JPS63240832A - Image measuring method for calculating contour line of lung part - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a measurement method for determining the contour of a lung field from chest radiation image data.

[Prior art]

Conventionally, a method has been proposed in which the outline of the lung field is extracted through image processing of a chest X-ray image, and it is determined from the shape of the obtained outline whether or not there is an abnormality in the subject (for example, 62-26047), to extract the contour line when performing such processing, the density data of the image is binarized using an appropriate threshold value, and the boundary between ``IJ'' and ``0'' is traced to extract the contour line. It is said that Alternatively, an edge image is created by applying filtering processing to the density data of the original image using a Laplacian or an edge extraction operator, and after removing noise and unnecessary edges from the resulting edge image, A contour image is obtained by performing thinning processing.

However, with the former contour extraction method, it is difficult to obtain a perfect threshold that separates the lung field from other parts because the density level differs depending on the position even in one part of a chest X-ray image. It is impossible. In addition, in the latter method, the type of original image, density level, pixel size,
Filtering conditions such as operator values vary greatly depending on the region for which the contour line is to be determined and other detailed conditions, and a considerable amount of specialized knowledge is required to determine the optimal conditions. In addition, when various body parts overlap in a complex manner, such as in an X-ray image of the human body, it is difficult to extract all the contour lines with a single filtering method, so complex processing such as combining several types of filtering is necessary. becomes necessary.

[Purpose of the invention]

An object of the present invention is to provide a method for easily and accurately extracting the outline of a specific region in a complex image such as a chest radiation image of a human body.

[Structure of the invention]

For this purpose, the present invention focuses on only one row or column of chest radiation image data expressed in a matrix, and the relationship between the values of the preceding and succeeding data in the one-dimensional density data string is predetermined. The points forming a specific pattern are taken as the contour points in that row or column, the contour points are found for the necessary range of rows or columns, and the line connecting these points is taken as the lung field contour line. did.

〔Example〕

Examples of the present invention will be described below. FIG. 1 is a diagram showing the configuration blocks of an apparatus for implementing the present invention. 1 is an image input device for reading an X-ray photographic image of the subject; 2 is an image storage device for storing density data of the read X-ray image; and 3 is a contour reading device for reading the density data stored in the image storage device. An image analysis device performs line extraction processing, a determination device 4 determines abnormality of the subject from the analyzed results, and a display device 5 notifies the user of the determined results.

First, the image input device 1 scans a chest X-ray image plane (film) with a spot of a light beam using a laser beam or the like using an optical deflector such as a polygon mirror, and records the intensity of the transmitted light at that time. This is a device that performs photoelectric conversion using a photodetector such as a sensor, sequentially A/D converts the obtained electric signals, and outputs them as density data of a digital image. Although the light beam is mechanically scanned here, a method of electronically scanning using a solid-state imaging device such as a COD or an imaging tube may also be adopted. Further, if a CCD line sensor or the like is used, it is also possible to read a plurality of chest X-ray image films continuously while mechanically conveying them one after another.

For the image storage device 2, a semiconductor memory such as a RAM is usually used due to its high processing speed, but a disk memory such as a magnetic disk, floppy disk, or optical disk, a magnetic tape, a magnetic bubble memory, etc. can also be used. .

The image analysis device 3 is a part constituting the main part of the present invention.

This device 3 sequentially reads the density data of the images stored in the image storage device 2 and performs processing to obtain data indicating the outline of the portion corresponding to the lung field and the surrounding area. This device 3
, a line memory capable of storing at least one row or one column of read image density data;
It consists of a processor that performs analysis, a contour data memory that stores extracted contour data, a main storage that stores analysis algorithms, etc., and that serves as a work area during analysis, an auxiliary storage, and the like. Here, it is also possible to process all of the image density data and contour data using a single main storage device, without using a configuration that includes a memory for storing the data.

The determination device 4 calculates various feature amounts of the contour data obtained by the image analysis device 3, and determines whether each item of the medical examination is out of the standard value range.

The display device 5 notifies the doctor, based on the output of the determination device 4, that there is an abnormality in the subject whose image data has been analyzed.

Next, the effect will be explained.

(Input of X-ray photographic image) The X-ray image is read by the image input device 1 with a sampling pinch of approximately 200 μm, and the density range θ~3 is
Quantized into gradations. At this time, the X-ray image is input with the user facing upright, and as shown in Figure 2, the image is in m-by-n matrix format with (1, 1) at the top left and (m, n) at the bottom right.
The image is stored in the image storage device 2.

Note that the sampling pinch is not fixed to 200 μm, but can be changed in the range of 25 μm to 1000 μm depending on the case.

Furthermore, the reading density range and the number of quantization levels can also be changed.

(Extraction of lung field contours) As shown in Figure 2, the lung field contours are extracted using the outer contour AL, AR1 inner contour BL, and BR1 diaphragm contour [CL, CR] for each of the left and right lung fields. Do about.

The outer and inner contour lines AL and AR1BLSBR are obtained as an array of values in the column where the contour line exists in each row of the image data, and the diaphragm contour lines CL and CR are obtained as the values in the row where the contour line exists in each column. It is found as an array of .

To extract the outer contour lines AL, AR and inner contour lines BL, BR of the lung field, one horizontal row of image density data is read from the image storage device 2 into the image analysis device 3, and the statistics of all the read data are calculated. Quantities are calculated and several concentration thresholds are determined from those values. Among the image density data (points) having values within the range of these thresholds, if the relationship between the values of a certain point and the points before and after it satisfies a specific pattern (condition) described later, that certain point The position of is stored as the column position of the lung field outline in that row. Specific patterns include ■Points where the minimum is reached...For example, C■Points where the slope is the maximum in Figure 3 (BL's a■Points where the slope is the minimum...For example, the third appearance)... For example, in Fig. 3 (b of bl, etc.), find some points that satisfy these patterns. Fig. 3 shows (b) the density change of the line in the column direction of the row ml x m4 indicated by (al).D is the density (transmittance).Furthermore, ■Position of the line to be analyzed on the image■Position of the determined point■Density value of the determined point■Slope of the value at the determined point■ Comprehensively judge the range of the threshold values used, etc., and decide which contour point the point is or whether it is noise.For example, if it is a point that constitutes the contour AR outside the right lung field, "Image 1 on the left side when facing
/3 and is less than the average value of all concentration data (
When density is expressed as transmittance, a high concentration portion shows a small transmittance. ), and the relationship between the values of the preceding and succeeding points is minimal, or the point J has a slope of O.

Such search algorithms can be used in traditional procedural programming languages such as FORTRAN and Pa5ca.
1, etc., but it can be described even more easily using recent rule-oriented languages such as 0PS5.0P383.

In this way, the positions of the points constituting the respective contours are determined for each row of image data, and the necessary parts are connected to obtain the respective contour lines.

Regarding the contour line of the diaphragm, image data for one column in the vertical direction m is read from the image reading device 1 into the line memory of the image analysis device 3, and the position of the row in the column where the contour exists is determined in the same manner as described above. Get the contour line. Figure 4 is (
(b) shows the density change in the lines in the row direction of columns n1 to ns indicated by al. As a search condition, a certain point is mainly used where the difference in value between two points several points before and after a certain point is maximum. In addition, in FIG. 4, the point where the slope is maximum is used (see d in FIG. 4 (bl)).

In this way, for each left and right lung field, the outer AL, AR,
The contour lines of the inner side BL, BR, diaphragm CL, and CR are determined and stored in the contour data memory in the image analysis device 3. Outer and inner contour data are stored as column position values in the memory at addresses corresponding to row positions. The diaphragm contour data is stored as a row position value on the memory at an address corresponding to a column position. Note that a method may also be used in which sets of row and column values for each point of the obtained contour are sequentially stored.

As described above, in this embodiment, the chest X
Focusing on only one row or column of line image data, we identify points in that one-dimensional density data sequence where the relationship between the values of the preceding and succeeding data forms a specific predetermined pattern. The points of the contour are determined for a necessary range of rows or columns, and the line connecting these points is defined as the contour of the lung field.

The point that forms the above-mentioned specific pattern is a point that is within a predetermined concentration threshold and where the difference in value between two points separated by a predetermined distance from a certain point is maximum, or a predetermined concentration threshold. Minimal points within the range of , etc. can be used.

It is possible to prepare a plurality of such specific patterns and select one pattern from the plurality, or to select a specific pattern depending on the position of the row or column where the measurement is being performed, or depending on the target contour part. Prepare multiple candidates for contour points in a row or column, and determine the contour points according to the position of the candidate points, the position of the row or column where the measurement is being performed, or the target contour part. You can select one.

(Determination of Abnormality Based on Shape of Contour) By measuring the shape of the contour of the lung field, the following abnormalities can be detected.

- Hypertrophy of the heart: Judgment is made from the extent to which the inner contour lines BL and BR of the lung field protrude to the outside of the heart.

■Protrusion of the aorta: Judgment is made from the extent to which the upper part of the inner left lung field contour line BL protrudes outward.

■Diaphragm abnormality: Determined based on the smoothness of the diaphragm contour lines CL and CR.

Furthermore, by using the information on the contour of the lung field, it is possible to limit the search range for detecting abnormalities such as (1) abnormal shadows in the lung field, (2) abnormalities in the lung apex, and (2) curvature of the vertebrae.

(Display of Judgment Results) The judgment results obtained as described above are displayed in a form that can be judged by the doctor using this support device. here,
If there is an abnormality, the chest X-ray photograph number, data of the subject, the item with the abnormality, and the measured value are displayed on the CRT terminal used to instruct operations. Here, as a display method, a printing type terminal can be used. Display items can be changed.

In addition, it is possible to display the image data that is the subject of measurement on an image display terminal, and display the areas determined to be abnormal by adding color or increasing the brightness. It is also possible to generate a warning sound in the event of an abnormality. Further, in an apparatus having a mechanism for automatically transporting chest X-ray film, abnormality can be notified by discharging only the film showing an abnormality from a transport port different from that for normal film.

In addition, in the above explanation, we have described cases in which judgments are made by reading only the X-ray photograph, but in some cases, such as an X-ray TV or an In the case of using a device that converts the signal into a digital signal, this signal can be directly A/D converted and stored in the image storage device 2. Also X
It is also possible to use images of other modalities, such as tomograms such as CT and MRl, which are not limited to line photographs. Furthermore, it is also effective to perform image processing such as subtraction and use an image from which shadows of bones have been removed.

(Verification of Example) Using the apparatus configured as described above, we investigated the results of lung field contour extraction for 100 chest X-ray photographs provided by doctors. was evaluated by the doctor as having a good contour. 7
Examples include abnormalities in the density level of images, the influence of images of intestinal gas, and blurred images due to diseases. Furthermore, when cardiac hypertrophy and the like were measured using the obtained contour information, it was found that in 95% of abnormal cases, the measured values exceeded the standard value range, and results that could be determined as abnormal were obtained.

〔Effect of the invention〕

As described above, by using the present invention, it is possible to easily and accurately extract the contour line of a lung field portion, etc. from image data in which complicated parts overlap, such as a lung field image of a chest X-ray photograph. Furthermore, it is possible to perform image measurement on a conventionally performed chest X-ray photograph and display whether or not there is an abnormality in the subject before the doctor interprets the image.

Therefore, it is possible to closely observe only the photographs that are suspected of having an abnormality, and only confirm the photographs that are not suspicious. Therefore, it is possible to improve the efficiency of conventional X-ray photograph examinations, which are performed by carefully observing each X-ray photograph one by one.

In addition, even if there are variations in the finished density of the photograph depending on the subject and the conditions of photographing and processing, image processing makes it possible to measure under constant conditions, so even if the photograph is difficult to interpret with the naked eye, abnormalities can be detected. It will be possible to determine the presence or absence of the disease, reducing the chance of diseases being overlooked.

[Brief explanation of drawings]

FIG. 1 is a block diagram of an apparatus for implementing the method of the present invention, FIG. 2 is an explanatory diagram of representation of image data in a matrix, and FIG. 3 is a block diagram of a device for implementing the method of the present invention. FIG. 4 is an explanatory diagram for extracting points constituting the inner and outer contours of a lung field. FIG.

Claims (5)

[Claims] (1) Focus on only one row or column of chest radiation image data expressed in a matrix, and
Points where the relationship between the values of the preceding and succeeding data in a dimensional density data string forms a predetermined specific pattern are defined as contour points in that row or column, and the contour points are determined for the necessary range of rows or columns. An image measurement method characterized in that a line connecting those points is used as an outline of a lung field. (2) A patent characterized in that the point that forms the above-mentioned specific pattern is a point that is within a predetermined concentration threshold and where the difference in value between two points separated by a predetermined distance is maximum. An image measurement method according to claim 1. (3) The image measurement method according to claim 1, wherein the points forming the specific pattern are minimal points within a predetermined density threshold. (4) A plurality of the above-mentioned specific patterns are prepared, and one pattern is selected from the plurality according to the position of the row or column where the measurement is being performed, or according to the target contour part. Characteristic claim 1
The image measurement method described in Items 1 to 3. (5) Prepare a plurality of the above-mentioned specific patterns, find a plurality of candidates for contour points in a row or column, and determine the purpose according to the position of the candidate point or the position of the row or column where measurement is being performed. 4. The image measurement method according to claim 1, wherein one point of the contour is selected depending on the region of the contour to be measured.