JP2923695B2 - Fundus image analyzer - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a fundus analyzing apparatus configured to extract, using digital image processing technology, minute light and dark unevenness in an image which cannot be generally recognized by the eyes. More specifically, it is formed so as to detect and evaluate the site of the defect of the retinal optic nerve fiber bundle that is closely related to glaucoma disease.

[0002]

2. Description of the Related Art There is a disease in the field of ophthalmology, which impairs the drainage mechanism of aqueous humor and increases intraocular pressure, thereby causing a depression of the optic disc, which is said to be the most vulnerable part in the eyeball. It is said that the retinal optic nerve fiber bundle connected to the optic nerve via the papilla is defective, and the defect is enlarged, the visual field is damaged, and it is said that the disease leads to blindness if left untreated. There is no established treatment method to treat the disease, and early detection of the disease, and control of the intraocular pressure at the time of the detection have been performed to prevent progression. Although efforts have been made, it is not a satisfactory test for the individual differences between subjects and the appearance of recently discovered low-tension glaucoma and the like, and it is urgently necessary to detect them early.

[0003] Generally, a bundle of retinal optic nerve fibers in the fundus occupies about one eye.
It is said that there are 600,000, and it is said that visual field impairment appears only after about half of the loss, and from the viewpoint of visual function test, establishment of the defect evaluation is desired, but the fiber bundle itself is transparent. Therefore, in everyday fundus observation medical treatment,
In direct inspection using an ophthalmoscope, etc., it is difficult to discriminate the difference between the fine reflected light from the normal fiber bundle and the defective fiber bundle. It is said. At present, it is difficult to capture objective quantitative changes because they are visually observed.

In order to objectively evaluate this, conventionally, the wavelength of 440 to 500 mm
It was recorded as a monochromatic black-and-white photo using a monochromatic light filter in the vicinity, and the aim was to extract the optic nerve fiber bundle loss part, but this method extracts it as a wedge-shaped dark part only if it can be discriminated even under direct vision However, at this point, there was a disadvantage that the visual field was obstructed and it was too late.

[0005]

The problem to be solved is that it is not possible to objectively extract the optic nerve fiber bundle defect before visual field defects appear.

[0006]

SUMMARY OF THE INVENTION The present invention focuses on a minute change in reflection from a normal portion and a defective portion of a retinal optic nerve fiber bundle, and realizes a technique for extracting a feature of the change.
Using this means, a device capable of performing a series of means of information inputting means, processing and processing means, and result display and evaluation means of a phenomenon of interest has been realized.

[0007]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention is an apparatus using a known video camera (fundus camera) and an analog / digital converter (AD conversion).

The outline will be described with reference to the block diagram of FIG. 1. First, a fundus image of the eye to be inspected is photographed using a known video camera. The red signal (R signal), green signal (G signal), and blue signal (B signal) from the video camera are digitally converted by an analog-to-digital converter, and the digital signal of each color image is stored in a frame memory.

The image stored in the frame memory is displayed on a monitor by operating the keyboard.

The computer (CPU) performs an image analysis described later in accordance with an operation from a keyboard.

The mouse is used to set the center of the nipple of the fundus image described later.

FIG. 3 shows a state in which a photographed image is displayed on a monitor screen. FIG. 3 shows a procedure of image analysis in this apparatus. The details will be described below with reference to the drawings.

First, a description will be given of a means for extracting a feature of a minute reflection change. As described above, the retinal optic nerve fiber bundle of interest has the property of having a reflection characteristic depending on the wavelength of light, and the optic nerve along the running of the retinal blood vessel 1 (see FIG. 3) emerging from the retinal optic disc of the fundus. Focusing on the denseness of the fiber bundles, the input color fundus image information is first decomposed into red, green, and red components. This method uses a 3CCD video camera, R, used for broadcasting stations.
It can be easily realized by using the G and B system video signals.

The video signals of the R, G, and B signals are converted from analog to digital and stored as digital image signals in the corresponding frame memories. That is, the image is recorded as a set of picture elements (pixels) 2 (see FIG. 4), and the degree of integration (memory capacity) determines the image quality and the degree of light / dark division of one picture element 2.

In general, when using a normal video system,
In the case of 512 scanning lines, 5 frames are used as the frame memory.
A memory having a capacity of 12 bits.times.512 bits.times.8 bits is easy to use, but other image scanners or the like can be used as an input means. Therefore, the memory capacity of the frame memory is not particularly limited.

A personal computer having a 16-bit or 32-bit computer which can normally obtain digital signal processing stored in a frame memory as digital image information of two-dimensional R, G, and B signals is used. Display color on top.

In order to extract the characteristics of the micro-reflection site, attention is paid to the above-mentioned morphological characteristics of the retina of the fundus, and the center 3 of the optic papilla is set as a concentric circle, and the periphery of the optic papilla is set as a small circle 4, which is further extended. Specify an arbitrary point along the optic fiber bundle running. Since this can be executed while looking at the image displayed on the monitor, it can be easily specified. It is possible to easily draw concentric circles by designating the center of the optic papilla using a mouse (specifically, by using a mouse grip) and by designating the diameters of the small circle 4 and the great circle 5, by creating software.

After the designation of the concentric circles, a radial line 6 centering on the center 3 of the optic disc is set at an arbitrary angle between the small circle 4 and the great circle 5 as shown in FIG. At this time, the diameter line 6 may or may not set the start point 7 and the end point 8. The digital information of the picture elements 2 (see FIG. 4) distributed along the radial line 6 is added (integrated) for each radial line 6, the designated angle of the radial line 6 is taken on the horizontal axis, and the R signal, A histogram (FIG. 5) is created for each of the G signal and B signal components.

Focusing on the envelope of this histogram (FIG. 5), the retinal retinal optic nerve fiber bundle shows a smooth change in a normal part, but there is a partial difference in brightness and darkness in the defective part, and a sharp partial drop is observed. In particular, as described above, the histogram based on the G signal component and the B signal component is remarkably extracted from the spectral reflection characteristics of the optic nerve fiber bundle, and it is possible to clarify a minute defect site.

In the color fundus image, retinal blood vessels 1
Is clearly recorded, and is displayed as an extreme decrease in the histogram at the angle portion overlapping the diameter line 6 on the histogram (FIG. 5), so that it is possible to easily discriminate.

As a display means for preventing the detection of a defective portion of the optic nerve fiber bundle from being overlooked, the continuity of the histogram for each continuous radial line is determined, and the calculation for extracting the discontinuous portion is easy. Coloring or flashing the histogram portion of a specific part can be realized by creating dedicated software.

Further, data is sampled along each optic nerve fiber bundle, the sum of the density of each point on the optic nerve fiber bundle is obtained, and the sum is compared to find a discontinuous portion of the envelope.

[0023]

According to the present invention, it is possible to extract a defective portion of the optic nerve fiber bundle in a short time and with high information with a simple structure of comparing the integral values centered on the nipple, and to prevent the visual field disorder from appearing. Early detection of glaucoma is possible.

[Brief description of the drawings]

FIG. 1 shows a block diagram of the present invention.

FIG. 2 shows a flowchart of the present invention.

FIG. 3 is an explanatory diagram showing a monitor image.

FIG. 4 is a schematic diagram for calculating a sum.

FIG. 5 shows a histogram.

[Explanation of symbols]

 Reference Signs List 1 retinal blood vessel 2 picture element 3 optic disc center 4 small circle 5 large circle 6 diameter line 7 start point 8 end point

Claims (1)

(57) [Claims] 1. The method of claim 1, wherein the brightness of an image on a plurality of meridians at predetermined angular intervals around the nipple of the fundus image is integrated, and a defect of the optic nerve fiber bundle in the fundus image is determined by comparing the plurality of radial lines. A fundus image analyzing apparatus characterized in that a unit is extracted.