characteristics to effectively distinguish adrenal hyperplasia from fat-free adenoma, reduces the probability of misdiagnosing hyperplasia as adenoma, avoids patients suffering from hyperplasia from unnecessary operations, and has good clinical value.

Description

Method, system and equipment for distinguishing adrenal hyperplasia from fat-free adenoma based on CT image characteristics
Technical Field
The application relates to the field of intelligent medical treatment, in particular to a method, a system, equipment and a computer readable storage medium for distinguishing adrenal hyperplasia from fat-free adenoma based on CT image characteristics.
Background
Adrenal hyperplasia and adipose adenoma are two different pathological conditions that are difficult to distinguish in some cases, mainly including imaging features, clinical manifestations, pathological features, biological behaviors, diagnostic criteria. Adrenal hyperplasia and fatty adenomas may exhibit similar imaging characteristics in imaging examinations (e.g., CT, MRI, etc.), such as changes in local density or signal, which may lead to diagnostic confusion; clinically, both may be manifested as endocrine symptoms such as primary aldosteronism, cushing's syndrome, etc., and thus cannot be distinguished from symptoms; pathologically, adrenal hyperplasia and the cellular morphology and tissue structure of certain types of lean adenomas (e.g., functional adrenal cortical adenomas) may be similar, resulting in difficult differentiation; in biological behavior, certain fatty adenomas may have secretory functions, functionally overlapping adrenal hyperplasia; finally, different doctors or medical institutions may employ different diagnostic criteria and methods, which may also affect the identification of these two diseases, and methods for differentiating adrenal hyperplasia from fatty adenoma are currently under continuous study.
Disclosure of Invention
Aiming at the problem that adrenal hyperplasia and fat-free adenoma are difficult to distinguish, the application provides a method for distinguishing the adrenal hyperplasia and the fat-free adenoma based on CT image characteristics, which specifically comprises the following steps:
acquiring CT images of adrenal gland parts;
performing feature calculation on the CT image to obtain a feature value, wherein the feature is a density feature and/or a volume feature, the density feature refers to lesion density of an interested region in an axial adrenal lesion image, and the volume feature refers to lesion volume of the interested region in the adrenal lesion image;
and inputting the characteristic values into a disease diagnosis model for classification, judging that adrenal gland hyperplasia is generated when the threshold value of the characteristic values is within a preset value, and judging that fat-free adenoma is generated otherwise.
Further, the measure of lesion density includes one or more of the following: flat scan CT value, CT attenuation value at portal vein, CT attenuation at abdominal aortic portal vein, maximum diameter of lesion, number of lesions, and position of lesion.
Further, calculating an average value of the measurements after repeating the measurements twice, and calculating an absolute strengthening value, a relative strengthening rate; the absolute enhancement value is calculated by firstly measuring a CT value of a portal vein period and a flat scan CT value, then subtracting the flat scan CT value from the CT value of the portal vein period, the relative enhancement value is calculated by firstly obtaining the flat scan CT value and the absolute enhancement value, then dividing the flat scan CT value by the absolute enhancement value, and the relative enhancement rate is calculated by firstly obtaining the absolute enhancement value and the CT value of the portal vein period of the abdominal aorta and then calculating the ratio of the absolute enhancement value to the CT value of the portal vein period of the abdominal aorta.
The measurement of the lesion volume includes one or more of the following: lesion volume, adrenal volume; the calculation of the volume characteristics comprises the steps of firstly obtaining the lesion volume and the adrenal volume, and then calculating the ratio of the lesion volume to the adrenal volume to obtain the volume ratio.
The disease diagnosis model comprises one or more of the following: logistic regression, random forest, support vector machine, XGboost, decision tree, extreme learning machine.
The preset value is obtained based on the maximum value of about sign index, when the lesion characteristic value exceeds the preset threshold value, the adrenal lesions are judged to be adenoma, otherwise, the adrenal hyperplasia is judged to be the adrenal hyperplasia.
The method further comprises a CT image area examination, which is to examine whether one or several of the following are included before the measurement: lesion edges, bleeding, calcification, artifacts, blood vessels, adipose tissue, necrosis, cystic areas, re-select the area of interest when any of the above is included, otherwise perform the eigenvalue measurement.
The application aims to provide a system for distinguishing adrenal hyperplasia from fatty adenoma based on CT image characteristics, which comprises the following steps:
a data acquisition unit: acquiring CT images of adrenal gland parts;
feature calculation unit: performing feature calculation on the CT image to obtain a feature value, wherein the feature is a density feature and/or a volume feature, the density feature refers to lesion density of an interested region in an axial adrenal lesion image, and the volume feature refers to lesion volume of the interested region in the adrenal lesion image;
regression classification unit: and inputting the characteristic values into a disease diagnosis model for classification, judging that adrenal gland hyperplasia is generated when the threshold value of the characteristic values is within a preset value, and judging that fat-free adenoma is generated otherwise.
The application aims to provide a device for distinguishing adrenal hyperplasia from fatty adenoma based on CT image characteristics, which comprises:
a memory and a processor, the memory for storing program instructions; the processor is used for calling program instructions, and when the program instructions are executed, any one of the methods for distinguishing adrenal hyperplasia from fat-free adenoma based on CT image characteristics is realized.
The present application aims to provide a computer readable storage medium having stored thereon a computer program which when executed by a processor implements a method of distinguishing between adrenal hyperplasia and adipose adenoma based on CT image features as any one of the above.
The application has the advantages that:
1. compared with the diagnosis results of radiologists, the method has better diagnosis performance, lower variability of the diagnosis results of the same case and better credibility of the diagnosis results.
2. The application creatively provides the method for measuring the lesion volume and the total volume of adrenal glands, calculating the ratio of the lesion volume to the total volume of adrenal glands and distinguishing the adrenal hyperplasia from the fat-free adenoma through the ratio, wherein the characteristic value shows good distinguishing capability.
3. The application calculates absolute enhancement value, relative enhancement value and relative enhancement rate by measuring the CT value of portal vein, flat scan CT value and CT value of abdominal aorta portal vein, and is used for assisting in judging the characteristics of the lesion area.
4. The application combines the CT value, the volume ratio and the lesion number of the portal vein to provide a distinguishing model of adrenal gland hyperplasia and lean fatty tumor, and compared with a doctor of radiology, the application can obtain excellent classifying performance and avoid unnecessary operation of patients with adrenal gland hyperplasia.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present application, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of a method for distinguishing adrenal hyperplasia from fatty adenoma based on CT image characteristics according to an embodiment of the present application;
FIG. 2 is a schematic diagram of a system for differentiating between adrenal hyperplasia and fatty adenoma based on CT image features according to an embodiment of the present application;
FIG. 3 is a schematic diagram of an apparatus for differentiating adrenal hyperplasia from adipose adenoma based on CT image features according to an embodiment of the present application;
FIG. 4 is a violin diagram of CT densitometry of the present application showing fatty adenoma and adrenal hyperplasia;
FIG. 5 is a three-dimensional visual analysis of lesions and adrenal glands of a 37 year old female suffering from cushing's syndrome and pathologically confirmed as adrenal cortex adenoma according to an embodiment of the present application;
FIG. 6 is a graph of ROC providing a logistic regression model for all study subjects according to an embodiment of the present application;
FIG. 7 is a graph showing the alignment of the portal CT values, volume ratio and lesion number in accordance with the present application;
FIG. 8 is a alignment chart provided by an embodiment of the present application and showing the goodness of fit of the model;
FIG. 9 is a ROC graph of a first set of logistic regression models provided by an embodiment of the application, with asterisks indicating the cut-off values of the models and dots indicating the sensitivity and specificity of subjective assessment by imaging physicians;
FIG. 10 is a graph of ROC of a second set of logistic regression models provided by embodiments of the present application.
Detailed Description
In order to enable those skilled in the art to better understand the present application, the following description will make clear and complete descriptions of the technical solutions according to the embodiments of the present application with reference to the accompanying drawings.
In some of the flows described in the specification and claims of the present application and in the above figures, a plurality of operations appearing in a particular order are included, but it should be clearly understood that the operations may be performed in other than the order in which they appear herein or in parallel, the sequence numbers of the operations such as S101, S102, etc. are merely used to distinguish between the various operations, and the sequence numbers themselves do not represent any order of execution. In addition, the flows may include more or fewer operations, and the operations may be performed sequentially or in parallel. It should be noted that, the descriptions of "first" and "second" herein are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, and are not limited to the "first" and the "second" being different types.
Fig. 1 is a schematic diagram of a method for distinguishing adrenal hyperplasia from fat-free adenoma based on CT image features, which specifically includes:
s101: acquiring CT images of adrenal gland parts;
in one embodiment, the adrenal glands are vital endocrine organs of the human body, and are named adrenal glands, as they are located above the kidneys on both sides. The left and right adrenal glands are located above the kidneys and are commonly wrapped by the fascia and adipose tissue of the kidneys. The left adrenal gland is half-moon shaped, and the right adrenal gland is triangular. Adrenocortical adenomas are benign tumors that occur in the adrenal cortex and secrete cortisol and aldosterone in a manner that is not affected by the adrenocorticotropic hormone (ACTH) or renin, angiotensin system, and are clinically manifested as cushing's syndrome or primary aldosteronism. Adrenocortical hyperplasia is overgrowth of the adrenal cortex, and causes include cushing's disease, ectopic corticotropin (ACTH) syndrome, and primary aldosteronism and congenital adrenocortical hyperplasia. The adrenal hyperplasia can be diffuse or nodular, the diffuse hyperplasia is shown as diffuse enlargement of adrenal gland and maintains the original shape, and the nodular hyperplasia is similar to adrenal adenoma in imaging.
In one embodiment, the clinical manifestations of adrenal hyperplasia and adenoma are similar, but the treatment is quite different. Adrenal hyperplasia is usually treated with drugs, while adenomas are treated with surgery. Adipose adrenal lesions are difficult to distinguish clinically and imagewise between adenomas and hyperplasia.
In one embodiment, the types of medical images include one or more of the following: CT images, nuclear magnetic resonance images, ultrasound images, X-ray images.
In one particular embodiment, the present application acquires CT images from a variety of different multi-detector CT scanners (Siemens Somatom Definition Flash, siemens Somatom Force and GE Discovery CT750 HD). After the pan-scan image acquisition, 80-90 mL volumes of iodine contrast agent were injected using a power injector, and arterial and portal vein phase images were acquired 25-30 seconds and 55-60 seconds after injection of contrast agent, respectively, for subsequent analysis.
In one embodiment, during the model training phase, the CT images are screened to obtain CT images that meet the requirements, wherein the exclusion criteria are: (a) Focus-on-scan CT < 10 HU [ measured by a radiologist resident x.b. with a 4-year reading experience ]; (b) Diffuse adrenal hyperplasia and corticotropin-dependent adrenal macronodular hyperplasia (AIMAH) [ reviewed by x.b. ]; (c) no clinical data and preoperative CT enhancement images; (d) lesion size < 10 mm. A total of 128 patients (83 fatty adenoma patients and 45 nodular hyperplasia patients) were included.
In one embodiment, the method further comprises a CT image region check, which is a check before measurement whether one or several of the following are included: lesion edges, bleeding, calcification, artifacts, blood vessels, adipose tissue, necrosis, cystic areas, re-select the area of interest when any of the above is included, otherwise perform the eigenvalue measurement.
In one embodiment, the training data of the present application is populated with 128 patients and uses available reference criteria based on image, clinical and pathology recordings. The demographic characteristics of the patients were as follows: 57 men and 71 women; average age, 49.3±12.2 years old. There was a statistically significant difference in gender (p=0.010) but no significant difference in age (p=0.472) between the two groups. 58 patients in the adenoma group had hypertension (58/83), while 40 of 45 patients in the proliferation group had hypertension (p=0.015). Most of the lesions in both groups were unilateral (111 vs 17, p=0.001) compared to bilateral. The proportion of bilateral lesions in the hyperplastic group (12/45) was higher than in the adenomatous group (5/83). There were 52 lesions on the left and 26 lesions on the right of the adenoma group. 45. In case of unilateral hyperplasia patients, 25 cases are located on the left side and 8 cases are located on the right side.
S102: performing feature calculation on the CT image to obtain a feature value, wherein the feature is a density feature and/or a volume feature, the density feature refers to lesion density of an interested region in an axial adrenal lesion image, and the volume feature refers to lesion volume of the interested region in the adrenal lesion image;
in one embodiment, the measure of lesion density comprises one or more of the following: flat scan CT value, portal phase CT value, abdominal aortic portal phase CT value, maximum diameter of lesion, number of lesions, and position of lesion.
In one embodiment, the average of the measurements is calculated after repeating the measurements twice, and the absolute reinforcement value, the relative reinforcement rate are calculated; the absolute enhancement value is calculated by firstly measuring a CT value of a portal vein period and a flat scan CT value, then subtracting the flat scan CT value from the CT value of the portal vein period, the relative enhancement value is calculated by firstly obtaining the flat scan CT value and the absolute enhancement value, then dividing the flat scan CT value by the absolute enhancement value, and the relative enhancement rate is calculated by firstly obtaining the absolute enhancement value and the CT value of the portal vein period of the abdominal aorta and then calculating the ratio of the absolute enhancement value to the CT value of the portal vein period of the abdominal aorta.
In one embodiment, the measurement of the lesion volume includes one or more of the following: lesion volume, adrenal volume; the calculation of the volume characteristics comprises the steps of firstly obtaining the lesion volume and the adrenal volume, and then calculating the ratio of the lesion volume to the adrenal volume to obtain the volume ratio.
In one particular embodiment, for densitometry, a circular or elliptical region of interest (ROI) is drawn over the adrenal lesions on the axial image to measure the CT values for the pan and portal phases. To ensure consistency, the ROI on the portal images is copied and pasted to the same location on the swipe image. For patients with multiple adrenal lesions, only the largest lesions (the largest diameter lesions in the axial image) were analyzed. In addition, CT values of the portal venous phase of the abdominal aorta were also measured. And simultaneously measuring the maximum diameter of the focus on the axial image, and recording the number and the positions of the focus. For each patient, all measurements were repeated twice and the average was taken for analysis while calculating absolute, relative fortification values.
Absolute enhancement value = focal portal CT value-plain CT value
Relative intensity value = absolute intensity value/flat scan CT value
In one specific embodiment, for volume measurements, the ROI of adrenal lesions and ipsilateral adrenal glands (including adrenal lesions) are manually drawn on each layer of image at a magnification of 6 to 12.5 x, all measurements being made at standard abdominal window settings. The volume ROI was reviewed and confirmed by a urogenital radiologist (h.s., with 17 years of experience). Based on the volume ROI, the lesion and adrenal volume data were evaluated using ITK-snap software, and the ratio of lesion volume to adrenal volume (volume ratio) was calculated.
In one embodiment, the application is processed to result in:
adrenal density: the average scan CT values (CT pre,25.1±8.3 HU vs 21.4±9.0 HU, p=0.023), portal CT values (CTp, 79.6±25.0 HU vs 66.4±25.8 HU, p=0.006) and absolute enhancement values (54.5±21.8 HU vs 45.0±24.0 HU, p=0.025) of the fatty adenoma were higher than proliferation, as shown in the violin plot of CT density for fatty adenoma and nodular hyperplasia shown in fig. 4. The average flat scan CT value (25.1 HU + -8.3 vs 21.4 HU+ -9.0) is (A) higher than the nodular hyperplasia, (B) the average portal CT value (79.6 HU + -25.0 vs 66.4 HU+ -25.8) is higher than the nodular hyperplasia, and (C) the absolute enhancement value (54.5 HU + -21.8 vs 45.0 HU+ -24.0) is higher than the nodular hyperplasia. There was no statistical difference in relative fortification values and relative fortification rates between the two groups (p=0.634, p=0.122).
Adrenal volume: the average volume of the focus of adenoma group and the adrenal gland volume on the same side are higher than those of hyperplasia group. Average lesion volumes of adenoma group and hyperplasia group were 9.1.+ -.1, respectively8.7 cm 3 And 3.0.+ -. 3.0cm 3 (p=0.033). The average volume of the total adrenal glands on the same side of the adenoma group and the hyperplasia group is 11.4+/-18.0 cm respectively 3 And 6.9.+ -. 2.8. 2.8 cm 3 (p=0.098). There was a significant statistical difference in the ratio of lesion volume to adrenal volume between the two groups (65.5.+ -. 26.3% vs. 36.8.+ -. 25.8%, P)<0.001 As shown in fig. 5, a 37 year old female with cushing's syndrome and pathologically confirmed as adrenal cortical adenoma had lesions and manual segmentation and three-dimensional visualization of adrenal glands. (A) adrenal gland manually segmented on CT image, (B) three-dimensional imaging of adrenal gland, (C) focus manually segmented on CT image, (D) three-dimensional imaging of focus.
Number, size of lesions: the difference in the number of lesions between the adenoma group and the hyperplasia group was statistically significant (p=0.039). Of 83 fatty adenoma patients, 76 (91.6%) had 1 nodule and 7 (8.4%) had 2 nodules. 45. Of the cases of nodular hyperplasia patients, 35 (77.8%), 8 (17.8%) and 2 (4.4%) had 1, 2 and 3 adrenal nodules, respectively. The proportion of patients with two and three adrenal nodules present in patients with nodular hyperplasia is significantly higher than in adenoma patients (p=0.028). The average size of adenomas and hyperplasia were 2.4.+ -. 1.0 cm (range, 1.03-8.08 cm) and 1.7.+ -. 0.5 mm (range, 1.02-3.13 mm) (P < 0.001), respectively.
S103: and inputting the characteristic values into a disease diagnosis model for classification, judging that adrenal gland hyperplasia is generated when the threshold value of the characteristic values is within a preset value, and judging that fat-free adenoma is generated otherwise.
In one embodiment, the disease diagnostic model includes one or more of the following: logistic regression, random forest, neural network, naive bayes, K-nearest neighbor, support vector machine, XGboost, adaBoost, gradient lifting, decision tree.
In one embodiment, the preset value is obtained based on a maximum value of about log index, and when the lesion characteristic value exceeds a preset threshold value, the adrenal lesion is judged to be adenoma, otherwise, the adrenal hyperplasia is judged to be adrenal hyperplasia.
In one embodiment, the method of the application for differentiating between adrenal hyperplasia and lean adenoma comprises: distinguishing by using density characteristics, distinguishing by using volume characteristics, and distinguishing by using density characteristics and volume characteristics; the density characteristic and the volume characteristic comprise an original measurement characteristic and a secondary calculation characteristic, wherein the density original measurement characteristic comprises a flat scan CT value, a portal vein CT value, an abdominal aorta portal vein CT value, a maximum diameter of a focus, the number of focuses and the positions of the focuses, and the density secondary calculation characteristic comprises an absolute strengthening value, a relative strengthening value and a relative strengthening rate; the original volume measurement features comprise lesion volume and adrenal volume, the secondary volume calculation feature is the ratio of the lesion volume to the adrenal volume, and the features are taken as distinguishing features singly or in any combination of N, wherein N is a natural number greater than or equal to 1.
In one specific embodiment, the present application invites four radiologists (readers 1-4, h.z., y.z., l.l.x., x.x.z., with radiological experience of 14 years, 5 years, and 4 years, respectively) to differentiate between fatty adenomas and adrenal hyperplasia as a comparison of the present application. The radiologist is blinded to the clinical history, patient information, and final diagnosis. 30 patients were randomly selected from all subjects and evaluated by each radiologist to calculate the inter-reader correspondence. The remaining 98 patients were randomized into two groups (group 1, n=49; group 2, n=49). Reader 1 and reader 3 evaluate the first group (n=79, 49[ group 1] +30), and reader 2 and reader 4 evaluate the second group (n=79, 49[ group 2] +30). In addition, after one month, the reader 1 and the reader 2 reevaluate all cases each previously diagnosed to evaluate the reader's internal consistency.
In one embodiment, the present application compares the group differences for all variables by t-test and chi-square test. And establishing a univariate Logistic regression model, and incorporating a variable with P < 0.05 in the univariate Logistic regression into the multivariate Logistic regression. Subject operating characteristics (ROC) curves were plotted and area under the curve (AUC), accuracy, sensitivity, specificity and threshold were calculated. The performance of four radiologists on adrenal lesions classification was assessed by calculating accuracy, sensitivity and specificity. Kappa values are used to show inter-reader consistency and intra-reader consistency for subjective evaluation. All statistical analyses were performed using SPSS version 27.0 (IBM, SPSS; chicago, ill., U.S.). The performance of AUCs was compared by the dilong test and ROC curves and calibration curves were plotted using R (version 4.2.1). Double sided P values < 0.05 were considered statistically significant.
In one embodiment, one-way logistic regression analysis shows that adenomas and hyperplasia differ statistically in sex, flat scan CT value, portal CT value, absolute enhancement value, lesion volume, volume ratio, lesion size and number. Multivariate logistic regression analysis was performed, showing that portal CT values, volume ratios, and lesion numbers were independent predictors of fatty adenomas (ratio: 1.029, [95% confidence interval: 1.004, 1.054],28.771 [95% confidence interval: 2.768, 299.018], and 0.190 [95% confidence interval: 0.057, 0.633 ]).
The present application establishes a multivariate logistic regression model comprising three factors, showing the ROC curve of the model (AUC 0.835, 95% confidence interval 0.764-0.907, accuracy 77.3%), as shown in fig. 6-10, fig. 6 shows the ROC curve of the logistic regression model for all subjects (AUC 0.835, n=128); FIG. 7 is a nomogram drawn in combination with portal CT values, volume ratio, and lesion number; fig. 8 is a calibration curve of the nomogram drawn to show the goodness of fit of the model. Fig. 9 and 10 are ROC curves of the first (auc0.824, n=79) and second (auc0.816, n=79) sets of logistic regression models, respectively. Asterisks indicate cut-off values of the logistic regression model, and dots indicate sensitivity and specificity of subjective evaluation by the reader. The AUC of the model was statistically different from that of reader 2 (p=0.042). The threshold was chosen based on the maximum value of the about log index, with > 0.676 as the threshold for adenoma, at which time the sensitivity of the model was 73.5% and the specificity was 80%. The alignment graph is built to evaluate the performance of the model and verify the goodness of fit of the alignment graph using a calibration curve.
In one embodiment, the model of the present application performed worse than the radiologist, and the subjective evaluation had relatively poor inter-reader consistency (Kappa range: 0.082-0.535), and the subjective evaluation had intermediate consistency between two radiologists before and after one month (reader 1, kappa=0.734; reader 2, kappa=0.583). The specificity and accuracy of the model of the application are superior to radiologists. For the first group of patients (n=79), the specificity of the model exceeded all radiologists (96.3% [ model ] vs 48.1%, 59.3% and 63.0%), and the accuracy of the model exceeded most radiologists (72.2% [ model ] vs 67.1%, 77.2% and 68.4%). For the second group of patients (n=79), the specificity and accuracy of the model exceeded all radiologists (specificity: 64.3% [ model ] vs 25.0%, 25.0% and 57.1%, accuracy: 77.2% [ model ] vs 72.2%, 72.2% and 75.9%), but the sensitivity of the model was slightly worse than the radiologists (84.3% [ model ] vs 98.0%, 98.0% and 86.3%). The Delong test showed that the AUC of reader 2 was statistically different from the AUC of the model (p=0.042).
In one embodiment, the present application combines a multivariate logistic regression model of flat scan CT values, volume ratios, and lesion numbers with an AUC of 0.835 (95% confidence interval of 0.764-0.907, sensitivity of 73.5%, specificity of 80%). The specificity and accuracy of the radiologist is worse than the model. The radiologist subjectively assessed intra-reader correspondence between 0.082 and 0.535, and the intra-reader correspondence for both radiologists was 0.734 and 0.583, respectively.
In one embodiment, CT density and volume can be used to identify adrenal hyperplasia and fatty adenoma. The logistic regression model in combination with portal CT values, volume ratios, and lesion numbers provides a valuable means to demonstrate better diagnostic performance than radiologists and is superior to radiologists in maintaining low variability and reproducibility. The application of the model can further reduce the probability of misdiagnosing hyperplasia as adenoma and avoid unnecessary operation of patients with hyperplasia.
In a specific embodiment, the application verifies the classification results of the model by experiments, such as the experimental results of model specificity, accuracy, AUC, PPV (positive predictive value), NPV (negative predictive value) in the training set of table 1 and the test set of table 2, based on the density characteristics and the volume characteristics of the CT image, through the logistic regression, the random forest, the neural network, the naive bayes, the K nearest neighbor, the support vector machine, XGboost, adaBoost, the gradient promotion and the decision tree algorithm. From table 1, it can be found that the random forest, the neural network, XGboost, adaBoost, the gradient lifting and the decision tree training effect show better effects, higher levels are obtained on various indexes, and in table 2, it can be found that the classification model of adrenal hyperplasia and lipoma has the highest sensitivity, the gradient lifting model, the XGboost model, the AUC area, the random forest model and the neural network model, the PPV is the naive Bayes model, the NPV is the XGboost model and the gradient lifting model. The evaluation indexes of the models are higher than those of radiologists, so that the classification method constructed based on the density characteristics and the volume characteristics has excellent classification performance and higher clinical significance.
Table 1 performance of training set classification results in various models
Table 2 performance of test set classification results in each model
Fig. 2 is a schematic diagram of a system for distinguishing adrenal hyperplasia from fat-free adenoma based on CT image features, which specifically includes:
a data acquisition unit: acquiring CT images of adrenal gland parts;
feature calculation unit: performing feature calculation on the CT image to obtain a feature value, wherein the feature is a density feature and/or a volume feature, the density feature refers to lesion density of an interested region in an axial adrenal lesion image, and the volume feature refers to lesion volume of the interested region in the adrenal lesion image;
regression classification unit: and inputting the characteristic values into a disease diagnosis model for classification, judging that adrenal gland hyperplasia is generated when the threshold value of the characteristic values is within a preset value, and judging that fat-free adenoma is generated otherwise.
Fig. 3 is a schematic diagram of an apparatus for distinguishing adrenal hyperplasia from fatty adenoma based on CT image features, which specifically includes:
a memory and a processor; the memory is used for storing program instructions; the processor is configured to invoke the program instructions, and when the program instructions are executed, any one of the methods described above is performed to distinguish between adrenal hyperplasia and adipose adenoma based on CT image features.
A computer readable storage medium storing a computer program for execution by a processor of any one of the above methods of distinguishing between adrenal hyperplasia and adipose adenoma based on CT image features.
The results of the verification of the present verification embodiment show that assigning an inherent weight to an indication may improve the performance of the method relative to the default setting. It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, which are not repeated herein. In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of the units is merely a logical function division, and there may be additional divisions when actually implemented, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical or other form. The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, each functional unit in the embodiments of the present application may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units. Those of ordinary skill in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be implemented by a program to instruct related hardware, the program may be stored in a computer readable storage medium, and the storage medium may include: read Only Memory (ROM), random access Memory (RAM, random Access Memory), magnetic or optical disk, and the like.
It will be appreciated by those skilled in the art that all or part of the steps in the method of the above embodiment may be implemented by a program to instruct related hardware, where the program may be stored in a computer readable storage medium, where the medium may be a rom, a magnetic disk, or an optical disk, etc.
While the foregoing describes a computer device provided by the present application in detail, those skilled in the art will appreciate that the foregoing description is not meant to limit the application thereto, as long as the scope of the application is defined by the claims appended hereto.

Claims (10)

1. A method for distinguishing adrenal hyperplasia from fat-free adenoma based on CT image features, the method comprising:
acquiring CT images of adrenal gland parts;
performing feature calculation on the CT image to obtain a feature value, wherein the feature is a density feature and/or a volume feature, the density feature refers to lesion density of an interested region in an axial adrenal lesion image, and the volume feature refers to lesion volume of the interested region in the adrenal lesion image;
and inputting the characteristic values into a disease diagnosis model for classification, judging that adrenal gland hyperplasia is generated when the threshold value of the characteristic values is within a preset value, and judging that fat-free adenoma is generated otherwise.
2. The method of distinguishing between adrenal hyperplasia and fatty adenoma based on CT image features of claim 1, wherein said measure of lesion density comprises one or more of the following: flat scan CT value, portal phase CT value, abdominal aortic portal phase CT value, maximum diameter of lesion, number of lesions, and position of lesion.
3. The method for distinguishing between adrenal hyperplasia and fatty adenoma based on CT image features according to claim 2, wherein the average of the measurements is calculated after repeating the measurements twice, and absolute, relative enhancement values are calculated; the absolute enhancement value is calculated by firstly measuring a CT value of a focal portal vein period and a flat scan CT value, then subtracting the flat scan CT value from the CT value of the portal vein period, the relative enhancement value is calculated by firstly obtaining the flat scan CT value and the absolute enhancement value, then dividing the flat scan CT value by the absolute enhancement value, and the relative enhancement rate is calculated by firstly obtaining the absolute enhancement value and the CT value of the abdominal aorta portal vein period and then calculating the ratio of the absolute enhancement value to the CT value of the abdominal aorta portal vein period.
4. The method of distinguishing between adrenal hyperplasia and fatty adenoma based on CT image features of claim 1, wherein the measurement of lesion volume comprises one or more of the following: lesion volume, adrenal volume; the calculation of the volume characteristics comprises the steps of firstly obtaining the lesion volume and the adrenal volume, and then calculating the ratio of the lesion volume to the adrenal volume to obtain the volume ratio.
5. The method of distinguishing between adrenal hyperplasia and fatty adenoma based on CT image features of claim 1, wherein the disease diagnostic model comprises one or more of the following: logistic regression, random forest, support vector machine, XGboost, decision tree, extreme learning machine.
6. The method of claim 1, wherein the predetermined value is based on a maximum value of about log indices, and wherein the adrenal lesions are determined to be adenomas when the lesion characteristic value exceeds a predetermined threshold, and wherein the adrenal lesions are determined to be adrenal hyperplasia otherwise.
7. The method of distinguishing between adrenal hyperplasia and adipose adenoma based on CT image features of claim 1, further comprising a CT image region examination, said CT image region examination being prior to measurement whether the examination includes one or more of the following: lesion edges, bleeding, calcification, artifacts, blood vessels, adipose tissue, necrosis, cystic areas, re-select the area of interest when any of the above is included, otherwise perform the eigenvalue measurement.
8. A system for distinguishing between adrenal hyperplasia and fatty adenoma based on CT image features, comprising:
a data acquisition unit: acquiring CT images of adrenal gland parts;
feature calculation unit: performing feature calculation on the CT image to obtain a feature value, wherein the feature is a density feature and/or a volume feature, the density feature refers to lesion density of an interested region in an axial adrenal lesion image, and the volume feature refers to lesion volume of the interested region in the adrenal lesion image;
regression classification unit: and inputting the characteristic values into a disease diagnosis model for classification, judging that adrenal gland hyperplasia is generated when the threshold value of the characteristic values is within a preset value, and judging that fat-free adenoma is generated otherwise.
9. An apparatus for differentiating between adrenal hyperplasia and fatty adenoma based on CT image features, comprising: a memory and a processor, the memory for storing program instructions; the processor is configured to invoke program instructions which when executed implement a method of distinguishing between adrenal hyperplasia and fatty adenoma based on CT image features as described in any of claims 1-7.
10. A computer readable storage medium having stored thereon a computer program, which when executed by a processor performs a method of distinguishing between adrenal hyperplasia and adipose adenoma based on CT image characteristics as claimed in any of claims 1 to 7.
CN202311221342.1A 2023-09-21 2023-09-21 Method, system and equipment for distinguishing adrenal hyperplasia from fat-free adenoma based on CT image characteristics Active CN116958151B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311221342.1A CN116958151B (en) 2023-09-21 2023-09-21 Method, system and equipment for distinguishing adrenal hyperplasia from fat-free adenoma based on CT image characteristics

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311221342.1A CN116958151B (en) 2023-09-21 2023-09-21 Method, system and equipment for distinguishing adrenal hyperplasia from fat-free adenoma based on CT image characteristics

Publications (2)

Publication Number Publication Date
CN116958151A true CN116958151A (en) 2023-10-27
CN116958151B CN116958151B (en) 2023-12-12

Family

ID=88449690

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311221342.1A Active CN116958151B (en) 2023-09-21 2023-09-21 Method, system and equipment for distinguishing adrenal hyperplasia from fat-free adenoma based on CT image characteristics

Country Status (1)

Country Link
CN (1) CN116958151B (en)

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102429679A (en) * 2011-09-09 2012-05-02 华南理工大学 Computer-assisted emphysema analysis system based on chest CT (Computerized Tomography) image
US20130129168A1 (en) * 2011-11-23 2013-05-23 The Regents Of The University Of Michigan Voxel-Based Approach for Disease Detection and Evolution
US20160140414A1 (en) * 2014-11-13 2016-05-19 Kabushiki Kaisha Toshiba Density measuring device, density measuring method, and computer program product
US20170039737A1 (en) * 2015-08-06 2017-02-09 Case Western Reserve University Decision support for disease characterization and treatment response with disease and peri-disease radiomics
CN111160812A (en) * 2020-02-17 2020-05-15 杭州依图医疗技术有限公司 Diagnostic information evaluation method, display method, and storage medium
CN113706435A (en) * 2020-05-09 2021-11-26 北京康兴顺达科贸有限公司 Chest enhanced CT image processing method based on traditional image omics
US20220015698A1 (en) * 2015-09-28 2022-01-20 Lan Jiang Method of identifying tumor drug resistance during treatment
CN114494163A (en) * 2022-01-13 2022-05-13 浙江工业大学 Image omics feature selection method based on momentum self-adaption and acoustic search
CN115661107A (en) * 2022-11-07 2023-01-31 中国医学科学院北京协和医院 Image analysis method, system and equipment based on bladder cancer risk stratification
US20230248998A1 (en) * 2023-04-12 2023-08-10 Buvaneswari Natarajan System and method for predicting diseases in its early phase using artificial intelligence

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102429679A (en) * 2011-09-09 2012-05-02 华南理工大学 Computer-assisted emphysema analysis system based on chest CT (Computerized Tomography) image
US20130129168A1 (en) * 2011-11-23 2013-05-23 The Regents Of The University Of Michigan Voxel-Based Approach for Disease Detection and Evolution
US20160140414A1 (en) * 2014-11-13 2016-05-19 Kabushiki Kaisha Toshiba Density measuring device, density measuring method, and computer program product
US20170039737A1 (en) * 2015-08-06 2017-02-09 Case Western Reserve University Decision support for disease characterization and treatment response with disease and peri-disease radiomics
US20220015698A1 (en) * 2015-09-28 2022-01-20 Lan Jiang Method of identifying tumor drug resistance during treatment
CN111160812A (en) * 2020-02-17 2020-05-15 杭州依图医疗技术有限公司 Diagnostic information evaluation method, display method, and storage medium
CN113706435A (en) * 2020-05-09 2021-11-26 北京康兴顺达科贸有限公司 Chest enhanced CT image processing method based on traditional image omics
CN114494163A (en) * 2022-01-13 2022-05-13 浙江工业大学 Image omics feature selection method based on momentum self-adaption and acoustic search
CN115661107A (en) * 2022-11-07 2023-01-31 中国医学科学院北京协和医院 Image analysis method, system and equipment based on bladder cancer risk stratification
US20230248998A1 (en) * 2023-04-12 2023-08-10 Buvaneswari Natarajan System and method for predicting diseases in its early phase using artificial intelligence

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
郭凯等，: "CT动态增强鉴别肾上腺乏脂肪腺瘤与肾上腺结节样增生的价值", 《医学影像学杂志》, vol. 30, no. 4, pages 632 - 635 *

Also Published As

Publication number Publication date
CN116958151B (en) 2023-12-12

Similar Documents

Publication Publication Date Title
US7418123B2 (en) Automated method and system for computerized image analysis for prognosis
Zhang et al. Quantitative CT texture analysis for evaluating histologic grade of urothelial carcinoma
Pickhardt et al. Abdominal CT with model-based iterative reconstruction (MBIR): initial results of a prospective trial comparing ultralow-dose with standard-dose imaging
Giger et al. Breast image analysis for risk assessment, detection, diagnosis, and treatment of cancer
Shen et al. Quantitative parameters of CT texture analysis as potential markers for early prediction of spontaneous intracranial hemorrhage enlargement
Sofue et al. Dual‐energy computed tomography for non‐invasive staging of liver fibrosis: Accuracy of iodine density measurements from contrast‐enhanced data
US10339648B2 (en) Quantitative predictors of tumor severity
Ganeshan et al. Hepatic enhancement in colorectal cancer: texture analysis correlates with hepatic hemodynamics and patient survival
Wienbeck et al. The role of cone-beam breast-CT for breast cancer detection relative to breast density
Bottari et al. Dual-source dual-energy CT in the evaluation of hepatic fractional extracellular space in cirrhosis
US11241190B2 (en) Predicting response to therapy for adult and pediatric crohn&#39;s disease using radiomic features of mesenteric fat regions on baseline magnetic resonance enterography
CN113017674B (en) EGFR gene mutation detection method and system based on chest CT image
Liu et al. Detection and segmentation of pelvic bones metastases in MRI images for patients with prostate cancer based on deep learning
Bonatti et al. Accuracy of unenhanced CT in the diagnosis of cerebral venous sinus thrombosis
Nguyen et al. Effect of phase of enhancement on texture analysis in renal masses evaluated with non-contrast-enhanced, corticomedullary, and nephrographic phase–enhanced CT images
Kang et al. The cone-beam breast computed tomography characteristics of breast non-mass enhancement lesions
Liu et al. The ability of ADC measurements in the assessment of patients with stage I endometrial carcinoma based on three risk categories
Kulkarni et al. Digital breast tomosynthesis: potential benefits in routine clinical practice
CN110234270B (en) Method and apparatus for providing quantitative volume maps of organs or assessment of organ health
Vag et al. Kinetic characteristics of ductal carcinoma in situ (DCIS) in dynamic breast MRI using computer-assisted analysis
Ueki et al. Efficacy of preoperative cine magnetic resonance imaging in evaluation of adhesion of renal cancer thrombus to inferior vena cava wall
Niu et al. Radiomic evaluations of the diagnostic performance of DM, DBT, DCE MRI, DWI, and their combination for the diagnosisof breast cancer
Danala et al. Developing new quantitative CT image markers to predict prognosis of acute ischemic stroke patients
Takahashi et al. Usefulness of z-score mapping for quantification of extent of hypoattenuation regions of hyperacute stroke in unenhanced computed tomography: analysis of radiologists' performance
CN116958151B (en) Method, system and equipment for distinguishing adrenal hyperplasia from fat-free adenoma based on CT image characteristics

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant