JP5837965B2 - Image receiving apparatus, image transmitting apparatus, and control method thereof - Google Patents

Description

  The present invention relates to an image processing method and apparatus, and more particularly to an image processing method and apparatus suitable for processing X-ray medical images.

  When a certain type of phosphor is irradiated with radiation (X-rays, α-rays, β-rays, γ-rays, electron beams, ultraviolet rays, etc.), a part of this radiation energy is accumulated in the phosphor, and visible light is visible in this phosphor. It is known that when irradiated with excitation light such as phosphors, phosphors exhibit stimulating luminescence according to the stored energy, and phosphors exhibiting such properties are stimulable phosphors (stimulable phosphors). be called. A radiographic image information recording / reproducing system using this storage phosphor has been proposed by the present applicant (Japanese Patent Laid-Open Nos. 55-12429, 56-11395, etc.). According to this system, radiation image information of a subject such as a human body is temporarily recorded on a sheet of stimulable phosphor, and this stimulable phosphor sheet is scanned with excitation light such as laser light to generate stimulated emission light. The obtained photostimulated light is photoelectrically read to obtain an image signal, which is displayed on a recording material such as a photographic photosensitive material or a display device such as a CRT based on the image signal, and a radiation image of the subject is visualized. Get as.

  In recent years, apparatuses for taking X-ray images in the same manner using semiconductor sensors have been developed. These systems have the practical advantage of being able to record images over a very wide radiation exposure range compared to conventional radiographic systems using silver salt photography. That is, X-rays with a very wide dynamic range are read by a photoelectric conversion means and converted into an electrical signal, and this electrical signal is used to output to a recording material such as a photographic photosensitive material or a display device such as a CRT, thereby exposing to radiation. A radiographic image that is not affected by variation in the amount can be obtained as a visible image.

  Further, the electrical signal obtained as described above can be converted into digital information, and the obtained digital information can be stored in a storage medium such as a memory. It has also been proposed to provide digital information to such an information processing apparatus to perform digital image processing and provide various diagnostic support.

JP-A-55-12429 JP-A-56-11395

  However, X-ray images contain a great deal of information, and there is a problem that the amount of data becomes enormous when storing and transmitting the images. Therefore, when storing and transmitting images, high-efficiency coding that reduces the amount of data by removing the redundancy of the images or changing the contents of the images to such an extent that image quality degradation is difficult to visually recognize Is used.

  For example, in JPEG recommended by ISO and ITU-T as an international standard encoding system for still images, DPCM is adopted for lossless compression, and discrete cosine transform (DCT) is used for lossy compression. Yes. For details on JPEG, refer to Recommendation ITU-T Recommendation T.264. 81, ISO / IEC 10918-1, etc., and detailed description thereof is omitted here.

  In recent years, much research has been conducted on compression methods using discrete wavelet transform (DWT). A feature of the compression method using DWT is that the blocking artifact seen in discrete cosine transform (DCT) does not occur.

  On the other hand, when compressing an X-ray image, a region of interest (an important region) is set as a means for efficiently improving the compression rate, and the compression rate of the region of interest is lowered to give priority to image quality over other regions. Can be considered. Even when an image is encoded losslessly (LOSSLESS), by encoding the region of interest preferentially, it becomes possible to decode and display the region of interest prior to reading. However, where to set the region of interest in the image has a medical diagnostic meaning and is not easy.

  From such a situation, the applicant analyzes the input image at the time of image compression, extracts an X-ray irradiation field region, further extracts an X-ray blank region from the extracted irradiation field region, A method and apparatus that preferentially encodes a region of interest by encoding the image corresponding to the region of interest by level-shifting and encoding a portion obtained by removing the unexposed region from the region of interest is proposed. The problem with this method is that the ratio of the missing region in the cut-out image (irradiation field region) is about 20%, and it is necessary to further narrow the region of interest when improving the compression ratio.

  On the other hand, the region of interest setting is often effective when displaying the contents of a compressed image file. For example, considering that one or more images composed of 2000 × 2000 pixels are displayed on a 1000 × 1000 pixel monitor, it is necessary to select whether to display a part of the image or to display it in a reduced size. Of course, a method of scrolling an image with a mouse or a trackball is also conceivable, but this does not achieve the purpose of displaying a plurality of images simultaneously. When a region of interest is set in each image when filing an image, it is possible to display only the region of interest, but even if there is no region of interest setting or there is a region of interest setting, display allocation In some cases, the area may be exceeded, and a reduced display is required whether a part is displayed or the whole is displayed.

  Further, as a problem when the image is displayed in a reduced size, there is a possibility that detailed observation (differentiated from diagnosis in the sense of referring to the image after the primary diagnosis) cannot be performed due to poor spatial resolution. As a technique for compensating for this, it has been proposed that a rectangular or circular magnifying glass moving on the CRT is set, and an operator (physician) moves the magnifying glass as appropriate to partially display the original image. However, there remains a problem that it is possible to appropriately select an enlarged region in an image having a poor spatial resolution.

  An X-ray image may be encoded and stored in order to reduce the data capacity, and it is desired to display the encoded image quickly. The quick display of the encoded image can also be realized by preferentially displaying a region effective for diagnosis from the encoded image data.

  Therefore, an object of the present invention is to preferentially read an effective area for diagnosis from encoded data.

  Another object of the present invention is to read the region of interest important for diagnosis or observation determined by the diagnosis support means preferentially and improve the image quality of the region, thereby effectively improving the accuracy of the diagnosis support. Is to increase.

  Another object of the present invention is to make it possible to capture image data with a required image quality according to the diagnostic purpose of diagnostic support, and to realize efficient image transfer.

In order to achieve the above object, an image receiving apparatus according to the present invention comprises the following arrangement. That is,
An image receiving device for receiving an encoded data stream of an original image from a transmitting device,
Designating means for designating the resolution of the reduced image of the original image according to information for diagnostic purposes;
Receiving means for receiving a first data stream, which is a data stream corresponding to a reduced image of the original image, having a resolution specified by the specifying means;
After receiving the first data stream, the receiving unit receives a second data stream that is the data stream of the original image and is not included in the first data stream .

Further, a method for controlling an image receiving apparatus according to the present invention for achieving the above object is as follows:
A control method of an image receiving device for receiving an encoded data stream of an original image from a transmitting device,
A designation step for designating the resolution of the reduced image of the original image according to the diagnostic purpose information, by the designation means of the image receiving device;
A receiving step of receiving a first data stream, which is a data stream corresponding to a reduced image of the original image, having a resolution specified in the specifying step;
In the receiving step, after receiving the first data stream, a second data stream that is the data stream of the original image and is not included in the first data stream is received .

  According to the present invention, a region effective for diagnosis can be preferentially read from encoded data, and a necessary image can be quickly acquired.

  In addition, according to the present invention, it is possible to preferentially read a region of interest important for diagnosis or observation, which is determined by medical diagnosis support, and to improve the image quality of the region. Can be increased.

  Furthermore, according to the present invention, it is possible to capture image data with a required image quality according to the diagnosis purpose of diagnosis support, and it is possible to realize efficient image transfer.

It is a block diagram which shows the function structure of the image processing apparatus by this embodiment. It is a flowchart explaining the operation | movement outline | summary of the image processing apparatus by this embodiment. It is a flowchart explaining a shadow diagnosis assistance process. It is a block diagram which shows the internal structure of a shadow extraction part. It is a figure explaining the characteristic of a shadow extraction part. It is a figure which shows the neural network for determining whether the site | part by which the shadow extraction was positive is positive. It is a figure explaining the process of the above shadow diagnosis assistance part. It is a flowchart explaining a texture disease diagnosis support process. It is a block diagram which shows the function structure of a texture disease extraction process. 3 is a block diagram illustrating a detailed functional configuration of a segmentation unit 91. FIG. It is a figure explaining the function structure of the segmentation part in a learning phase. It is a figure explaining the functional structure of the segmentation part 91 in a utilization phase. It is a figure explaining segmentation. It is a figure which shows a specific example. It is the figure which showed the structure in the case of producing | generating a code sequence dynamically between the image input part 11 and the image server 151. FIG. It is the figure which showed the code sequence transmission at the time of dividing | segmenting an image into a tile and encoding. It is the figure which showed the code sequence transmission at the time of encoding an image by DWT. It is the figure which showed the mode of the layer structure at this time. It is a figure explaining the layer arrangement | positioning in coding data. 1 is a block diagram illustrating a schematic configuration of an image processing apparatus according to an embodiment. It is a figure explaining the relationship between a code block and a region of interest.

  Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings.

  In the following embodiments, when reading an encoded X-ray medical image from a storage medium, an image processing apparatus that automatically determines a region effective for diagnosis, preferentially reads and displays data related to the region. explain.

  FIG. 20 is a block diagram illustrating a schematic configuration of the image processing apparatus according to the present embodiment. Reference numeral 21 denotes a CPU, which implements various processes described below with reference to FIG. 1 and the like by executing a control program stored in the ROM 22 or a control program loaded from the external storage device 25 to the RAM 23. A ROM 22 stores a control program executed by the CPU 21 and various data. Reference numeral 23 denotes a RAM that provides a load area for the control program and a work area for the CPU 21.

  Reference numeral 24 denotes a display which performs various displays under the control of the CPU 21. An external storage device 25 is composed of a hard disk or the like, and stores a control program, image data, and the like. Reference numeral 26 denotes an input device that captures an operation input from the operator. A network interface 27 is communicably connected to the network 28. Reference numeral 20 denotes a bus, which connects the above components.

  Reference numeral 28 denotes a network such as a LAN or the Internet. A server 29 stores the encoded image data. Reference numeral 29 denotes an image server that provides a data stream of image data that has been series-converted and encoded.

  FIG. 1 is a block diagram illustrating a functional configuration of the image processing apparatus according to the present embodiment. In FIG. 1, an image input unit 11 is premised on input of encoded image data. In the following description, it is assumed that a DWT-converted image is converted into an encoded stream and input. However, the present invention is not limited to the DWT conversion and is not limited to a DCT or KL-converted image. It is also possible to consider application examples in which encoding is performed. In information source coding, DCT, KL conversion, DWT, and the like are collectively referred to as sequence conversion. Note that DCT and KL conversion are common techniques as described in “ANIL K. JAIN, Fundamental Of Digital Image Processing, 1989, Prentice-Hall Inc.”.

  The encoded image data is normally transferred from the image server (29) connected to the network, but the data input by the image input unit 11 is transferred to the decoding unit 12. As a decoding unit for decoding a DWT image, one according to ISO / IEC 15444 (hereinafter JPEG2000) is used, and details thereof are disclosed in JPEG2000 standard documents and the like. The image 13 that is the output of the decoding unit 12 is displayed on the display 24 by the image display unit 15 and is provided to the diagnosis support unit 16. Of course, the image support unit 16 may be configured to capture the image 13.

  It should be noted that the initial image used for the diagnosis support unit 16 may have the minimum image quality necessary for the target diagnosis support, and it is necessary to read all the data streams converted for the diagnosis support. It is not. The target image quality necessary for diagnosis support, that is, the layer configuration at the time of encoding, is set in the initial stream setting unit.

  In the following description, detection support for tumor shadows in the chest front image is shown as an example, but a relatively low-resolution reduced image is sufficient for this diagnosis support. Accordingly, the initial stream setting unit 18 is set so that data corresponding to the target image is partially transferred as the initial image. As a precondition, there are a plurality of types of diagnosis support, and the required image quality is determined by the type of diagnosis support. For example, in the case of a tumor shadow, since a relatively large target is detected, fine resolution is not necessary, and an image quality with a spatial resolution of 1/5 and a density resolution of about 1/2 is set as the initial image. On the other hand, in the case of a ground glass-like shadow, since a relatively fine resolution is required, an image quality having a spatial resolution of 1/2 and a density resolution of about 1/3 is set as an initial image. These initial setting values are strongly related to parameters set in the diagnosis support algorithm and are determined empirically. When an image satisfying the conditions set in this way is input, the diagnosis support unit 16 starts a diagnosis support process. When a further data stream is required in the diagnosis support process, data in a necessary area is input via the input control unit 14. The positive region 17 determined by the diagnosis support unit 16 is displayed as an overlay on the reduced image displayed on the image display unit 15 as contour information, and is provided to the input control unit 14.

  The above operations are summarized as follows. FIG. 2 is a flowchart for explaining the outline of the operation of the image processing apparatus according to this embodiment. First, in step S21, the initial stream setting unit 18 sets an initial stream suitable for target diagnosis support. In step S <b> 22, the image input unit 11 captures encoded image data and provides it to the decoding unit 12. At this time, the input control unit 14 controls the image input unit 11 so that image data is input according to the contents set by the initial stream setting unit 18.

  Subsequently, in step S23, the encoded image data input from the image input unit 11 is decoded, and the obtained image data 13 is displayed by the image display unit 15. In step S24, the diagnosis support unit 16 analyzes the decoded image data 13 to detect a diseased part, and outputs positive region data 17. In step S25, the input control unit 14 sets a region to be preferentially captured in the encoded image input in step S21 based on the positive region data 17 output from the diagnosis support unit 16. If it is determined in step S26 that the image data has not been captured, the process returns to step S22. Then, the input control unit 14 controls the image input unit 11 to capture the encoded image according to the set area. The encoded images thus input from the image input unit 11 are sequentially decoded by the decoding unit 12 and displayed on the image display unit 15. Further, the progress of taking in the encoded image data contributes to the improvement of the accuracy of the diagnosis support process in step S24 and the refinement of the image display in step S23.

  By operating as described above, an area effective for diagnosis is automatically selected from the encoded data, and the area is read preferentially, so that information effective for diagnosis can be displayed with priority. It becomes possible. Since information effective for diagnosis is preferentially displayed in this way, the same effect as displaying an X-ray image quickly can be obtained.

  Next, the diagnosis support unit 16 will be described. The diagnosis support unit 16 requires different processing depending on the purpose of diagnosis support, but can be broadly classified into shadow diagnosis processing and texture disease diagnosis processing. Shadow diagnosis processing is a process used to detect calcification in breast images, tumors in the form of tumors, or tumors in mammograms (MASS). On the other hand, the texture disease diagnosis process is used to support diagnosis of interstitial pneumonia found in a chest image.

  First, shadow diagnosis support will be described with reference to FIGS.

  FIG. 3 is a flowchart for explaining the shadow diagnosis support process. First, in step S31, the decoded image data 13 is input to the shadow extraction unit, and shadow extraction is performed.

  FIG. 4 is a block diagram showing the internal configuration of the shadow extraction unit. A high-frequency image and a low-frequency image are created from the input image data 13 by the high-pass filter 41 and the low-pass filter unit 42. The computing unit 43 takes the difference between these two images and creates shadow candidate image data 44. Examples of characteristics of the high-pass filter 41 and the low-pass filter 42 are shown in FIGS. Further, the characteristic formed by the difference between the high-frequency image and the low-frequency image is shown in FIG. 5C, which can be considered as a matched filter for detecting a circular pattern of a certain size from the image (FIG. 5). As shown in FIG. 5C, it can be understood that a filter that extracts only a region corresponding to a certain frequency in an image effectively extracts only a certain size in the image).

  As can be seen from the above description, the size of the circular pattern to be extracted can be adjusted by adjusting the characteristics of the high-pass filter 41 and the low-pass filter 42 in the shadow extraction process. Generally, in the case of a chest front image, it is very difficult to select and extract a tumor having a diameter of 5 mm or less from the image. The reason is that the image contains many signals of the same size, and many of them are not sick signals. Therefore, a matched filter having a diameter of about 10 mm is assumed here. However, there is no limitation that the diameter of the circular pattern to be extracted by the matched filter is one type, and a plurality of types of circular patterns may be used. For example, three types of filters having a circular pattern diameter of 8 mm, 12 mm, and 16 mm may be sequentially applied, and the shadow candidates extracted by the respective filters may be passed to the subsequent processes.

Next, in step S32, a feature amount is calculated by the pathological feature amount extraction process for the shadow candidates extracted in the above-described shadow extraction process (S31) (for each of the shadow candidates). The feature quantities calculated here are area, circularity, and threshold sensitivity. Each feature amount will be described below. However, it is necessary to binarize the shadow candidates when calculating the area and the circularity. The binarization threshold is empirically set to a 10% value of the histogram of the matched filter output image.
(1) Area S = [number of pixels included in shadow candidate] × [area of one pixel]
(2) Circularity C = A / S: Here, A is an area where the circle and the shadow candidate overlap (overlap) when a circle having an effective diameter D is arranged at the center of gravity of the shadow candidate. (3) Threshold sensitivity = | 2 × S10−S5−S15 |: The threshold sensitivity is the change in the shaded area when the threshold is changed. S5, S10, and S15 change the threshold to 5%, 10%, and 15%, respectively. "||" represents an absolute value of the area when measured.

  In step S33, based on the feature amount calculated as described above, it is determined whether each shadow extracted in step S31 is positive or false positive. Here, a false positive is a shadow that is not positive. In the present embodiment, the above-described three types of feature quantities are used, but are not limited to these three types.

  Note that the determination in step S33 is performed using a determination unit configured by a neural network. Hereinafter, the determination unit will be described separately for a learning phase and a usage phase. In this embodiment, a neural network is used for the determination unit. However, the present invention is not limited to this, and there are cases where linear separation is possible depending on the purpose of diagnosis support. A low linear separation may be used.

  In the learning phase, an internal coefficient of a neural network is learned by inputting a feature amount while presenting the result for a shadow whose result is positive or false positive. Various neural networks have been developed. An example is Feed Forward type error back-propagation neural network developed by Rammelhart (reference: DE Rumelhart and JL McCelland, Parallel Distributed Processing: Explorations in the Microstructure of Cognition, Vol. 1: Foundation. Cambridge: The MIT Press, 1986.), Radial Basis Function Neural Network (simply RBF-NN) (reference: C. Bishop, "Improving the Generalization Properties of Radial Basis Function Neural Networks," Neural Comp., Vol. 3, pp. 579 -588, 1991.).

  In this embodiment, a 3-input RBF-NN as shown in FIG. 6 is used. Structurally, it adopts a Feed Forward type and has a three-layer structure of an input layer, an intermediate layer, and an output layer. In the input layer, input nodes corresponding to the number of feature amounts extracted in the pathological feature amount extraction process in step S32 are arranged. The RBF neuron arranged in the intermediate layer has an output characteristic that has a Gaussian distribution as a non-linear element. The number of RBF neurons depends on the number of learning cases and the complexity of the problem, but about 100 is appropriate for setting the calculation time appropriately. The number of outputs of the neural network is 1. If it is positive, 1 is output, and if it is negative, 0 is output. In this embodiment, a neural network is prepared for each purpose of diagnosis support. However, since the neural network is configured by software, in practice, a coefficient for each diagnosis support is stored and used for each diagnosis support. Will be set.

In the use phase, after setting an internal coefficient learned in accordance with the purpose of diagnosis in a neural network, three types of feature amounts for a shadow whose determination result is unknown are presented to obtain a network output. If learning is performed with a positive output of 1 and a negative output of 0, a positive determination is output when the output is 0.5 or more, and a negative determination is output when the output is 0.5 or less. Each of the circular patterns (shadows) determined to be positive is a positive area determination process (step S34).
Used for.

  The purpose of the positive area determination process in step S34 is to make one positive area when a plurality of shadow patterns determined to be positive are close to each other, and to make the positive area rectangular in order to specify the encoded area It is to be. However, the restriction to limit the designation of a positive region to a rectangular shape is not essential, and in JPEG2000, a circular region of interest (positive region) can be set by using MAXSHIFT.

  FIG. 7 is a diagram for explaining the process of the above-described shadow diagnosis support process. FIG. 7A is a chest front image input as image data 513, and FIG. 7B shows the result of applying the shadow extraction process of step S31 to this image. Next, the pathological feature amount extraction processing in step S32 extracts a pathological feature amount for each of the four shadow candidates shown in (B). FIG. 7C shows the result of determining whether or not each shadow is positive in the neural network based on the extracted feature amount in step S33. In (C), the rectangular shadow pattern of the right lung shown in (A) is excluded as negative. Furthermore, in the positive area determination process in step S34, (D) is the result of determining the positive area for the positive shadow pattern in (C). Thus, the positive region is set so as to include all the positive shadows, and the positive region is set by integrating the shadows of the left lung. As a condition for integration, if there are other positive regions within a predetermined range from a certain positive region, they are finally handled as one positive region. As an example of the area integration algorithm, after performing a predetermined number of morphological dilations on the image of FIG. 7C, the number of individual areas for labeling is counted, and the number of individual areas decreases. If so, it is determined that joining of regions has occurred. Then, the ROI (region of interest) is determined so as to include all the original positive regions that are combined and included in the region, and FIG. 7D is output.

  Next, texture disease diagnosis support will be described with reference to FIGS. However, in the following description, an explanation will be given taking an example of interstitial lung disease for a chest front image.

  FIG. 8 is a flowchart for explaining texture disease diagnosis support processing. In FIG. 8, a texture disease extraction process is applied to the image data 13 output from the decoding unit 12 in step S81, and a texture disease candidate region of interest (ROI) is extracted. FIG. 9 is a block diagram showing a functional configuration of the texture disease extraction process. The image data 13 as an input image is input to the segmentation unit 91 and the lung field region is segmented. Interstitial lung disease is a disease in which the mediastinal region is excluded from the target and the search for the shadow pattern is different from the local region, so in order to define the local region of the lung field Lung field extraction is required.

  The segmentation unit 91 will be further described with reference to FIG. FIG. 10 is a block diagram showing a detailed functional configuration of the segmentation unit 91. Reference numeral 101 denotes a segment feature amount extraction unit that calculates a feature amount for each pixel of the image data 13. Reference numeral 102 denotes a determination unit that determines a disease candidate region using a neural network.

  The configuration of the segmentation unit 91 differs depending on the learning phase and the use phase. Since the segmentation unit 91 of this embodiment is configured by a neural network, a phase in which learning data is input to form a coefficient inside the neural network is called a learning phase, and segmentation is performed on the input image data. The phase is called the usage phase. In the following phases, segmentation is performed in units of pixels. However, the segmentation method is not limited to that in units of pixels, and segmentation can also be performed by a method of tracking an outline from an image. Contour tracking type segmentation is described by O. Tsujii, MT Freedman, and SK Mun, "Lung contour detection in chest radiographs using 1-D convolution neural networks," Electronic Imaging, 8 (1), pp. 46. -53, January 1999.

  FIG. 11 is a diagram illustrating the functional configuration of the segmentation unit in the learning phase. The feature amount is calculated by the segment feature extraction unit 101 for each pixel of the input image. The calculated feature amount includes a calculation based on a pixel value, a calculation based on a texture, a calculation based on a relative address from an anatomical structure, and the like. For details, see the paper by the present inventor (O. Tsujii, MT Freedman, and SK Mun, "Automated Segmentation of Anatomic Regions in Chest Radiographs using Adaptive-Sized Hybrid Neural Network," Med. Phys., 25 (6), pp. 998-1007, June 1998.).

  The feature quantities used are not limited to those shown in the above examples and papers, but other papers MF McNitt-Gray, HK Huang and JW Sayre, "Feature Selection in the Pattern Classification Problem of Digital Probabilistic features described in Chest Radiograph Segmentation, "IEEE Trans. Med. Imag., 14, 537-547 (1995)."

  Next, the determination unit (hereinafter also referred to as a neural network unit) 102 will be described. Various neural networks 102 have been developed as described above. In this embodiment, RBF-NN is used. Structurally, a Feed Forward type is adopted, and a three-layer structure of an input layer, an intermediate layer, and an output layer as shown in FIG. However, for convenience, the number of input NODEs is 3, the number of intermediate neurons is 4, and the number of output NODEs is 1.

  Input nodes corresponding to the number of feature amounts extracted by the segment feature amount extraction unit 101 are arranged in the input layer. The RBF neuron arranged in the intermediate layer has an output characteristic that has a Gaussian distribution as a non-linear element. The number of RBF neurons depends on the number of learning cases and the complexity of the problem, but about 5000 is appropriate for setting the calculation time appropriately. The number of outputs of the neural network corresponds to the number of segments for each part. For example, in the case of an image of the front of the chest as shown in FIG. 13A, since there are two anatomical segments as shown in FIG. 13B, two output NODEs are prepared.

  In this embodiment, a neural network is prepared for each imaging region (for each target image). However, since the neural network is configured by software, the coefficients for each region are actually stored and used. In this case, the coefficient for each part is set.

  In the learning phase, an answer example (teacher output image), which is a segmentation result, is presented to the neural network 102 together with the feature amount for each pixel. The teacher output image for the input image is provided from the manual segmentation processing unit 111. A teacher image corresponding to the input chest image shown in FIG. 13A is shown in FIG. In the segmentation processing unit 111, teacher images are created so as to have output values 1 and 2 so as to correspond to the segment classification corresponding to the chest front image. However, other numbers, for example, zero are set for areas that are not targeted by the segment. Creation of the teacher image by the segmentation processing unit 111 can be performed by graphic software that can specify an area in the image, but it is also possible to use commercially available Photoshop (manufactured by Adobe Corporation in the United States) or the like. In the case of RBF-NN, learning of the neural network can be obtained analytically by performing the least square method that minimizes the output error, and the calculation of the internal coefficient can be performed in a short time.

  Next, the usage phase will be described. FIG. 12 is a diagram illustrating the functional configuration of the segmentation unit 91 in the use phase.

  The feature amount extraction for each pixel in the image by the segment feature amount extraction unit 101 is performed in the same manner as in the learning phase. As the internal coefficient for the neural network 102, the coefficient for the target imaging region is loaded before use. The feature quantity for each pixel is presented to the neural network, and the network outputs an output value to the output NODE. In the case of the chest front image, there are two output NODEs, and the input pixels are classified into segments corresponding to the NODEs that output an output closest to YES among these output NODEs. The coefficient loaded in the above is a coefficient that is learned for each imaging region in the neural network and is formed inside the neural network as a result of learning. These coefficients are originally converged in the learning process in accordance with a specific imaging region and diagnosis support, starting from an appropriate initial value. This converged coefficient determines the behavior of the neural network and is usually composed of a plurality of coefficients. The number of coefficients depends on the type and size of the neural network to be used.

  FIG. 13C shows an example of a segment image in which the classification for all the pixels in the image is completed in this way. In general, the segment image output from the neural network unit 102 includes noise. For example, a lung field region having a small area may be generated separately from the lung field region. Such noisy small regions are removed in the post-processing process, but this technique is described in the above-mentioned document: O. Tsujii, MT Freedman, and SK Mun, "Lung contour detection in chest radiographs using 1-D convolution. See neural networks, "Electronic Imaging, 8 (1), pp. 46-53, January 1999.

  The output of the discrimination unit (neural network unit) 102 in FIG. 10 is a segmentation output image 103. In the case of the chest front image, it is the lung field portion, and an example thereof is shown in FIG.

  Next, an area division process 92 (FIG. 9) is applied to the segmentation output image 103. The area dividing process is a process of defining a local area for calculating the texture disease parameter. A specific example is shown in FIG. FIG. 14A shows a chest front image, and the segmentation unit 91 that has input the image as image data 13 outputs the image shown in FIG. 14B as a segmentation output image 103. When the region division processing 92 is applied to the segmentation output image 103 in (B), an image divided into regions as shown in FIG. 14C is obtained. That is, the dividing process is a process of dividing the target region (in this case, the lung field in this case) of the segmentation output image into a rectangular shape so as to have an area of a certain value or less. The roughly rectangular region of interest defined in this way is called a texture disease candidate ROI (93).

  As described above, the output of the texture disease extraction process in step 81 is the texture disease candidate ROI 93. In step S82, pathological feature amounts are extracted (calculation of pathological feature amounts) for each of the plurality of texture disease candidates ROI 93, and it is determined whether each candidate is positive or negative based on the obtained feature amounts. .

The process after the feature amount calculation is the same as the shadow diagnosis support process (FIG. 3). That is, a rectangular positive region including the texture disease candidate ROI determined as positive (the hatched region in FIG. 14C) is defined (S83, S84), and this positive region is output (S85). . However, the feature amount calculation is performed according to the purpose of diagnosis support. Regarding features used for texture disease diagnosis support,
(1) Shigehiko Katsurakawa, et al .: Possibility of computer diagnosis support for interstitial disease, Journal of Japanese Medical Association y-ray Society, 50: 753-766, 1990.
(2) Yasuo Sasaki, et al .: Quantitative evaluation by texture analysis of pneumoconiosis standard photographs, Japan Medical Front Lane Association, 52: 1385-1393, 1992.
And the like.

  As described above, when the diagnosis support unit 16 outputs the positive region 17, this output is input to the input control unit 14 and used for image input control by the image input unit 11. The purpose of the input control unit 14 is to read data for the positive region 17 (hereinafter also referred to as a region of interest), which improves the accuracy of diagnosis support by making the image of the positive region detailed. Contributes to image display. When the accuracy of diagnosis support is improved, the diagnosis support unit 16 is applied again based on the input image with improved accuracy. Hereinafter, the input control unit 14 will be described.

  The input control unit 14 designates the image server to preferentially transfer data by designating a tile area or code block area for the region of interest. A request for selective data to the image server is made via the image input unit 11, but depending on the configuration of the image server, all the codes of LOSSLESS are stored and moved in response to a request from the image input unit 11. It is possible to construct a stream such as a layer dynamically.

  FIG. 15 is a diagram illustrating a configuration when a code string is dynamically generated between the image input unit 11 and the image server 151. In the figure, the image input unit 11 transmits the position of the region of interest in the image and a transmission request for the portion to the image server 151. The image server 151 transmits a necessary part from the code string encoded in advance and stored in the external storage device 152 to the image input unit 11. Hereinafter, an example of the output form of the code string related to the region of interest in this configuration will be described.

  FIG. 16 is a diagram showing code string transmission when an image is divided into tiles and encoded. In the figure, an image is divided into tiles that are rectangular areas, and each tile is independently encoded and stored in the external storage device 152. A transmission request from the image input unit 11 is input to the parser 161 in the image server 151. The parser 161 reads out the code sequence of the tile related to the region from the input position information of the region of interest from the external storage device 152 and outputs it.

  Here, the code sequence of each tile is composed of a plurality of layers (10 layers from Layer 0 to Layer 9 in FIG. 16) as described above, and the code sequence of the output tile is encoded data of a necessary layer. Or all layers. The output code strings are combined into one, and a marker code representing the start of the tile and the tile number is inserted at the head of the encoded data of each tile.

  In FIG. 16, when no code string is transmitted before that time, the input control unit 14 controls the image input unit 11 to input a data stream according to the information amount set by the initial stream setting unit 18. Thereby, the image input unit 11 issues a transmission request to the parser 161 so as to input layers 9 and 8 of all tiles, for example. The parser 161 collects the encoded data layers 9 and 8 of each tile from the tile 1 to the tile 16 to form a code string and outputs it. Subsequently, if the requested region (region of interest) is the tile 5 as a result of the analysis by the diagnosis support unit 16, the input control unit 14 controls the image input unit 11 to input the tile 5 preferentially. As a result, the image input unit 11 further makes a transmission request for the tile 5. Receiving this, the parser 161 collects all the remaining layers (layers 7 to 0) of the tile 5 to form a code string and outputs it. In this way, all the layers are transmitted for the tile 5 designated as the required area, and the minimum necessary layers are transmitted for the other tiles. In this way, if the encoding is performed reversibly, the decoded image is reversible with respect to the portion of the tile 5 that is the region of interest (the same image quality as the original), and the other portions are observed. Therefore, the image is reproduced with a sufficient image quality.

  On the other hand, it is possible to adopt a different code string construction method, which will be described below with reference to FIG. FIG. 17 is a diagram showing code string transmission when an image is encoded by DWT. In JPEG2000, subband coefficients or quantization indexes generated by DWT are bit-plane encoded in units of code blocks. In this example, the encoded data of each code block is stored in the external storage device 152 in a form that is not configured as a code string.

  Here, in the encoded data of each code block stored in the external storage device 152 shown in FIG. 17, the encoded data portion set by the initial stream setting unit 18 according to the diagnostic purpose has already been transmitted. ing. For example, data corresponding to a predetermined bit plane or layer from the top is already transmitted by the initial stream setting unit 18, and the diagnosis support unit 16 performs processing based on an image obtained by decoding the transmitted encoded data. It shall be.

  Next, when the region of interest is output by the diagnosis support unit 16 as shown in FIG. 21A, the input control unit 14 specifies a code block including the region of interest as shown in FIG. A transmission request is output to the parser 171 so as to further transmit the encoded data of these code blocks.

  In response to a transmission request from the image input unit 11, the parser 171 configures and outputs the encoded data of the code block related to the region of interest as a code string. At this time, for example, as shown in FIG. 21, for the selected code block, all the encoded data is transmitted, and decoded together with the encoded data that has been transmitted previously, Can compose images with high image quality.

  FIG. 18 shows an example of the layer structure in the JPEG2000 code string. In the above example, in the layer 9 of the code string shown in FIG. This is the same as transmitting only data in a new code string. Alternatively, it is possible to obtain an image in which the image quality of the region of interest and its peripheral part is improved by forming an additional layer composed of code blocks around the region of interest and transmitting it simultaneously.

  In this way, the parser 171 generates and outputs a code string in which the encoded data of the code block related to the region of interest is arranged in the upper layer as shown in FIG. 18 in response to a transmission request from the image input unit 11. In this way, the encoded data in the intermediate state before the code string configuration is stored in the external storage device 152, and the optimum code string is dynamically generated in response to a request from the outside, so that the tile division described above is not performed. Even in this case, it is possible to generate and output a code string having a region of interest with high image quality in response to a request on the image display side.

  In addition, supplementary explanation will be added by taking encoding as an example regarding the layer in JPEG2000. In JPEG2000, the region of interest can be preferentially encoded by a hierarchical encoded data structure called a layer. FIG. 18 is a diagram showing the state of the layer configuration at this time. In the figure, code blocks 3 and 4 are included in the region of interest. Based on the input information of the region of interest, the code string configuration unit 5 configures a code string by including only the code of the code block included in the region of interest in Layer 9.

  Here, when the progressive form is Layer-Resolution level-Component-Position in JPEG2000, the arrangement of layers in the entire code string is as shown in FIG. 19, and the encoded data of the region of interest is stored in Layer 9 located at the head of the code string. Only included.

  In addition, the quantization process is not performed in the quantization process, and a predetermined upper layer is configured to have an encoded data amount corresponding to a predetermined compression rate, as shown in FIG. As described above, it is possible to perform lossless encoding as a whole while making the compression rate correspond to each layer. However, since the highest layer (Layer 9) includes only the region of interest, the encoded data length differs if the size of the region is different.

  Since a medical image is large, its details cannot always be displayed for observation on a monitor having a limited number of pixels. In order to solve this problem, it is conceivable to display a large area by determining a region of interest. However, it is difficult to appropriately determine a region of interest from a small image, and means for determining the region of interest on behalf of a person is required. In the above embodiment, the region of interest determined by the diagnosis support technology is used, and the region is automatically displayed in high definition.

  That is, according to the above-described embodiment, when performing diagnosis support for medical images, the diagnosis support is performed with the minimum necessary images without reading all the images, and the portion determined to be positive by the diagnosis support means In contrast, a detailed image of the area is selectively captured. Therefore, efficient data transfer can be performed, and detailed image display can be quickly performed on important parts.

  Also, depending on the purpose of image diagnosis, it is possible to capture an appropriate initial image, and unnecessary data transfer does not occur for diagnostic support means that require only a rough image, thus reducing network traffic. It is possible.

  As a result of the diagnosis support means, for those that do not have a requested area, it is conceivable that an area in which the missing part is deleted from the irradiation field area that the inventor has already applied for is set as the area of interest. Further, by intentionally not setting (displaying) the region of interest, there is an effect of clearly indicating that there is no finding to the doctor who is viewing the image.

  Another object of the present invention is to supply a storage medium storing software program codes for implementing the functions of the above-described embodiments to a system or apparatus, and the computer (or CPU or MPU) of the system or apparatus stores the storage medium. Needless to say, this can also be achieved by reading and executing the program code stored in the.

  In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiments, and the storage medium storing the program code constitutes the present invention.

  As a storage medium for supplying the program code, for example, a floppy disk, a hard disk, an optical disk, a magneto-optical disk, a CD-ROM, a CD-R, a magnetic tape, a nonvolatile memory card, a ROM, or the like can be used.

  Further, by executing the program code read by the computer, not only the functions of the above-described embodiments are realized, but also an OS (operating system) operating on the computer based on the instruction of the program code. It goes without saying that a case where the function of the above-described embodiment is realized by performing part or all of the actual processing and the processing is included.

  Further, after the program code read from the storage medium is written into a memory provided in a function expansion board inserted into the computer or a function expansion unit connected to the computer, the function expansion is performed based on the instruction of the program code. It goes without saying that the CPU or the like provided in the board or the function expansion unit performs part or all of the actual processing, and the functions of the above-described embodiments are realized by the processing.

Claims (15)

  An image receiving device for receiving an encoded data stream of an original image from a transmitting device,
  Designating means for designating the resolution of the reduced image of the original image according to information for diagnostic purposes;
  Receiving means for receiving a first data stream, which is a data stream corresponding to a reduced image of the original image, having a resolution specified by the specifying means;
  The receiving means receives the first data stream, and then receives a second data stream that is the original image data stream and is not included in the first data stream. apparatus. Decoding means for decoding the data stream transmitted by the transmission device to obtain a two-dimensional image;
Analyzing the two-dimensional image obtained by decoding the first data stream by the decoding means, and analyzing means for determining a region of interest in the two-dimensional image ,
Said second data stream, image reception apparatus according to claim 1, wherein said analyzing means said generated from the encoded data stream of the original image based on the determined region of interest by, characterized and this . 3. The image receiving apparatus according to claim 2 , wherein the first data stream guarantees a minimum image quality that enables execution of processing according to the diagnostic purpose by the analyzing unit. . The image receiving device according to claim 3, wherein the region of interest is a region determined to be positive as a result of medical diagnosis support according to the diagnostic purpose . It said second data stream, image reception apparatus according to any one of claims 2 to 4, characterized in that the data stream representing the details of the image corresponding to the region of interest where the analyzing means outputs. Said second data stream, a region corresponding to the region of interest, any one of claims 2 to 5, wherein the is to the first image quality than the image quality obtained by the data stream 1 The image receiving device according to item . 7. The display device according to claim 2 , further comprising a display unit that displays the two-dimensional image decoded by the decoding unit, and displays the region of interest determined by the analyzing unit in the display. or image receiving apparatus according to item 1.   Based on the resolution information from the receiving device that designates the resolution of the reduced image of the original image according to the information for diagnosis purposes, a part of the data of the original image, the receiving device in advance Identifying means for identifying data to be transmitted to
  Transmission means for transmitting data other than the data specified by the specifying means out of the data of the original image to the receiving apparatus after transmitting the data specified by the specifying means to the receiving apparatus. A featured image transmission apparatus.   An image receiving system for receiving an encoded data stream of an original image from a transmitting device,
  Designating means for designating the resolution of the reduced image of the original image according to information for diagnostic purposes;
  Receiving means for receiving a first data stream, which is a data stream, corresponding to a reduced image of the original image having a resolution designated by the designation means;
  The receiving means receives the first data stream, and then receives a second data stream that is the original image data stream and is not included in the first data stream. system.   Based on the resolution information from the receiving device that designates the resolution of the reduced image of the original image according to the information for diagnosis purposes, a part of the data of the original image, the receiving device in advance Identifying means for identifying data to be transmitted to
  Transmission means for transmitting data other than the data specified by the specifying means out of the data of the original image to the receiving apparatus after transmitting the data specified by the specifying means to the receiving apparatus. A featured image transmission system.   A control method of an image receiving device for receiving an encoded data stream of an original image from a transmitting device,
  A designation step in which the designation means of the image receiving device designates the resolution of the reduced image of the original image according to information for diagnostic purposes;
  A receiving step for receiving a first data stream, which is a data stream corresponding to a reduced image of the original image, having a resolution specified in the specifying step;
  In the receiving step, after receiving the first data stream, a second data stream that is a data stream of the original image and is not included in the first data stream is received. Control method of the device.   A method for controlling an image transmission device, comprising:
  Based on the resolution information from the receiving device that designates the resolution of the reduced image of the original image according to the diagnostic purpose information, the specifying unit of the image transmission device uses a part of the original image data. A specifying step for specifying data to be transmitted to the receiving device in advance;
  After the transmission unit of the image transmission device transmits the data specified in the specifying step to the receiving device, the data other than the data specified in the specifying step among the data of the original image is transmitted to the receiving device. A method for controlling the image transmitting apparatus. Program for causing a computer to function as each unit of the image receiving apparatus according to any one of claims 1 to 7.   The program for functioning a computer as each means of the image transmission apparatus of Claim 8. A computer-readable medium storing the program according to claim 13 or 14.

Images