Image analysis – Applications – Biomedical applications
Reexamination Certificate
1998-09-14
2003-07-15
Johns, Andrew W. (Department: 2621)
Image analysis
Applications
Biomedical applications
Reexamination Certificate
active
06594380
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an image discrimination apparatus and an image discrimination method for extracting the data for a feature from an image, and for employing this feature data to automatically identify the category of a subject in the image or the posture of the subject when radiographed; and in particular to an image discrimination apparatus and an image discrimination method for employing the data for a feature, such as the shape for a profile in an arbitrary area of an autoradiograph, to identify a portion or the posture of a subject, such as the front or the side of the lungs, when radiographed.
2. Related Background Art
Currently, in accordance with recent developments in digital techniques, an autoradiograph of a chest is digitized, image processing of the digital image is performed, and the resultant image is either displayed on a CRT or is printed.
In chest radiography it is common for the posture of a subject to differ in consonance with the purpose of the radiograph. For example, the radiographic posture will vary when a radiograph is taken from the front or from the side of the lungs, i.e., the portion to be radiographed (hereinafter referred to as a category) varies in accordance with the purpose. The distribution of densities differs depending on the posture, and the width of the density distribution in a target area is also varied. Therefore, in general, different image processing procedures should be employed for each radiographic posture in accordance with whether an image is to be displayed or is to be printed. Thus, for an image processing apparatus the radiographic posture must be input so that the apparatus will perform the image processing that is appropriate for the posture.
However, a great deal of manual labor is required when the radiographic posture is input each time image processing is performed, and input errors tend to accumulate.
To resolve this shortcoming, an automatic discrimination method is proposed whereby the data for the feature of an image is employed to automatically detect the posture of a subject. According to this method, a cumulative histogram for an image is prepared, and the value of a change rate substantially in the center of the cumulative histogram is employed to identify the category of the image.
With another automatic discrimination method, a specific amount of data for a feature is extracted from an image in order to easily understand the image data in an area to be examined, and the data for the feature, such as the maximum or the minimum value for the density of the body, is employed to determine an image processing condition.
To extract the data for a feature according to this method, as is shown in
FIG. 1
, for example, a density value histogram for an overall image (solid line A in
FIG. 1
) is prepared, and the maximum value and the minimum value are acquired from the shape of the histogram. Furthermore, to extract the data for a feature while a threshold value for a binary operation is being changed, the labeling process is performed to obtain an area, and for the area the maximum value and the minimum value are calculated under a specified condition.
When the conventional automatic discrimination method is used, however, a parameter for identifying the category of an image is fixed in advance, and the category of the image can not be identified unless a static condition prevails when it is exposed.
Furthermore, since when this method is used the category of an image is identified based on a change rate appearing substantially in the center portion of a cumulative histogram, deterioration of the accuracy of the discrimination process will occur when the interval between the maximum density value and the minimum density value changes.
In addition, with this method, when the exposure condition is changed, a parameter for identifying the category of an image must also be changed, and since a great deal of skill is required to perform this alteration, the parameter can not easily be changed.
Further, according to the conventional feature data extraction method, which was employed to prepare the density value histogram shown in FIG.
1
and to extract the data for a feature, the density value histogram in
FIG. 1
was formed by adding the distribution of the internal density value of a body (broken line a in
FIG. 1
) to the distribution of the density value (broken line b in
FIG. 1
) in a plain image area and in the internal area of a body that contacts the plain image area at a constant interval. A problem has arisen in that the maximum density value max, acquired from the shape of the histogram, differs from the actual maximum density value max
2
in the internal area of the body.
Moreover, since when using the conventional feature data extraction method the processing is performed sequentially by performing the labeling process at the same time as the threshold value for the binary operation is being changed, the processing is not performed efficiently and requires an extended period of time for its execution.
When one of the conventional feature data extraction methods that have such problems is employed as the automatic discrimination method, a natural result is that deterioration of the discrimination accuracy occurs.
SUMMARY OF THE INVENTION
It is, therefore, one objective of the present invention to provide an image discrimination apparatus and an image discrimination method that can resolve the above described shortcomings.
It is another objective of the present invention to provide an image discrimination apparatus and an image discrimination method for accurately and efficiently extracting the data for a feature of an image, and for automatically and precisely identifying the category of the image by using the extracted data for the feature.
To achieve the above objectives, according to one aspect of the present invention, an image discrimination apparatus, which extracts the data for a feature from an autoradiograph that is produced when a subject is irradiated with radioactive rays by a radioactive device and which employs the obtained data for the feature to identify the category of the subject in the autoradiograph, comprises:
area extraction means for entering an autoradiographic signal, for detecting a plain image area indicated by the autoradiographic signal, and for extracting an area other than the plain image area that is detected and an area, which is in contact with the plain image area, having a predetermined width;
maximum density value calculation means for calculating the maximum density value for one pixel of those pixels that constitute the area, and the coordinates of the location of the pixel in the autoradiograph;
feature data calculation means for calculating the data for the feature of the autoradiograph by employing the maximum density value for the pixel and the coordinates for the pixel in the autoradiograph, both of which are obtained by the maximum density value calculation means; and
discrimination means for employing the data for the feature obtained by the feature data calculation means to identify the category of the subject in the autoradiograph.
It is an additional objective of the present invention to provide an image discrimination apparatus and an image discrimination method for accurately and efficiently extracting the data for a feature for an image and for employing the data for the feature to precisely and automatically identify the posture of a subject in the image.
To achieve this additional objective, according to another aspect of the present invention, an image discrimination apparatus, which identifies the posture of a subject in an autoradiograph produced when the subject is radiated with radioactive rays by a radioactive device, comprises:
comparison means for entering an autoradiographic signal, for detecting a density value for a pixel that constitutes a first area, which corresponds to a first autoradiograph portion indicated by the autoradiographic signal, and a density value for a pixe
Canon Kabushiki Kaisha
Johns Andrew W.
Morgan & Finnegan , LLP
Nakhjavan Shervin
LandOfFree
Image discrimination apparatus and image discrimination method does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Image discrimination apparatus and image discrimination method, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image discrimination apparatus and image discrimination method will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3076226