RU2358640C2 - Method of improving quality image during fibroesophagogastroduodenoscopy and device to this end - Google Patents

Abstract

FIELD: information technology.

SUBSTANCE: present invention relates to computer engineering and medicine and can be used for improving quality of images during fibroesophagogastroduodenoscopy. The method of obtaining an image during fibroesophagogastroduodenoscopy involves combined digital processing of two frames and output of the reconstructed image for diagnosis. On the image frame, points lying in the light-exposed region are sought for, the illumination values of which exceed a preset threshold value, under the condition that, the number of identified points of the light-exposed region do not exceed the preset threshold value. Image correction of the light-exposed region is done through median filtering, otherwise a second image is obtained in the current position of the endoscope, and correction of the light-exposed region of the image is done by replacing points lying in the light-exposed region of the first image with points of the second image lying in the corresponding positions. In the device for realising this method, there is an image input unit, output unit of the reconstructed image, unit for controlling sensitivity of the image sensor, image processing unit, eliminator, median filtering unit, first memory device, second memory device, third memory device, fourth memory device and a register bank.

EFFECT: invention improves quality of endoscopic images of esophageal mucosa, the stomach and duodenum through reconstruction of light-exposed regions.

2 cl, 4 dwg, 1 ex

Description

The invention relates to computing and medicine and can be used to improve image quality with fibroesophagogastroduodenoscopy.

A known method of endoscopic examination of the stomach (patent No. 2253350 dated 06/10/2005), providing for fibroesophagoduodenoscopy. In this case, the organ lumen is filled with water through the suction channel of the endoscope. In this study, the distal end of the endoscope is completely immersed in the liquid and an obturator probe with a latex balloon at the end, which is installed behind the pyloric ring, is carried out parallel to the endoscope. During the study, the elevated position of the head end of the couch relative to the foot by 15-20 degrees is used.

The disadvantages of this method include the complexity of the technical implementation of fibroesophagoduodenoscopy, due to the presence of a number of additional necessary conditions.

A device for a wide-angle endoscope lens (patent No. 2197007 dated 01.20.2003) is known, which contains four components, the first of which is made in the form of a negative meniscus, and the fourth is in the form of a collective, and the aperture diaphragm located between the second and third components at the front focus of the third and fourth components. The second positive component contains the main lens and two lenses installed in front of it, the first of which is negative with a greater curvature of the second surface along the rays, the second is the meniscus, convex to the image, and the third, the main, is a biconvex lens. The third component is a negative lens made of a material whose refractive index changes relative to the base indicator in the direction perpendicular to the optical axis, the fourth component-team is made in the form of two positive lenses, the second of which has the last surface aligned with the image surface.

The disadvantage of this device is that the optical components used in the system can introduce distortions into the resulting image caused by aberrations and reflections inside the optical system and increase the area of the illuminated areas.

Closest to the proposed method is fluorescence endoscopy (patent No. 2290855 from 08/10/2005), which involves irradiating with white light or radiation a specific spectral composition of a biological object and registering the reflected radiation with subsequent formation of an image on the monitor screen. In this case, the reflected radiation is divided by a beam splitter into various channels. Formation of the video image on the monitor screen is carried out in a white light study mode, in a fluorescence study mode, the image is recorded as a separate frame with excitation from a single short light pulse. Immediately before this, the image of the same biological object is recorded as a separate frame in the study mode in white light. In this case, the peak power density of a short light pulse on the surface of a biological object exceeds the power density of white light. Then carry out joint digital processing of these two frames.

The main disadvantages of this method include the possibility of a significant increase in the illuminated areas during fluorescence studies, which negatively affects the quality of the image and reduces its information content.

Closest to the proposed is a device that implements a method of fluorescence endoscopy (patent No. 2290855 from 08/10/2005), containing a fluorescence endoscope, a combined lighting unit connected by a lighting bundle with an optical channel of the endoscope, reflected light channels equipped with means for receiving an image in reflected light and in the light of fluorescent radiation, filters, processor and monitor. Means for receiving the image are continuously recorded on the monitor image of the object in white light, and at the time of the light pulse register a separate fluorescence image. At the same time, the processor is equipped with an algorithm for digital processing of the superposition of the fluorescence frame and the frame in white light and outputting the resulting frame to the monitor.

The disadvantages of this device include the fact that when shooting at the time of a light pulse, the probability of increasing the area of the illuminated areas increases, which negatively affects the quality of the received image and reduces its information content.

An object of the invention is to improve the quality of endoscopic images of the mucous membrane of the esophagus, stomach and duodenum by restoring the exposed areas.

The technical problem is solved by the fact that in the known method, including joint digital processing of these two frames, the search for points belonging to the illuminated region, the brightness values of which exceeds a predetermined threshold value, is introduced, provided that the number of detected points of the illuminated region does not exceed a predetermined threshold value , the image correction of the illuminated area is carried out by median filtering, otherwise a second image is obtained at the current position of the endoscope, and the correction the illuminated region of the image is carried out by replacing the points belonging to the illuminated region of the first image with the points of the second image located at the corresponding positions.

The technical problem is solved in that in a known device that implements a method of fluorescence endoscopy, containing input, image processing and output of the restored image, a replacement unit, a median filtering unit, a sensitivity control unit, a register unit, a first storage device, a second storage device, a third are introduced a storage device and a fourth storage device.

The invention can be used to improve the quality of images obtained by the endoscope, by restoring the illuminated areas of the image and meets the criterion of "industrial applicability".

The invention is illustrated by drawings, where figure 1 shows the algorithm for processing an endoscopic image, figure 2 is a diagram of a device that implements the restoration of illuminated areas in the image, figure 3 and figure 4 - glare on the image of the mucous membrane obtained by endoscopy.

In an endoscopic examination, the accuracy of the diagnosis is largely dependent on the quality of the image obtained by the endoscope. However, when using a lighting system, illuminated areas may appear on the image obtained by the endoscope, as a result of which instead of the image of a certain part of the studied object, there will be a glare, which is a bright spot formed as a result of specular reflection of light from the source [Computer vision, trans. from English / L. Shapiro, J. Stockman - M .: Binom. Laboratory of Knowledge, 2006, 752 pp., - S.268].

To restore areas of the image distorted by glare, a sequential analysis of all points of the image is performed, which consists in finding those points that belong to the illuminated areas. When a point is detected whose brightness value exceeds a predetermined threshold value, which is a value, above which a point is considered to belong to a glare, the number of points in this glare is calculated. Provided that the number of flare points does not exceed a predetermined threshold value, the image area illuminated by it is restored using a median filter. Otherwise, in the current position of the endoscope with a reduced sensitivity of the image sensor, one more frame is obtained, and the distorted points of the first image are replaced by the points of the second located at the corresponding positions.

The device comprises an image input unit (BVI) 2 and a reconstructed image output unit (BV) 10, an image sensor sensitivity control unit (BCH) 1, an image processing unit (BOI) 8, a replacement unit (BZ) 9, a median filtering unit (MF) 11, block of registers (BR) 3, the first storage device (memory) 4, the second memory 5, the third memory 6, the fourth memory 7, and the first and second group inputs and the first and second group inputs and outputs of BOI 8 are connected respectively to the second and the first group outputs and the third and second group inputs / outputs BR 3, input d BOI 8 is connected to the output of BVI 2, the third and fourth group inputs-outputs of BOI 8 are connected to the group input-output of the fourth memory 7 and to the group input-output of the third memory 6, respectively, the second output of BOI 8 is connected to the first input of BV 10, group the input of which is connected to the first group output of the first memory 4, and the group output is designed to output information, the first input-output of the BOI 8 is connected to the input-output of the MF 11, the first and second group inputs and outputs of which are connected to the first group input-output of the BR 3 and to the first group entry the output of the first memory 4, respectively, and the group input to the first group output of the third memory 6, the second input-output of the BOI 8 is connected to the input-output of the BZ 9, in which the first and second group inputs are connected to the group output of the second memory 5 and to the second group the output of the third memory 6, respectively, and the group input-output is connected to the second group input-output of the first memory 4, the third group input of BOI 8 is connected to the second group output of the first memory 4, in turn, the group inputs of the first memory 4 and second memory 5 are connected respectively to the first and in BVI 2 group outputs, the input of which is connected to the output of the register block 3, BVI 2 group input is used to enter information from the image sensor, BKCH 1 group output is used to set the image sensor intensity, BKH 1 group input-output is connected to the fourth group input -BR 3 output, and the input - with the first output of BOI 8.

The register block 3 contains five registers: A, B, C, D and E. Register A is designed to store the threshold brightness value. Register B contains the offset of the current point relative to the beginning of the memory block. Register C is designed to store the sensitivity value of the image sensor controlled by the BCF 1. Register D is a flag register and contains the flags EF, FST, LST, BR, and EXC, which take the following values: EF = 1 - the device stops processing images, EF = 0 - continues, FST = 0 - the image was retrieved (with reduced sensitivity of the image sensor), FST = 1 - the first frame was received at the current endoscope position, LST = 1 - the last image point was processed, LST = 0 - the processing of the current image points continues, BR = 1 - brightness value and the point in question is greater than the threshold brightness value, BR = О - the brightness value of the point under consideration is less than the brightness threshold value, EXC = 1 - the brightness value of at least one point of the image under study requires replacement, EXC = 0 - the brightness values of all image points do not require replacement. Register E contains a threshold value for the number of flare points for applying a median filter.

The sensitivity control unit 1 of the image sensor controls the sensitivity of the image sensor, the current value of which is stored in register C.

The image input unit 2 is a means for receiving an image and receives an image and records the received frame in the first memory 4.

The image processing unit 8 implements:

- sequential analysis of all points of the processed frame (the offset of the current point from the beginning of the image stored in the first storage device (memory) is in register B);

- comparison of the brightness value of the current point with the threshold value stored in register A;

- counting the number of points in the current glare;

- comparison of the obtained value with the maximum possible number of points for applying the median filtering stored in register E;

- placing in the third memory in the position corresponding to the offset stored in the register In, the sign of the use of the median filter or sign of replacement;

- the formation of a signal indicating the need to reduce the sensitivity of the image sensor at the time of receiving one frame.

The replacement unit 9 is intended to replace the values of the first memory 4 with the values of the second memory 5 in those positions at which the third memory 6 contains a sign of replacement.

отсортированный массив n действительных чисел, то медиана множества чисел, хранящихся в массиве А, равна А[(n-1)/2] [Компьютерное зрение, пер. с англ. / Л.Шапиро, Дж.Стокман - М.. - Бином. Лаборатория знания, 2006, 752 с], и записи новых значений яркости обработанных точек в первое ЗУ 4 вместо уже имеющихся на соответствующие позиции.MF 11 is designed for median filtering of areas of the image for points of which at the corresponding positions in the third memory 6 there is a sign of the need to use a median filter that replaces the pixel value with the median of values from the vicinity of this pixel, and if A [i], i =

sorted array of n real numbers, then the median of the set of numbers stored in array A is A [(n-1) / 2] [Computer vision, trans. from English / L. Shapiro, J. Stockman - M .. - Binom. Laboratory of Knowledge, 2006, 752 s], and recording new brightness values of the processed points in the first memory 4 instead of those already available at the corresponding positions.

BV 10 is designed to output the reconstructed image for its further analysis, in particular for diagnosis.

The first, second, third and fourth memory units are memory blocks implemented on the basis of static or dynamic RAM [Lebedev ON The use of memory chips in electronic devices: Reference, manual. M .: Radio and communications, 1994. - 216 s]. In the first memory 4, the image obtained by the BWI 2 is stored, in the second memory 5, the frame obtained at the same endoscope position, but with a reduced image sensor sensitivity, is stored in the third memory 6 at the corresponding image points stored in the first memory 4 and in the second memory 5, signs of the need for replacement or signs of the need to use a median filter are stored in the positions, which indicate whether it is necessary to replace the brightness of a point at a given position in the first memory 4 to replace the brightness of a point in the same position of the second memory 5, apply median filtering to a given point or leave the brightness unchanged. The fourth memory 7 serves to store the displacements of the points that belong to the flare currently being studied.

The device operates as follows in accordance with the algorithm presented in figure 1.

The device performs sequential processing of endoscopic images of the mucous membrane obtained by the endoscope until the endoscopic analysis of the stomach is completed (block 1 of the algorithm shown in FIG. 1). The processing of each image is as follows.

Before image processing starts, the flags are set to the following values: FST = 1, LST = 0, BR = 0, EXC = 0.

After the operation of setting flags BCH 1 (in block 2 of the algorithm shown in figure 1) sets the sensitivity of the sensor in register C, at which an image frame will be received, applying the corresponding value from the group input-output.

Next, from the group output of the BCH 1 to the first group input of the BVI 2, the sensitivity value of the image sensor is received and from the group input to the BVI 2 the image is received. After that, BVI 2 from the first group output writes the image to the group input of the first memory 4 with the value “1” of the FST flag, and with the value “0” - from the second group output to the group input of the second memory 5 (block 4 of the algorithm). The value of the flag FST is fed to the input of the BVI 2 from the output of the BR 3.

Then, from the second group output of the first memory 4 to the third group input of the BOI 8, the point brightness value is received (block 6, Fig. 1), and the offset of the current point relative to the beginning of the first memory 4 goes to the second group input-output of the BOI 8 from the second group input - output BR 3, which corresponds to the register B.

After BOI 8 compares the obtained brightness value of the current point with the threshold value (block 7, Fig. 1) received at the second group input from the first group output of BR 3, which corresponds to register A. Provided that the received value is less than the threshold, BOI 8 sends from the second group input-output to the second group input-output of BR 3 a signal indicating an increase in the value in register B so that after it stores the offset of the next point relative to the beginning of the first memory 4. After BOI 8, it starts processing heel next point (block 5, Figure 1).

If the brightness value of the current point is greater than the threshold (block 8 of the algorithm), BOI 8, receiving data on the fourth group input-output from the group input-output of the memory 6, checks for the presence in the third memory 6 of a sign of the need to use a median filter or a sign of the need to replace for the current point at the position corresponding to the offset stored in register B. If the test result is positive, BOI 8 starts processing the next point (block 5, figure 1). Otherwise, BOI 8 counts the number of points in the glare to which the current point belongs (block 9 of the algorithm). In this case, BOI 8 from the third group input-output writes to the group input-output of the fourth memory 7 displacement (relative to the beginning of the first memory 4) points belonging to the current glare.

Then, at the first group input, BOI 8 receives from the second group output of BR 3, which corresponds to register E, the threshold value of the number of flare points for applying the median and compares the obtained value with the number of points of the current flare (block 10 of the algorithm). If the received value is less than the threshold, a signal is received from the fourth group input-output of the BOI 8 to the group input-output of the third memory 6, indicating the need to record signs of the use of the median filter at positions whose offsets are sent to the third group input-output of the BOI 8 from the group input the output of the fourth memory 7 (block 11, figure 1). Otherwise, BOI 8 (block 16 of the algorithm) records in the third memory 6 in the same position the signs of the need for replacement and from the first group input / output to the third group input / output BR 3 gives the value of the EXC flag equal to "1".

After, according to a given algorithm, the device has analyzed and processed all the image points, BOI 8 from the first group input-output to the third group input-output of BR 3 supplies the LST flag value equal to "1". After that for those points that belong to the glare, depending on the size of the illuminated area, in the third memory 6 in the positions corresponding to the displacements of the points relative to the beginning of the first memory 4, there is either a sign of the need to use a median filter, or a sign of the need for replacement.

After all the points of the image are examined, BOI 8 sends a signal to the input-output of MF 11 from the first input-output, indicating that at those points for which at the positions of the third memory 6, corresponding to their displacements relative to the beginning of the first memory 4 , there is a sign of the need to use the median filter received by MF 11 to the group input from the first group output of the memory 6, median filtering should be applied (block 13 of the algorithm). MF 11, after applying the median filter to each of these points, their new values from the second group input / output are written through the first group input / output to the first memory 4 instead of the existing ones.

After that (block 14, figure 1) BOI 8 checks the value of the EXC flag. If it is “1”, then BOI 8 (block 15 of the algorithm) sends a signal to the BCH input through its first output, indicating that the sensitivity of the image sensor should be reduced by the time of receiving one frame, and from the first group input-output to the third group input-output BR 3 gives the value of the flag EXC equal to "0".

Then, the newly received image of BVI 2 through the second group output is written to the group input of the second memory 5. BOI 8 (block 12 of the algorithm), having received the signal that the second image is in the second memory 5 at the input from the output of BVI 2, it sends from the second input -exit to the input-output of the BZ 9 signal that it is necessary to replace the brightness values of the points of the first memory 4 by applying to its second group input-output from the input-output of the BZ 9 brightness values of the points received from the group output of the second memory 5 to the first group input БЗ 9, in positions in which t etem memory 6 at respective offsets of data points from the beginning of the storage unit is an indication to replace the derived knowledge base 9 to the second group input from the second output of the third memory group 6.

After that, BOI 8 sends a signal to the input of the BV 10 through the second output that an image is received from the first group output of the first memory 4 to the BV 10 group input, in which the illuminated areas are completely restored and which can be transferred from the group output for further analysis.

After that, the device proceeds to processing the next image, if one is received (block 1 of the algorithm).

A clinical example of the use of the invention

Patient Bn V.A., medical history No. 2355/67, 26 years old, on admission complained of periodic mild aching pains in the epigastric region that occur 1 hour after eating, sometimes on an empty stomach, accompanied by nausea.

From the anamnesis it was found out that periodic hungry, nocturnal pains first began to bother about 7 years ago while serving in the army, with fibroesophagogastroduodenoscopy, a peptic ulcer was diagnosed. Exacerbations occur annually in the autumn-spring period. He repeatedly underwent outpatient and inpatient treatment at the place of residence. The last deterioration is about 2 weeks. Seeking medical help due to increased pain. Heredity: a peptic ulcer in the father. Objectively: the stomach is painful and tense in the epigastrium, Mendel's symptom is positive. Blood tests, urine without features.

When fibroesophagogastroduodenoscopy, the claimed invention by restoring the exposed areas of the image revealed an ulcer in the antrum of the stomach with a diameter of 1.5 cm (figure 3).

The following changes were also revealed: chronic gastritis, chronic duodenitis, deformity of the pyloric bulbar division of the II degree. No ulcers in the duodenum.

Clinical diagnosis: Peptic ulcer with localization of a chronic ulcer in the antrum, exacerbation. Chronic gastritis, exacerbation. Chronic duodenitis, exacerbation. Cicatricial deformity of the pyloric bulbar department of the II degree.

Thus, the invention improves the quality of endoscopic images of the mucous membrane of the esophagus, stomach and duodenum and, as a result, the accuracy of the diagnosis based on the endoscopic examination.

Claims (2)

1. A method of obtaining an image during fibroesophagogastroduodenoscopy, including the joint digital processing of two frames and outputting the reconstructed image for diagnosis, characterized in that on the image frame they search for points belonging to the illuminated region, the brightness values of which exceed a predetermined threshold value, provided that the number the detected points of the illuminated area does not exceed a predetermined threshold value, the image correction of the illuminated area is carried out by means of honey filtering, otherwise get the second image at the current position of the endoscope, and the correction of the illuminated area of the image is carried out by replacing the points belonging to the illuminated area of the first image located at the corresponding positions of the points of the second image. 2. The device for implementing the method according to claim 1, comprising input, processing, and output blocks of the reconstructed image, characterized in that a replacement unit, a median filtering unit, a sensitivity control unit, a register unit, a first storage device, a second storage device are additionally introduced, a third storage device and a fourth storage device, wherein the first and second group inputs and the first and second group inputs and outputs of the image processing unit are connected respectively to the second and the first group outputs and the third and second group inputs and outputs of the register unit, the input of the image processing unit is connected to the output of the image input unit, the third and fourth group inputs and outputs of the image processing unit are connected to the group input / output of the fourth storage device and to the group input the output of the third storage device, respectively, the second output of the image processing unit is connected to the first input of the output image recovery unit, the group input of which is connected to group output of the first storage device, and the group output is for outputting information, the first input-output of the image processing unit is connected to the input-output of the median filtering unit, the first and second group inputs and outputs of which are connected to the first group input-output of the register block and to the first group input-output of the first storage device, respectively, and the group input to the first group output of the third storage device, the second input-output of the communication image processing unit n with an input-output of a replacement unit, in which the first and second group inputs are connected to the group output of the second storage device and to the second group output of the third storage device, respectively, and the group input-output is connected to the second group input-output of the first storage device, the third group the input of the image processing unit is connected to the second group output of the first storage device, in turn, the group inputs of the first storage device and the second storage device the two are connected respectively to the first and second group outputs of the image input unit, the input of which is connected to the output of the register unit, the group input of the image input unit is used to enter information from the image sensor, the group output of the sensitivity control unit is configured to set the image sensor intensity value, group input - the output of the sensitivity control unit is connected to the fourth group input-output of the register unit, and the input to the first output of the image processing unit.

Images