JPH039737A - Image diagnosing apparatus for judging malignancy of thyroma - Google Patents

Abstract

PURPOSE:To judge quality of knotting thyroma from image information by an ultrasonic tomograph by arranging a means for inputting a degree having a specified feature, an arithmetic processing means for characterizing a degree of malignancy of thyroma numerically, a diagnosing means for judging malignancy of thyroma and a means for displaying the results of diagnosis. CONSTITUTION:A diagnosing device for malignancy of thyroma has a personal computer mainframe 1, an operating section comprising a mouse input device 2 and a keyboard 4 and a display as CRT display device 3. The display device 3 displays seven query items for ultrasonic images in characters and allows responses from a subject to be inputted to queries. The display device 3 has an ID display screen 5 for displaying number of images, date, names of subject and the like inputted from the keyboard 4, a menu screen 6 for showing an operation process and a results display screen 7 for displaying the results of diagnosis. When responses are inputted to all items, an input ending key is depressed and then, an information processing key is depressed, an image diagnosis is performed to carry out an arithmetic processing of an input data. When a results of diagnosis display key is depressed, a display is given on the results display screen.

Description

[Detailed Description of the Invention] Background of the Invention When diagnosing thyroid disease, cases that appear to be a thyroid nodule through examination such as visual inspection and palpation may undergo X-rays and blood tests to further confirm and differentiate the diagnosis. , auxiliary diagnosis such as ultrasonic tomography is performed. Among these, ultrasound examinations are non-invasive to the examinee and are easy to use.
Although it is widely used, its diagnostic accuracy is still not sufficient. The need for criteria for ultrasonographic diagnosis of nodular goiter is being discussed and considered in academic societies, but it has not yet been established. In ultrasound diagnosis, many various factors are intricately related, such as shape, margin, border, echo, internal echo, etc., and what factors are important for determining whether a nodule is benign or malignant, and how It is not yet clear whether it is compatible. Furthermore, the examiner's criteria for determining each item are not constant and vary from image to image, and the weight (importance) of each item is not constant but varies from image to image.

In other words, the degree of judgment for each item depends on the subjectivity of the individual, the judgment criteria are not constant, and the judgment of the individual is also affected by the impression on the previous screen and by the physical condition and psychological state of the day. Furthermore, once one has the impression that a certain observation item is malignant, other observation items tend to be judged based on that view.

These are major features of the fuzzy nature that intervenes in ultrasound diagnosis and image diagnosis. However, even though there is no suspicion of malignancy in other observation items, if there is a strong suspicion of malignancy in some observation items, or characteristics that indicate malignancy are observed, the examiner should examine that one item or This is important because malignancy is determined based solely on the results of the binomial test performed.

Because of the fuzzy nature of image information itself, there has been a strong desire to develop a more accurate image diagnostic apparatus.

SUMMARY OF THE INVENTION An object of the present invention is to provide an image diagnostic apparatus for determining whether a nodular goiter is benign or malignant based on image information obtained from an ultrasonic tomogram.

The image diagnostic apparatus for determining whether goiter is benign or malignant according to the present invention is capable of at least six items of shape, margin, internal echo, high intensity, cyst, and surrounding paper echo zone in an ultrasound tomogram of nodular goiter. Means for inputting the degree of having predetermined characteristics, arithmetic processing means for quantitatively characterizing the degree of malignancy of goiter based on the degree of each input item, and the sum of the quantities obtained by the arithmetic processing of the above six items. The present invention is characterized by comprising a diagnostic means for determining benign or malignant goiter based on the above, and a means for displaying the diagnosis result.

Preferably, the above input, arithmetic processing, and summation calculation are performed for seven items including the characteristics regarding the boundaries of the ultrasonic tomographic image.

To put it more clearly, the device according to the present invention: - quantifies the ambiguity of the characteristics of each of the above six items, preferably the above seven items, using a fuzzy scale, and - calculates the weight of each item (evaluation score for malignancy). ) in a stable form in response to fuzzy input, ■ prepare scores in advance that can detect potentially malignant adenomas even with just 2 of the 11 items, and ■ determine whether an adenoma is benign or malignant using the total score from all items. It is characterized by making a comprehensive judgment.

In this invention, six items, preferably seven items, such as shape, are adopted, and the extent to which the characteristics of these items contribute to malignancy is scored and quantified. This quantification is based on the results of an analysis using discriminant analysis of statistical analysis (multivariate analysis, quantification type II) based on clinical data of nodular goiter.

In addition, in the image diagnostic apparatus according to the present invention, the characteristics of the seven items described above are given through fuzzy input, and the degree to which these items contribute to the malignancy of the adenoma is expressed in the form of a membership function based on data analysis. Therefore, in determining whether a thyroid nodule is benign or malignant by ultrasonic diagnosis, where the diagnostic accuracy of images has traditionally been questionable, this image diagnostic apparatus enables highly accurate ultrasonic diagnosis. Furthermore, by quantitatively clarifying the relationship between six or seven image items and whether the image is benign or malignant, image diagnosis with a high accuracy rate can be expected.

When the raw data of 60 cases were actually diagnosed using this image diagnostic equipment, all 21 cases were determined to be malignant, and 3 cases were determined to be malignant.
Of the 9 benign cases, 36 were benign and 3 were “unknown”
It was determined that

DESCRIPTION OF THE EMBODIMENTS FIG. 1 shows the appearance of an image diagnostic apparatus for benign or malignant goiter.

This diagnostic device includes a personal computer main body 1,
It includes an operation section consisting of a mouse input device 2 and a keyboard 4, and a display section consisting of a CRT display device 3.

The display device 3 displays seven question items for the ultrasound image in text, and also displays the examiner's answers to the questions (degree of degree) so that they can be input at a given position on a straight line. The display device 3 also displays the image number, date, and examiner core (diagnosis input person name) input from the keyboard 4.
An ID display screen 5 for displaying information such as information, a menu screen 6 for displaying operation processes such as information input, information processing, result display, and termination, and a result display screen 7 for displaying diagnostic results.

A numeric keypad 8, an input correction key 9, an image diagnosis redo key 10, an image diagnosis end key 11, and other function keys are provided on the keyboard 4 of the operation section, and a power switch 12 is also arranged.

Furthermore, the floppy disk unit device 1 of this device
3. The system floppy FDI for starting up the diagnostic imaging device is in the upper row of 3, and the floppy FDI for data files is in the lower row.
D2 are respectively stored.

Answers for each item of image information are input from the mouse device 2 or the keyboard 4, and the examiner's degree of affirmation for the item (fuzzy information) is expressed by position. The degree of input is displayed with a red cursor on the screen, allowing examiners to confirm the information they have entered. If you feel that the input level is inappropriate, you can change it by moving the cursor to the position you feel is appropriate.

When the input end key is pressed when inputting answers for all items is completed, inputting of image information is completed. Next, when the information processing key is pressed, image diagnosis is performed and the input data is processed. When you press the diagnosis result display key, the results are displayed on the result display screen.

The floppy disk FD1 for the image diagnosis system is for a system for inputting and processing image information, and the floppy disk FD2 stores response data and image processing results for one or more images. It is meant to be kept.

When the power of the device is turned on and the floppy disk for the fuzzy image diagnosis system and the floppy disk for data are inserted, the image analysis system automatically starts and the image diagnosis menu screen is displayed on the display device 3 as shown in Figure 2. Ru.

On the menu screen, the patient's name, ID number, image number,
You can enter the date, image analyzer's name, etc.

On this menu screen, select the type of ultrasound tomography (10MHZ
% 7.5 MHz, etc.). Items such as shape, margin, boundary, internal high-intensity echo, cyst, and peripheral low-echo zone can be selected as image information items.

If preset items are acceptable, input them using the mouse cursor.

On the input screen, as shown in FIG. 2 or 3, input items for image information such as shape and boundary are displayed. While looking at the actual medical image (ultrasound tomogram), the examiner indicates the degree of each item using the cursor position (in this case, the input method can start from any item). Focus on the item ``Border'' and check the degree of the item ``When it is judged to be somewhat unclear, change the level (from 0 to 1
) Move the cursor to where you think this is on the straight line, and press the YES key on the mouse there. The position will then be fixed with a small red cursor. The position does not necessarily have to be one point, but if it is thought to have a certain width, the cursor can be used to indicate the minimum position (starting point) and maximum position (end point) (both ends of the width) of the degree in each item.

As shown in FIG. 4, membership functions and evaluation scores for malignant tumors are preset for each of the seven items.

The examiner inputs the degree of each of the seven items as described above. FIG. 5 shows the input method. Use the mouse to enter the position of each item with the minimum degree as the starting point and the maximum as the end point. - When input as point information, add a width of +0.05 to the left and right of the point and use it as the starting point and ending point. This input width is displayed on the screen in yellow, and is stored in the calculation as an isosceles triangular distribution with xl)X2 as the base.

Next, referring to FIG. 6, the grade of the point having the maximum grade in the common subset of the input triangular distribution and the item category membership function is calculated. Next, the union of the membership functions regarding the scores of each category is truncated at the maximum grade value mentioned above, and the score corresponding to the center of gravity is determined as the evaluation score for the item. As a result, a score corresponding to the robust center of gravity is given to the instability of the determination of each item of image information. Scores are calculated for all seven items using the same procedure.

FIG. 7 shows the process of comprehensive judgment based on seven items. The scores for the 7 items are totaled to form a total score (Y).

The criteria for determining benign or malignant are determined as follows for the total score Y obtained above.

When Y > 2.0, it is malignant. When 2.0≧Y≧2.0, the judgment is unclear. When Y < -2,0, it is benign. It is expressed as a degree of - A method is adopted in which a score is calculated for each item and the overall score is used to determine whether the item is benign or malignant. This is because malignant features may appear prominently. For this reason, scores are distributed so that even binomial lesions are determined to be malignant.

In this image diagnostic apparatus, input of each item, arithmetic processing, and overall judgment are performed using the following method % formula % (1) When inputting the degree of each item, the input method (1) in Figure 8 is used. As shown, the position of the point on the (0,1) line for each item can be entered as a single point (x), and as shown in input method (2), the starting point (
It is also possible to input in a section having a width surrounded by the end point (xl) and the end point (x2).

In both input methods (1) and (2), the input position information is further converted into the form of a singleton or triangular membership function.

Conversion from − point input information to triangular membership function using input method (1) is performed by adding ±a( to the left and right of the point position (X)
For example, this is realized by adding a width of a −0,05) and configuring a membership function of an isosceles triangle.

Conversely, when constructing a singleton using input method (2), the midpoint of X s X 2 is found and the information as a singleton is converted at that position.

(2) Arithmetic Processing As explained using FIG. 8, the input information is converted into the form of a singleton or an isosceles triangle membership function. In the arithmetic processing, evaluation scores (scores) for the malignancy of each item are calculated for these two types of input.

The score calculation process when the input is a triangular membership function is shown in FIG. If the input is a singleton, you can also do it in the following way. That is, one method is to read the intersection of the singleton and category membership functions and calculate the weighted average of the weights (scores) of each category, as shown in the upper part of FIG. 9. The other method, as shown in the lower part of FIG. 9, is to use a singleton to truncate the membership function of the scores and find the center of gravity of the union of the membership functions of the scores of each category.

(3) Comprehensive Judgment In this image diagnostic apparatus, a malignancy evaluation score was calculated for each of the seven items of image information, and the scores of the seven items were comprehensively judged.

As shown in Figure 7, one method for comprehensive judgment is to find the center of gravity of the membership function of the score for each item to obtain a score, then add up the scores for these seven items to calculate the total score. There are methods to determine benign or malignant. Another method, as shown in Figure 1O, is to leave the membership function itself representing the score for each item, find the union of these membership functions for the seven items, and calculate this newly created comprehensive There is a method of determining based on a score membership function. In this case, you can find the center of gravity of this comprehensive membership function, or calculate the area with a score of 2.0 or more, and if the area is 10% or less of the total area, there is "low suspicion of malignancy" , 10 to 30%, it can be determined that there is a slight suspicion of malignancy, 30 to 50%, that there is a strong suspicion of malignancy, and 50% or more, that there is a very strong suspicion of malignancy.

FIG. 11 shows the flow of the processing procedure in the above-mentioned apparatus.

[Brief explanation of drawings]

FIG. 1 is a perspective view showing the appearance of an image diagnostic apparatus for determining whether goiter is benign or malignant. FIG. 2 is a front view of the display screen of the image diagnostic apparatus. FIG. 3 shows each item of image information. FIG. 4 shows membership functions for each item of image information and evaluation scores for malignancy. FIG. 5 shows an input method for one item of image information. FIG. 6 shows the arithmetic processing process from the input position information to the calculation of the evaluation score for malignancy. FIG. 7 shows the process of benign or malignant judgment diagnosis from input of seven items of image information to comprehensive judgment of malignancy. FIG. 8 shows the types of input methods and input data conversion methods for each item of image information. FIG. 9 shows two scoring methods when input data is converted to a singleton. FIG. 10 shows a method for calculating the total score when the score for each item is expressed by a membership function. FIG. 11 is a flow chart showing the overall processing procedure of the image diagnostic apparatus. that's all

Claims (3)

[Claims] (1) A means for inputting the degree to which an ultrasound tomogram of nodular goiter has predetermined characteristics with respect to at least the six items of shape, margin, internal echo, high brightness, cyst, and peripheral hypoechoic zone; a calculation processing means for quantitatively characterizing the degree of malignancy of the goiter based on the degree of each of the above six items; a diagnostic means for determining the benignity or malignancy of the goiter based on the sum of the quantities obtained by the calculation processing of the six items; and an image diagnostic device for determining benign or malignant goiter, which is equipped with a means for displaying diagnostic results. (2) The apparatus according to claim (1), wherein the input, arithmetic processing, and summation calculation are performed for seven items including features regarding boundaries of ultrasonic tomographic images. (3) A claim characterized in that the degree of malignancy of the adenoma in each of the above items is characterized by a membership function (
1).