JP2022090165A - Image processing device, image processing method and program - Google Patents

Abstract

To provide an image processing device, an image processing method and a program, capable of performing image processing for enhancing especially a line component of an interest object more than other parts, concerning a radiation image.SOLUTION: An image processing device 4 includes an image processing unit performing filter processing for enhancing a line component to a radiation image of a subject obtained by radiography, and processing for reducing a line component in an area other than an interest object Ar1 of the subject in the radiation image more than a line component of the interest object Ar1, an image generation unit for generating an output image Pd based on the radiation image and on a processing result by the image processing unit, and an image output unit for outputting the output image Pd.SELECTED DRAWING: Figure 4

Description

The present invention relates to an image processing apparatus, an image processing method and a program.

In recent years, there has been known a technique for sharply showing a diagnosis target site (object of interest) or the like reflected in an image by performing various image processing on a radiation image taken by a radiation image capturing apparatus.

For example, in Patent Document 1, the signal value of one image data is decomposed into a plurality of band limiting signal components having different frequency bands, the frequency and the response are associated with each other, and a plurality of tables having different response characteristics are used as presets. Store and convert multiple decomposed band-limited signal components based on the table corresponding to one of the presets, and reconstruct the converted multiple band-limited signal components into one emphasized image data. Disclosed is an image processing apparatus that generates a frequency-enhanced image by adding an enhanced image data multiplied by a predetermined enhancement coefficient to one image data.
With such a device, it is possible to create an image that emphasizes a fine structure by selecting a high-frequency preset.

Further, Patent Document 2 includes a two-dimensional histogram in which a lung nodule-enhanced image and a linear shadow-enhanced image are created from a chest X-ray image, and a lung nodule-enhanced image is taken on the p-axis and a linear shadow-enhanced image is taken on the q-axis. The hilar pulmonary vascular shadow is extracted as a linear shadow from the lung nodule-enhanced image using the extraction curve of 1, and the hilar lung is based on the lung nodule-enhanced image and the image obtained by applying the maximum filter to the lung nodule-enhanced image. By suppressing the brightness value of the vascular shadow, a linear shadow suppression image is created, and the false positive shadow is extracted using the two-dimensional histogram and the second extraction curve, and the total number of pixels of the false positive shadow is calculated in advance. An image processing method is disclosed in which a step of extracting false positive shadows is repeated until a determined threshold value is exceeded.
According to such a method, linear shadows in an image can be emphasized.

Japanese Unexamined Patent Publication No. 2018-166847 Japanese Unexamined Patent Publication No. 2017-18339

However, where various structures are reflected in the image, for example, as described in Patent Document 1, when the frequency enhancement process for emphasizing the high frequency component is performed, the fine structure is emphasized and becomes easy to see. The image lacks the contrast of each component.
It is preferable to emphasize the medium frequency component in order to add contrast, but there is a conflict between the emphasis of the high frequency component that can make fine structures such as trabecula look sharp and the emphasis of the medium frequency component that can increase the contrast. It was difficult to achieve both processing.

Further, unlike the case where the lung nodule is emphasized from the chest X-ray image assumed in Patent Document 2, various tissues (line components) appearing as linear shadows such as cartilage are present around the bone including the trabecula. exist. Therefore, if the line enhancement filter is simply applied, the line component corresponding to the object of interest such as the trabecula becomes clear, but the line component other than the object of interest is also emphasized, which becomes noise.

The present invention has been made in view of the above circumstances, and is an image processing apparatus, an image processing method, and an image processing method capable of performing image processing on a radiographic image in which the line component of interest is particularly emphasized more than other parts. The purpose is to provide a program.

The invention according to claim 1 is an image processing apparatus.
Filter processing for emphasizing the line component in the radiation image of the subject obtained by radiography, and reducing the line component in the region other than the subject of interest than the line component of the subject of interest in the radiation image. The image processing unit that performs processing,
An image generation unit that generates an output image based on the radiation image and the processing result of the image processing unit.
An image output unit that outputs the output image and
It is characterized by having.

The invention according to claim 7 is an image processing method.
Filter processing for emphasizing the line component in the radiation image of the subject obtained by radiography, and reducing the line component in the region other than the subject of interest than the line component of the subject of interest in the radiation image. The processing, the image processing process to perform, and
An image generation step of generating an output image based on the radiation image and the processing result in the image processing step,
The image output process for outputting the output image and
It is characterized by including.

The invention according to claim 8 is a program.
On the computer
Filter processing for emphasizing the line component in the radiation image of the subject obtained by radiography, and reducing the line component in the region other than the subject of interest than the line component of the subject of interest in the radiation image. Image processing function to perform processing,
An image generation function that generates an output image based on the radiation image and the processing result of the image processing function,
It is characterized by realizing.

By performing image processing as in the present invention, it is possible to perform image processing on a radiographic image in which the line component of interest is emphasized more than other parts.

It is a figure which shows the whole structure of the radiation imaging system in this embodiment. It is a block diagram of the main part of the image processing apparatus in this embodiment. It is a flowchart which shows the image processing in this embodiment. It is explanatory drawing which shows an example of the image processing in this embodiment schematically. (A) is a diagram showing an example of an original image in which a region of interest is captured, and (B) is a diagram showing an example of a line detection filter image in which a line detection filter is applied to the image shown in (A). (A) is a diagram showing an example of an original image in which a region of interest is captured, and (B) is a diagram showing an example of weighting for each portion. (A) is a diagram showing an example of an original image in which a region of interest is captured, and (B) is a diagram showing an example of weighting of each region recognized region. (A) and (B) are diagrams showing an example of an adjustment screen for adjusting a coefficient when a weighted filter image is applied to an original image. It is a flowchart which shows the image processing in one modification of this embodiment. It is explanatory drawing which shows typically the image processing in one modification of this embodiment. It is explanatory drawing which shows typically the image processing in one modification of this embodiment.

Hereinafter, an embodiment of an image processing apparatus, an image processing method, and a program according to the present invention will be described with reference to the drawings.

[Overall configuration of radiation imaging system]
FIG. 1 is a diagram showing an overall configuration of a radiation imaging system to which the image processing apparatus according to the present embodiment is applied.
As shown in FIG. 1, the radiation imaging system 100 includes a radiation generator 1, a radiation imaging device (hereinafter, simply referred to as “photographing device 2”), a console 3, an image processing device 4, and the like.

Each of the devices 1 to 4 is connected via a communication network NT such as a LAN (Local Area Network), a WAN (Wide Area Network), or the Internet, and can communicate with each other. Each device constituting the radiation imaging system 100 conforms to the DICOM (Digital Image and Communications in Medicine) standard, and it is preferable that communication between the devices is performed according to DICOM.
Further, the radiographic imaging system 100 is communicably connected to HIS (Hospital Information System), RIS (Radiology Information System), PACS (Picture Archiving and Communication System), etc. (not shown). You may.

The radiographic imaging system 100 may be installed in an imaging room (not shown), or a part or all of the radiographic imaging system 100 may be movable (for example, a system mounted on a round-trip car). May be good.
Note that FIG. 1 illustrates a case where each of the devices 1 to 4 is provided in the radiographic imaging system 100, but the configuration of the radiographic imaging system is not limited to the illustrated example. For example, a plurality of photographing devices 2 may be associated with a single or a plurality of consoles 3 via a communication network NT or the like. Further, a plurality of consoles 3 may be connected to the image processing device 5 via the communication network NT.

[Radiation generator]
The radiation generator 1 includes a radiation source (not shown) and a radiation irradiation control device.
The radiation source is arranged at a position facing the photographing device 2 with the subject (subject) in between, and irradiates the subject with radiation (X-rays) under the control of the radiation irradiation control device.
The radiation irradiation control device is connected to a console 3 (console for photographing), and controls a radiation source based on the radiation irradiation conditions input from the console 3 to perform radiography.

[Shooting device]
The photographing device 2 obtains a radiation image of a subject by radiography. Specifically, the photographing apparatus 2 generates digital data of a radiation image according to the radiation emitted from the radiation generating apparatus 1 and transmitted through a subject (not shown).
The photographing device 2 is, for example, an FPD (Flat Panel Detector) configured to be portable. The FPD may be used in a state of being set on a shooting table (not shown) corresponding to a standing position, a lying position, or the like, or may be used as a single panel.
The radiation image generated by the photographing apparatus 2 may be a still image or a moving image.
The imaging device 2 may be any as long as it can generate a radiographic image, and is not limited to the FPD. For example, CR (computed ragiography) or the like may be used.

[console]
The console 3 outputs radiation irradiation conditions and the like to the radiation generator 1 (radiation irradiation control device of the radiation generator 1) to control radiation imaging by the imaging device 2, and outputs image reading conditions and the like to the imaging device 2. It functions as a shooting console that controls the reading operation of radiographic images. The console 3 is a computer including a control unit (not shown) composed of a CPU (Central Processing Unit), a RAM (Random Access Memory), and the like.
Further, the console 3 of the present embodiment performs various corrections and the like on the radiation image (original image, original image) acquired by the photographing device 2.
The console 3 makes necessary corrections and the like, and then transmits the image data of the radiographic image to the image processing device 4.
It is not essential to perform various correction processes on the console 3. For example, the console 3 may be a console for photographing, and may merely give various instructions regarding photographing to the radiation generating device 1 and the photographing device 2.
In this case, the image processing device 4 makes various corrections and the like on the radiation image (original image, original image) acquired by the photographing device 2.

[Image processing device]
The image processing device 4 performs various image processing on the radiographic image of the subject obtained by radiography to generate an output image Pd in which the line component of interest such as the trabecula portion Ar1 is emphasized.
As shown in FIG. 2, the image processing device 4 includes a control unit 41, a storage unit 42, an operation unit 43, a display unit 44, a communication unit 45, and the like, and each unit is connected by a bus 46.

The control unit 41 is a computer composed of a CPU, RAM, and the like. The CPU of the control unit 41 reads out the system program and various processing programs stored in the storage unit 42 in response to the operation of the operation unit 43 and develops them in the work area of the RAM, and will be described later according to the expanded program. Various processes such as image processing are executed, and the operation of each part of the image processing device 4 is centrally controlled.
In the present embodiment, the control unit 41 functions as an image processing unit, an image generation unit, an image output unit, and the like.

The control unit 41 as an image processing unit performs filter processing for emphasizing the line component of the radiation image of the subject obtained by radiography, and the line component of the subject's object of interest in the radiation image other than the object of interest. The process of reducing the line component in the region of No. 1 is performed.
The control unit 41 as an image generation unit generates an output image Pd based on a radiation image and a processing result by the control unit 41 as an image processing unit.
The control unit 41 as an image output unit transmits the generated output image Pd to, for example, the display unit 43 and the storage unit 42 of the image processing device 4, various display units, storage units, printing units, etc. outside the device (not shown). Output.
The details of the functions of the control unit 41 as an image processing unit, an image generation unit, and an image output unit will be described later.

The storage unit 42 is composed of, for example, a non-volatile semiconductor memory, a hard disk (HDD: Hard Disk Drive), or the like. The storage unit 42 stores data such as parameters or processing results required for processing by various programs and programs including a program for executing various image processing by the control unit 41. These various programs are stored in the form of a readable program code, and the control unit 41 sequentially executes an operation according to the program code.

The operation unit 43 includes a keyboard equipped with cursor keys, number input keys, various function keys, and a pointing device such as a mouse, and controls an instruction signal input by key operation on the keyboard or mouse operation. Output to 41. Further, the operation unit 43 may include a touch panel on the display screen of the display unit 44, and in this case, the instruction signal input via the touch panel is output to the control unit 41.

The display unit 44 is composed of a monitor such as an LCD (Liquid Crystal Display) or a CRT (Cathode Ray Tube), and performs various displays according to instructions of a display signal input from the control unit 41.
In the present embodiment, the output image Pd is generated by image processing in the control unit 41 as described later. When the image data of the output image Pd is output from the control unit 41 to the display unit 44, the image based on the image data of the output image Pd is displayed on the display screen of the display unit 44.

The communication unit 45 includes a LAN adapter, a modem, a TA, and the like, and controls data transmission / reception with each device connected to the communication network NT.

[About the image processing method in this embodiment]
In the present embodiment, the image processing method is based on filter processing for emphasizing the line component of the radiation image of the subject obtained by radiography with the photographing device 2 or the like, and the line component of the subject of interest in the radiation image. An image processing step of reducing the line component in a region other than the object of interest, an image generation step of generating an output image based on a radiation image and a processing result in the image processing step, and an output image output. It includes an image output step to be performed.
Specific processing will be specifically described below with reference to FIGS. 3 to 11.

FIG. 3 is a flowchart showing an outline of image processing in the present embodiment. Further, FIG. 4 is an explanatory diagram schematically illustrating an outline of image processing.
As shown in FIG. 3, first, the control unit 41 of the image processing device 4 that functions as an image processing unit is subjected to radiographic imaging by various imaging devices 2 such as an FPD (in the present embodiment, the original image Po; FIG. When (see step 4) is acquired, a filter process for emphasizing the line component is performed on the obtained radiation image (original image Po) of the subject (step S1).

Specifically, a line detection filter (line enhancement filter) is applied to a radiation image (original image Po) to generate a line detection filter image Pf (see FIG. 4 and the like) in which a line component is emphasized.
The radiographic image (original image Po) is not limited to the one acquired by the radiographing apparatus 2 provided in the radiographic image capturing system 100. It may be stored in a database or the like for storing image data, or may be acquired by imaging in another radiographic imaging system 100.
Further, the radiation image (original image Po) referred to here may be image data that has not been subjected to any processing, or image data that has been subjected to preprocessing such as smoothing processing by various mean value filters, Gaussian filters, or the like. You may.

In the present embodiment, the filter processing by the control unit 41 as the image processing unit applies a line detection filter to the entire region of interest Ar0 in the radiation image (original image Po). The range to which the line detection filter is applied is not limited to the entire region of interest Ar0. It may be a narrow range narrowed down from the region of interest Ar0, or a wider range than the region of interest Ar0 may be targeted.
Here, the "region of interest Ar0" is a predetermined region including the "object of interest". Since the radiation generator 1 irradiates radiation targeting the region of interest Ar0, the region of interest Ar0 substantially coincides with the irradiation field region of radiation.
Further, the "object of interest" is, for example, a diagnosis target site or the like, and is a target portion for which a structure or the like is desired to be observed and analyzed in detail.

5 (A) is an example of a radiation image, and FIG. 5 (B) is an example of a line detection filter image in which a line detection filter is applied to the radiation image of FIG. 5 (A).
As shown in FIGS. 5 (A) and 5 (B), in the present embodiment, the radiographic image is an image of a certain area around the joint including the bone region Ar2, and is particularly bone in the bone region Ar2. An example will be described in which the trabecular portion Ar1 is the object of interest, and the bone region Ar2 including the trabecular portion Ar1 and the soft tissue region Ar3 (cartilage portion or the like) around the bone region Ar2 are designated as the region of interest Ar0.

As the line detection filter (line enhancement filter) applied to the radiographic image, for example, a Kasvand filter (for example, a Kasvand filter in four directions), a filter using a Hessian matrix, a VanderBrug filter, or the like can be preferably applied. The applicable line detection filter is not limited to these, and various filters can be used. The filter to be applied may be switched depending on the type of the radiographic image (for example, the part where the line component is to be emphasized).
The trabecula (trabecula portion Ar1), which is the "object of interest" in the present embodiment, is a mesh-like structural portion formed by the bone substance, which is found at the terminal portion of the bone portion and the like. By applying a line detection filter such as a Kasvand filter to the radiation image (original image Po) in which the trabecular part Ar1 is reflected, the line components in the image can be obtained as shown in FIGS. 4 and 5 (B). It is emphasized and the reticulated structure of the trabecula (trabecular part Ar1) becomes clear.

Next, the control unit 41 as an image processing unit weights the line detection filter image Pf (multiplies the weights) to reduce the noise portion, and generates the weighted filter image Pw (step S2).
That is, when the line detection filter is simply applied to the entire region of interest Ar0 in the radiographic image, as shown in FIG. The line component of Ar3 etc.) is also emphasized. Therefore, this becomes noise, and the trabecular part Ar1 is not clear. Therefore, weighting (multiplication of weights) is performed for each portion so that the line component of the trabecular part Ar1 which is the object of interest is particularly emphasized, and the “object of interest” of the subject in the radiographic image (the trabecular portion in the present embodiment). A process is performed to reduce the line component of a region other than the "object of interest" (for example, the soft region Ar3 in the present embodiment) rather than the line component of Ar1).
It should be noted that the weighting may be performed so as to reduce the line component of the region other than the "target of interest" (for example, the soft tissue region Ar3) rather than the line component of the "target of interest" (in the present embodiment, the trabecula portion Ar1). In addition to the weighting that weakens the line component of the region other than the "object of interest" (for example, the soft tissue region Ar3), the weighting that emphasizes the line component of the "object of interest" (the trabecular part Ar1 in the present embodiment) may be performed. good.

The procedure for generating the weighted filter image Pw is not limited to that exemplified here.
For example, the weighted filter image Pw may be generated by weighting (multiplying the weights) the radiation image (original image Po in the present embodiment) and then applying the line detection filter.
Further, even if the line detection filter is weighted (weight multiplication) and the weighted line detection filter is applied to the radiation image (original image Po in the present embodiment), the weighted filter image Pw is generated. good.

Weighting is performed by weights estimated (calculated) by learning (machine learning, deep learning) realized by a so-called convolutional neural network such as U-NET or Seg-NET.
In the weight learning, for example, the input is a "radiation image" (the "radiation image" may be the "original image Po" or the "processed image Pa" described later), and the output is a "weight", and a teacher image is given. Learn weights for noise removal (supervised learning).
The weight learning method is not particularly limited. Various learning (machine learning) methods (algorithms) can be used for weight learning.
By multiplying (multiplying) the weight estimated by learning on the line detection filter image Pf, the line component of the part to be emphasized is emphasized, and the weighted filter image Pw in which the emphasis of the line component of the other part is suppressed is obtained. Can be done.

The weight obtained by learning may differ depending on the type of object of interest. That is, the weights may be learned according to the part / part where the line component is to be emphasized.
For example, among the radiographic images (original image Po; see FIGS. 4, 5 (B), etc.) taken around the joint including the bone region Ar2 as the region of interest Ar0, we want to emphasize only the line component of the trabecular part Ar1. In this case, the weight that emphasizes the line component of the trabecular part Ar1 (object of interest) (for example, the weight of the trabecular portion Ar1 (object of interest) is set to "1", and the weight of the other parts is set to "0". Weighting) is learned.
Further, for example, the weighting level may be changed for each part or target, and learning may be performed so as to weight in a plurality of stages. Specifically, with respect to the radiographic image (original image Po; FIG. 6 (A)), as shown in FIG. 6 (B), the trabecular portion which is the tip portion of the bone region Ar2 in the region of interest Ar0. The weight of Ar1 (object of interest) is set to "1", the weight of the portion of the bone region Ar2 other than the trabecular part Ar1 (object of interest) is set to "0.5", and the weight of the region Ar0 of interest is the bone region Ar2. The weighting that the weight of the region other than the region (for example, the soft region Ar3 such as cartilage) is set to "0" is learned.

Further, when it is desired to emphasize other objects (for example, a catheter) reflected in the region of interest Ar0, the weights for emphasizing this are also learned. Specifically, the weight is learned by including an image including an object such as a catheter as a teacher image.
Further, the weight learned for each part may be switched depending on the type of the captured image (for example, normal imaging or long imaging). Specifically, for example, if you want to emphasize the line component only for the large epiphysis in long photography, you can emphasize the line component only for the large epiphysis above a certain level in the long radio image. Condition and learn weights.
In addition, multiple patterns of weights to be learned are prepared, for example, the weight of the first pattern is adopted in the case of the photographed image of the chest, and the weight of the second pattern is adopted in the case of the photographed image of the limbs. , A plurality of learning weight patterns may be provided according to the type of captured image so that they can be switched as appropriate.

If the radiation image includes an image of a portion outside the region of interest Ar0 (an image of the irradiation field region), the weight calculation of that portion may be omitted. As a result, unnecessary calculation processing can be omitted and weighting can be performed quickly.
The weight may be a weight calculated in a complex manner by considering the weight coefficient calculated by learning (machine learning) together with the feature amount of each pixel of the radiation image, the area recognition result, and the like.
Further, in the weight learning, the part, the dose information at the time of image capture, and the like may be included as learning parameters.
Further, it may be possible to perform a plurality of types of weight learning so that the user can appropriately select and switch the desired weight. For example, it is possible to learn a weight that emphasizes line emphasis although some noise remains, a weight that does not leave noise but weakens line emphasis, and the like so that the user can switch what kind of weighting is applied.

Further, the weighting is not limited to the case of acquisition by learning.
For example, region recognition may be performed on a radiation image, and weighting may be set for each recognized region. Specifically, with respect to the radiographic image (original image Po; FIG. 7 (A)), as shown in FIG. 7 (B), the region of interest Ar0 is the bone region Ar2 and the other portion (cartilage portion or the like). The region is recognized separately from the soft region Ar3), the weight of the bone region Ar2 is set to "1", and the weight of the portion other than the bone region Ar2 (soft region Ar3) is set to "0". Depending on the accuracy of the area recognition, the area may be further divided and the weight may be set for each area.
Further, as for the weighting, the adjustment amount of the weight may be changed depending on the feature amount.
For example, the weight may be determined based on the strength of the line detection filter, the standard deviation of pixels, the edge strength, and the like.
Further, the weighting may be adjusted by combining each method exemplified above (that is, weight learning by machine learning, weighting for each region by region recognition, adjustment by feature amount, etc.).
After smoothing the calculated weight, it may be applied (multiplied) to the line detection filter image Pf or the like.

When the weighting process is performed by the control unit 41 as the image processing unit and the weighted filter image Pw is generated, the control unit 41 as the image generation unit controls the radiation image (original image Po) and the image processing unit. The output image Pd is generated based on the processing result of the unit 41 (step S3).
Specifically, as shown in FIGS. 3 and 4, in the present embodiment, the control unit 41 as an image generation unit adds a weighted filter image Pw by a coefficient to the radiation image (original image Po). To generate an output image Pd.

The degree of "coefficient multiplication" in the addition process is set as appropriate.
The degree of "coefficient multiplication" may be adjusted depending on the type and site of the region of interest Ar0 and the object of interest (trabecula portion Ar1 in the present embodiment) to be particularly emphasized.
The adjustment may be automatically selected and set according to the characteristics of the part or the like, or may be manually adjusted by the user by the user interface (hereinafter referred to as “UI”).
As the UI, for example, the adjustment screen 442 including the slider bar 441 as shown in FIGS. 8 (A) and 8 (B) is displayed on the display unit 44 (see FIG. 2).

For example, in the example shown in FIG. 8A, by setting the line enhancement level to “0.1”, the enhancement of the line component of the output image Pd after the addition process is weakened. Further, in the example shown in FIG. 8B, by setting the line enhancement level to “0.5”, the enhancement of the line component of the output image Pd after the addition processing is stronger than in the example shown in FIG. 8A. It has become.
By making it possible to adjust the "coefficient" in the addition process on the adjustment screen 442, the user can appropriately check the image on the adjustment screen 442 and add the weighted filter image Pw to the radiation image (original image Po). Can be changed, and an output image Pd having a desired line enhancement level can be easily obtained.
The adjustment screen 442 shown in FIGS. 8A and 8B and the adjustment method of the “coefficient” using the slider bar 441 are examples, and various adjustment screens 442 and adjustment methods can be applied. ..

Then, when the output image Pd is generated, the control unit 41 as the image output unit outputs the output image Pd as appropriate (step S4).
The output destination of the output image Pd is not particularly limited. The control unit 41 as an image output unit can display, for example, the output image Pd on the display unit 44, output the output image Pd to a printing device or the like (not shown) via various I / Fs (not shown), or an internal storage unit 42. And output to various external storage means connected to the image processing device 4 for storage.

It should be noted that the "radiation image" to which the line detection filter or the like is applied up to this point is the "original image Po" (that is, the captured image without any processing or the captured image after smoothing processing, etc., FIGS. 3 and 3). Although the case of (see 4 etc.) has been described as an example, the “radiation image” is not limited to the “original image Po”.
As shown in FIGS. 9 and 10, the "radiation image" may be a "processed image Pa" obtained by subjecting the "original image Po" to frequency processing (frequency enhancement processing) or the like.
As the frequency processing (frequency enhancement processing) applied in the present embodiment, for example, there is a processing for emphasizing a high frequency component in a radiographic image. By emphasizing the high frequency component, it is possible to generate an image in which fine structural parts such as line components are emphasized.

In this case, for example, as shown in FIGS. 9 and 10, a filter process for emphasizing the line component is performed on the radiation image (original image Po) to generate a line detection filter image Pf (step S11). , This is weighted to generate a weighted filter image Pw (step S12).
Further, the radiation image (original image Po) is subjected to frequency processing to generate a processed image Pa (step S13).
Then, a weighted filter image Pw multiplied by a coefficient is added to the processed image Pa to generate an output image Pd (step S14), and then output is appropriately performed (step S15).

Further, the "radiation image" to be filtered and weighted for emphasizing the line component is not limited to the "original image Po".
For example, as shown in FIG. 11, a weighted filter image Pw may be generated by performing a filter process or a weighting process on the “processed image Pa” after the “original image Po” has been subjected to frequency processing or the like.

The processing applied to the radiation image (original image Po) to generate the processed image Pa is not limited to the frequency processing.
For example, the "radiation image" is a "original image Po" with dynamic range compression processing, frequency enhancement processing, gradation processing, scattered ray correction processing, noise suppression processing, streak correction processing, gain correction processing, offset correction processing, and defect correction processing. It may be a "processed image Pa" to which at least one of the above is applied.
The process applied to the "original image Po" is not limited to the one exemplified here.

As described above, in the present embodiment, the filter processing for emphasizing the line component is performed, and the weighting process for reducing the line component of the portion other than the object of interest is performed and added to the radiation image to generate the output image Pd.
As a result, the line component is emphasized only in the part to be emphasized, such as the trabecular part Ar1, and the other line components are suppressed (reduced), so that a high-definition output image Pd with less noise and good contrast can be obtained. can.

〔effect〕
As described above, according to the present embodiment, the control unit 41 of the image processing device 4, as an image processing unit, applies a line component to a radiation image (for example, "original image Po") of a subject obtained by radiography. Filter processing for emphasizing and reducing the line component of the region other than the subject of interest (for example, the soft region Ar3) from the line component of the subject of interest (for example, the bone beam portion Ar1) in the radio image. The control unit 41 as an image generation unit performs the process of causing the image to be generated, and the control unit 41 as the image generation unit is based on the radiation image (for example, “original image Po”) and the processing result by the image processing unit (“weighted filter image Pw” in the present embodiment). The "output image Pd" is generated, and the control unit 41 as an image output unit outputs the "output image Pd".

When a line detection filter (line enhancement filter) is simply applied to a radiographic image, both the line component of the object of interest (for example, trabecular part Ar1) to be emphasized and the line component of other parts (for example, soft region Ar3) are emphasized. As a result, an image that is not sharp and is not always suitable for observing and analyzing the object of interest is generated.
In particular, in the field of plastic surgery for the purpose of observing and analyzing joint parts, there are various structures appearing as line components in images such as cartilage parts in addition to bone parts around the joint parts. Therefore, for example, when it is desired to observe and analyze only the trabecular part Ar1 in detail, if the line component such as the cartilage part is also emphasized, this becomes noise and the trabecular part Ar1 becomes difficult to see.
In addition to filter processing, there is frequency processing (frequency enhancement processing) that emphasizes a specific frequency (high frequency component) as an image processing technique that makes it easier to see fine parts such as line components. It is necessary to emphasize the high frequency component in order to make the fine structural part easy to see, but it is necessary to emphasize the medium frequency component in order to improve the contrast of the image. However, since frequency processing can only emphasize a specific frequency, it is not possible to perform processing that achieves both enhancement of fine structural parts and improvement of contrast of an image, and only an image lacking sharpness can be obtained.

In this regard, when performing a line detection filter (line enhancement filter) process for emphasizing a line component and a weighting process for noise removal as in the present embodiment, the target of interest is the trabecular part Ar1 or the like. It is possible to generate an output image Pd in which the line component of an unnecessary portion is suppressed while emphasizing the line component of.
Therefore, the line component of interest such as the trabecula portion Ar1 can be expressed in detail, and the line component such as the cartilage portion is not conspicuous. That is, fine structures such as the trabecula portion Ar1 can be easily seen without noise, and a sharp, high-quality image can be obtained. This makes it possible to perform appropriate image diagnosis and the like using the output image Pd.
And, by improving the image quality, it is possible to save the time and effort of the shooting engineer to make various adjustments, etc., and it is possible to perform efficient shooting, so it is expected that the number of shots will increase and the efficiency of hospital management will be improved. can do.

Further, in the present embodiment, the control unit 41 as an image processing unit applies a line detection filter to the entire region of interest Ar0 in the radiographic image as a filtering process.
In this way, since image processing is performed on the necessary portion, wasteful processing time can be omitted and quick and appropriate image processing can be performed.

Further, in the present embodiment, the line detection filter may include a Kasband filter in four directions.
By applying the Kasband filter in four directions as the line detection filter, it is possible to detect and emphasize the line component of a very fine structural part such as the trabecular part Ar1 with high accuracy.

Further, in the present embodiment, the control unit 41 as an image processing unit performs a process of reducing the line component of the region other than the object of interest in the radiation image from the line component of the object of interest of the subject by the weight learned by the convolutional neural network. conduct.
In this way, in order to learn the weight for noise removal by advanced machine learning (DeepLearning), the part unnecessary for trabecular diagnosis such as the cartilage part while emphasizing the line component of interest such as the trabecular part Ar1. It is possible to obtain a sharp image with high accuracy, such as suppressing the line component of.

Further, the learned weight may be different depending on the type of the object of interest. When the weighting is set finely in this way, it is possible to generate a sharp image that more closely meets the demands of doctors and the like.

Further, the radiation image has at least dynamic range compression processing, frequency enhancement processing, gradation processing, scattered ray correction processing, noise suppression processing, streak correction processing, gain correction processing, offset correction processing, and defect correction processing on the original image Po. It may be the one to which either is applied.
By performing a filter process or a weighting process for emphasizing the line component on the radiation image after various processes (processed image Pa), it is possible to generate an image having higher sharpness.

[Modification example]
Although the embodiments of the present invention have been described above, it goes without saying that the present invention is not limited to such embodiments and can be variously modified without departing from the gist thereof.

For example, in the above embodiment, the case of a radiographic image in which the periphery of the joint portion or the like is set as the region of interest Ar0 and the trabecular bone portion Ar1 is the target of interest is exemplified, but the radiographic image to which the processing described in the present embodiment is applied is The image is not limited to such an image in the shaping field.
For example, a radiographic image obtained by photographing another part such as the chest may be subjected to the processing shown in the present embodiment to obtain an output image Pd.

Further, in the present embodiment, the case where the image processing device 4 is provided separately from the console 3 is illustrated, but the configuration of the radiation imaging system 100 is not limited to this.
For example, the console 3 may also serve as an image processing device, and the control unit of the console 3 may function as an image processing unit, an image generation unit, and an image output unit.

Further, in the present embodiment, the case where the control unit 41 as the image generation unit generates the output image Pd by adding the processing result (weighted filter image Pw) by the image processing unit to the radiation image (original image Po or the like) is exemplified. However, the output image Pd may be generated based on the radiation image (original image Po, etc.) and the processing result by the image processing unit, and is not necessarily the radiation image (original image Po, etc.) and the processing result by the image processing unit. It is not limited to the case where it is generated by the addition of.
For example, the control unit 41 as an image generation unit may generate an output image Pd by subtracting the processing result by the image processing unit from the radiation image (original image Po or the like).

Needless to say, the present invention is not limited to the above-described embodiments and modifications, and can be appropriately changed as long as the gist of the present invention is not deviated.

3 Console 4 Image processing device 41 Control unit 42 Storage unit 44 Display unit 100 Radiation imaging system Ar0 Area of interest Ar1 Trabecula (object of interest)
Ar2 Bone area Ar3 Soft area Po Original image Pf Line detection filter image Pw Weighted filter image Pa Processed image Pd Output image

Claims (8)

Filter processing for emphasizing the line component in the radiation image of the subject obtained by radiography, and reducing the line component in the region other than the subject of interest than the line component of the subject of interest in the radiation image. The image processing unit that performs processing,
An image generation unit that generates an output image based on the radiation image and the processing result of the image processing unit.
An image output unit that outputs the output image and
An image processing device characterized by comprising. The image processing apparatus according to claim 1, wherein the filter processing by the image processing unit applies a line detection filter to the entire region of interest in the radiographic image. The image processing apparatus according to claim 2, wherein the line detection filter includes a Kasband filter in four directions. The image processing unit is characterized in that the weight learned by the convolutional neural network is used to reduce the line component of a region other than the object of interest than the line component of the object of interest of the subject in the radiation image. The image processing apparatus according to any one of claims 1 to 3. The image processing apparatus according to claim 4, wherein the learned weight differs depending on the type of the object of interest. The radiation image has at least one of dynamic range compression processing, frequency enhancement processing, gradation processing, scattered ray correction processing, noise suppression processing, streak correction processing, gain correction processing, offset correction processing, and defect correction processing on the original image. The image processing apparatus according to any one of claims 1 to 5, wherein the image processing apparatus is provided. Filter processing for emphasizing the line component in the radiation image of the subject obtained by radiography, and reducing the line component in the region other than the subject of interest than the line component of the subject of interest in the radiation image. The processing, the image processing process to perform, and
An image generation step of generating an output image based on the radiation image and the processing result in the image processing step,
The image output process for outputting the output image and
An image processing method comprising. On the computer
Filter processing for emphasizing the line component in the radiation image of the subject obtained by radiography, and reducing the line component in the region other than the subject of interest than the line component of the subject of interest in the radiation image. Image processing function to perform processing,
An image generation function that generates an output image based on the radiation image and the processing result of the image processing function,
A program characterized by realizing.

Images