TWI470203B - An image analysis system applied to the detection of cancerous cells and a method of use thereof - Google Patents

Description

Image analysis system and method for detecting cancer cells

The present invention relates to a system and method for image analysis, and more particularly to an image analysis system and method for detecting cancer cells.

The traditionally used cell detection method is biomedical detection, which uses diffusion and random collision to complete the biomedical reaction, but the traditional biomedical test takes several hours or even several working days to complete, the process is time-consuming, and requires human intervention. Operation, it is not easy to achieve the required accurate measurement; after the biochip and flow cell technology to complete the cell detection to improve detection time and accuracy, but because the instrument is expensive, and requires more biological sample volume and complicated The chemical step can complete the detection, so it is not conducive to practical use; with the advancement of multi-spectral color reproduction technology, the CCD camera, microscope, filter and spectrophotometer are gradually used for cell detection, which is generally due to different spectrum algorithms. The system is divided into SVM detection and Winer's estimation method detection. The SVM detection system classifies cells to detect cell type, but since SVM detection requires liquid crystal tunable filter and needs to accurately control the exposure time of CCD for different liquid crystal tunable filters, Therefore, there is a greater demand for instruments and equipment, and Winer's The estimation method requires more optical parameters, so it is insufficient and is not convenient for practical use.

In order to solve the problems of high equipment, time consuming, and complicated procedures of the above existing cell detection methods, the main object of the present invention is to provide an image analysis system and method for detecting cancer cells.

The technical means used in the present invention is to provide an image analysis system for detecting cancer cells, comprising: an observation module comprising a stage unit, an illumination unit and an image enlargement unit, the stage unit For carrying a suspected cancer cell sample to be detected, the illumination unit provides a suspected cancer cell sample projected onto the stage unit by an illumination source, and the image enlargement unit is disposed on the stage unit to be suspected on the stage unit The image of the cancer cell sample is enlarged for easy identification; an image taking module is located in the output path of the observation module and includes a photosensitive coupling unit, a lens unit and a capture unit, and the photosensitive coupling element is received An image from a suspected cancer cell sample magnified by the image magnifying unit, the lens unit being disposed in the photo-sensing unit and focusing and providing an image of the suspected cancer cell sample amplified by the image magnifying unit to the photo-sensing unit Obtaining a magnified image of a clear suspected cancer cell sample, the capture unit being coupled to the photosensitive coupling Taking a magnified image of the focused suspected cancer cell sample; and a multi-spectral color image reproduction module that performs spectral analysis and color on the magnified image of the focused suspected cancer cell sample captured by the capture unit Gain and color image reproduction, the magnified image of the focused suspected cancer cell sample captured by the capture unit is reprocessed and provided to the physician to assist in determining the degree of cancer cell detection.

In addition, the present invention provides an image classification applied to cancer cell detection. The analysis method is carried out by using the foregoing image analysis system, which comprises a process for establishing a cancer cell spectrum database, which comprises: a penetration spectrum analysis step, which uses multi-spectral technology to analyze the cancer of the four cancer stages Cells to obtain an average transmission spectrum, thereby obtaining spectral characteristics of cancer cells of the four cancer stages; and a database establishment step of establishing the cancer cell spectrum based on spectral characteristics of cancer cells of the four cancer stages a database; and a multi-spectral color reproduction image detection cell process, which is performed on the basis of the spectrum database of the cancer cells, and the detection cell flow system of the multi-spectral color reproduction image includes: a step of capturing images, Acquired cancer cell image; a cell position circle selection step, which uses an algorithm to circle the cell location of the suspected cancer cell image to determine the position of the cell in the image; a penetration spectrum analysis step, which will The circled cancer cell location utilizes multispectral analysis to derive an average transmission spectrum; a cancer cell classification step that is suspected of cancer The images of the cells are classified into four stages of cancer cells according to different penetrating spectra; a color gain step is to increase the color difference of the images of the four stages of cancer cells; a color image reproduction step is performed by principal component analysis, linear Regression and color adaptation conversion to obtain the spectral performance of the circled cancer cells; and a cell detection, which identifies cancer cells of different phases through the alignment of the cancer cell spectrum database.

The image analysis method applied to cancer cell detection of the present invention requires no complicated chemical steps, and the time required for detecting cells is shorter than that of the conventional biomedical cell detection method, and the present invention provides non-contact type. Cell detection is less likely to cause contamination or damage to the sample; compared to SVM detection and Winer's estimation method detection, the present invention The instrumentation required is relatively simple, and there is no need to control the exposure time, and the parameters used are relatively small.

In order to be able to understand the technical features and practical functions of the present invention in detail, it can be implemented in accordance with the contents of the specification, and further illustrated in the preferred embodiment as illustrated in the following:

The present invention is an image analysis system for detecting cancer cells. Referring to the preferred embodiment of FIG. 1, the invention includes an observation module 10, an image capturing module 20, and a multi-spectral color image reproduction module 30.

The observation module 10 includes a stage unit 12, an illumination unit 11 and an image enlargement unit 13 for carrying a suspected cancer cell sample to be detected. The illumination unit 11 is located on the stage. Below the unit 12, an illumination source is provided to project and penetrate a suspected cancer cell sample on the stage unit 12, and the image amplifying unit 13 is disposed on the stage unit 12 to image the suspected cancer cell sample on the stage unit 12. It can be enlarged to facilitate identification.

More preferably, as shown in FIG. 2 , the viewing module 10 can further include a filter unit 14 , which is located on a light source projection path of the illumination unit 11 , so that the illumination unit 11 The provided illumination source is filtered to produce a filtered source of the desired wavelength band and is provided with a suspected cancer cell sample on the stage unit 12, thereby increasing the color difference of the suspected cancer cell sample, the filter unit 14 comprising The red, green, blue, cyan, purple, and yellow filters can be used in combination; in addition, the lighting unit 11 can also be replaced with lighting units of different color systems to provide light source effects of different wavelength bands.

The image capturing module 20 is located in the output path of the viewing module 10 and includes a photosensitive coupling unit 22 (CCD), a lens unit 21, and a capturing unit 23, and the photosensitive coupling element 22 is The plurality of rectangular photosensitive elements are arranged in an array and a column, and the photosensitive elements of the horizontal and vertical dimensions are recorded as pixels of the electronic image, and the photosensitive coupling unit 22 receives the suspected cancer amplified by the image amplifying unit 13. More preferably, the photosensitive coupling unit 22 senses a light source image from the illumination unit 11 provided by the illumination unit 11 and penetrating the suspected cancer cell sample and passing through the image amplifying unit 13 . The unit 21 is disposed on the photosensitive coupling unit 22 to focus and provide an image of the suspected cancer cell sample amplified by the image amplifying unit 13 to the photosensitive coupling unit 22 to obtain a magnified image of a clear suspected cancer cell sample. Preferably, the lens unit 21 can focus on the penetrating light source provided by the lighting unit 11 below the stage unit 12, and the capturing unit 23 is connected to The photosensitive coupling unit 22 captures an enlarged image of the focused suspected cancer cell sample, and the capturing unit 23 can be a camera or a spectrophotometer. More preferably, the spectrophotometer can be a spectrophotometric model of the model CS1000A. meter.

The multi-spectral color image reproduction module 30 performs the spectrum analysis 31, the color gain 32, and the color image reproduction 33 on the enlarged image of the focused suspected cancer cell sample captured by the capture unit 23, and the capture unit 23 The magnified image of the sampled, suspected cancer cell sample taken is reprocessed and provided to the physician to assist in determining the extent of cancer cell detection.

Since the spectrum analysis, the color gain, and the color image reproduction technique have been disclosed in the invention of the Republic of China Application No. 098137687 "Color Image Reproduction Method", the details of the technology will not be described in detail, and the application of the present invention will be described below. Image analysis method for cancer cell detection.

In the multi-spectral color image reproduction module, a cancer cell spectrum database needs to be built first, and a multi-spectral color reproduction image detection cell is performed according to the cancer cell spectrum database.

Referring to FIG. 3, the cell germplasm database establishment process 40 includes: the cell cancer staging pre-verification step can be achieved by various methods including thermal effects, cytopathology, flow cytometry, and frequency spectrum. Cytopathology, the present invention is described by taking a thermal effect experiment as an example; a cell thermal effect experimental step: the cells are used for suspected cancerous cells to perform different hyperthermia temperatures and different hyperthermia times, and are calculated by cell survival rate calculation. The survival rate of the cells, and the relationship between the temperature survival rate and the temperature holding time of the experimental results, the temperature survival rate curve is plotted, as shown in Fig. 4 and Fig. 5, which is the second phase of the bladder cancer through the thermal effect time. -8301, bladder cancer stage III J82 and bladder cancer stage 4 TCC-sup cells were heated to 43 °C in Fig. 4 and 45 °C in Fig. 5, and bladder cancer was determined by a graph of temperature survival rate and temperature holding time relationship. The cells are in the second, third and fourth phases; the Arrhenius model is used to obtain quantitative data of activation energy and frequency factors, and the normal cells can be compared with the second phase by algorithm. Cancer cells are calculated to derive data from the first phase of cancer cells. Please refer to Table 1 below for a quantitative data table of activation energy and frequency factors based on the Arrhenius model to analyze the four cancer stages of suspected cancerous bladder cells. ;

Penetration spectrum analysis step 41: The multi-spectral technique is used to analyze the bladder cancer cells of the four cancer stages to obtain an average transmission spectrum. Please refer to FIG. 6 to realize partial second stage bladder cancer cell penetration. Spectrum, the dotted line is the fourth stage of bladder cancer cell penetration spectrum, thereby obtaining the spectral characteristics of the bladder cancer cells of the four cancer stages; the database establishment step 42: based on the four cancer stages of bladder cancer The spectral characteristics of the cells establish a database of cancer cell spectrum.

Based on the cancer cell spectrum database, the detection cells of a multi-spectral color reproduction image can be performed. Referring to FIG. 7, the process 50 for detecting cells of the multi-spectral color reproduction image includes: capturing the image step 51: It uses the suspected cancer cell image obtained by the image capturing module; cell position circle selection step 52: it uses an algorithm to circle the cell position of the suspected cancer cell image to determine the position of the cell in the image. Penetration spectrum analysis step 53: the average transmission spectrum is obtained by multi-spectral analysis of the circled cancer cell position; the cancer cell classification step 54: the classification of the cancer cells by the downward trend of the penetration spectrum; As cancer cells of different degrees of cancer progress with the degree of canceration, their nuclei will become more and more swollen, so that the ratio of nuclei occupying the whole cell gradually increases, and the cell-based permeability is lower than that of the cytoplasm. Cancer cell classification by the downward trend of the penetration spectrum; color gain step 55: which increases the color difference of the images of cancer cells of different stages; Increasing image chromatic aberration, multi-threshold setting, external filtering, boundary detection, color space conversion, noise removal, etc. can be used to process the color of the image; color image reproduction step 56: using principal component analysis, linear Regression and color adaptation conversion to obtain the spectral performance of the circled cancer cells; the principal component analysis method is a commonly used method for multivariate statistics, and is applied to color technology. The main purpose of principal component analysis is to define a large amount of spectrum. The main axis of the information and the data of the information are simplified, mainly by reorganizing the original data, calculating the highly correlated and independent variables, and then obtaining the main components by analysis to obtain the variation of most of the data in the original data. Sexuality, thereby producing the main variable of the spectrum of known cancer cell images, the main base of the spectrum; linear regression is widely used in statistical analysis methods, the purpose of which is to understand whether the target variable can be linear with the independent variable The equation is used to represent and use it to explain the characteristics of the target variable, that is, whether the relationship between the variables X and Y is close, and finally Through the linear regression, the Y value can be obtained from the value of the variable X, thereby increasing the number of bases of the spectrum, and transmitting the base obtained by the principal component analysis with linear regression to make the image reproduction more accurate; the color adaptation conversion is based on von Based on the concept and theory of Kries, von Kries proposed that "human visual receivers and human perceptions should be independent and not mutually influential", so they experience color in the human eye. In the process of adapting to the conversion, the appropriate mode should be used to convert and convert the "color tristimulus" of the observed object into a "cone-like sensor value" related to human vision to predict the color in different observation environments. Performance capability; the method can convert the color of the object observed under the original source end light source to the chromaticity value expressed by the target end light source through the ratio between the source end and the destination end and the conversion matrix of different modes; Detection 57: It identifies and selects cancer cells of different phases through the alignment of the cancer cell spectrum database, and can represent cancer cells of different stages by circles of different colors.

The above is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any one of ordinary skill in the art can use the present invention without departing from the technical features of the present invention. Equivalent embodiments of the local changes or modifications made by the disclosed technology are still within the scope of the technical features of the present invention.

10‧‧‧Observation module

11‧‧‧Lighting unit

12‧‧‧stage unit

13‧‧‧Image magnification unit

14‧‧‧ Filter unit

20‧‧‧Image capture module

21‧‧‧ lens unit

22‧‧‧Photosensitive coupling element

23‧‧‧Capture unit

30‧‧‧Multi-spectral color image reproduction module

31‧‧‧ spectrum analysis

32‧‧‧Color gain

33‧‧‧Color image reproduction

40‧‧‧The process of establishing a cancer cell spectrum database

41‧‧‧Permeability Spectrum Analysis Steps

42‧‧‧Database establishment steps

50‧‧‧Multi-spectral color reproduction image detection cell flow

51‧‧‧ Capture image steps

52‧‧‧ cell position circle selection step

53‧‧‧Permeability Spectrum Analysis Steps

54‧‧‧ Cancer Cell Classification Steps

55‧‧‧Color Gain Step

56‧‧‧Color image reproduction steps

57‧‧‧cell detection

1 is a block diagram of a system in accordance with a preferred embodiment of the present invention.

2 is a system block diagram of another preferred embodiment of the viewing module of the present invention.

3 is a flow chart showing the establishment of a cancer cell spectrum database of the present invention.

Fig. 4 and Fig. 5 are graphs showing the relationship between the survival rate and the temperature holding time of bladder cancer cells at different temperatures.

Figure 6 is a spectrum of the penetration of bladder cancer cells in the second and fourth phases.

Figure 7 is a flow diagram of the detection of cells in a multispectral color reproduction image of the present invention.

10. . . Observation module

11. . . Stage unit

12. . . Lighting unit

13. . . Image magnification unit

20. . . Image capture module

twenty one. . . Lens unit

twenty two. . . Photosensitive coupling element

twenty three. . . Capture unit

30. . . Multi-spectral color image reproduction module

31. . . Spectrum analysis

32. . . Color gain

33. . . Color image reproduction

Claims (7)

An image analysis system for detecting cancer cells, comprising: an observation module, comprising: a stage unit, an illumination unit and an image enlargement unit, wherein the stage unit is configured to carry a suspected cancer cell sample to be detected The illumination unit provides a suspected cancer cell sample projected onto the stage unit by an illumination source, and the image enlargement unit is disposed on the stage unit to enlarge the image of the suspected cancer cell sample on the stage unit to facilitate An image capture module is disposed in an output path of the observation module and includes a photosensitive coupling unit, a lens unit and a capture unit, and the photosensitive coupling element receives the suspected cancer amplified by the image amplification unit An image of a cell sample, the lens unit being disposed in the photosensitive coupling unit and focusing an image of a suspected cancer cell sample amplified by the image magnifying unit and providing the photosensitive coupling unit to obtain a clear magnification of the suspected cancer cell sample Image, the capture unit is coupled to the photosensitive coupling unit to capture a focused suspected cancer cell sample And a multi-spectral color image reproduction module, wherein the amplified image of the focused suspected cancer cell sample captured by the capture unit is subjected to spectrum analysis, color gain, and color image reproduction, and the capture unit is A magnified image of the sampled, suspected cancer cell sample taken is reprocessed and provided to the physician for assistance in determining the extent of cancer cell detection; wherein the spectral analysis utilizes an algorithm to magnify the focused suspected cancer cell sample The position of the cell in which the cancer cell image is suspected is circled to determine the position of the cell in the image; and the position of the circled cancer cell is analyzed by multispectral analysis to obtain an average transmission spectrum; and further suspected The images of cancer cells are classified into four stages of cancer cells according to different penetration spectra. The image analysis system for detecting cancer cells according to claim 1, wherein the observation module comprises a filter unit, and the filter unit is located on a light source projection path of the illumination unit to enable the illumination. The illumination source provided by the unit is filtered to produce a filtered source of the desired band to provide a suspected cancer cell sample on the stage unit. The image analysis system for detecting cancer cells according to claim 1, wherein the filter unit comprises red, green, blue, cyan, violet, and yellow filters and can be used in combination. The image analysis system for detecting cancer cells according to claim 1, wherein the capturing unit is a camera or a spectrophotometer. The image analysis system for detecting cancer cells according to claim 1, wherein the photosensitive coupling unit senses an image of the penetrating light source from the stage unit passing through the image amplifying unit. An image analysis system for detecting cancer cells according to claim 1, wherein said lens unit focuses a reflected light source from the stage unit. An image analysis method applied to cancer cell detection, comprising: a process for establishing a cancer cell spectrum database, comprising: a penetrating spectrum analysis step, which uses multi-spectral technology to analyze cancer cells of four cancer stages Obtaining an average transmission spectrum, thereby obtaining spectral characteristics of cancer cells of the four cancer stages; and a database establishing step of establishing a spectrum database of the cancer cells according to spectral characteristics of the cancer cells of the four cancer stages; And a multi-spectral color reproduction image detection cell process, which is Based on the cancer cell spectrum database, the multi-spectral color reproduction image detection cell flow system includes: a capture image step, a suspected cancer cell image; a cell position circle selection step, which utilizes an algorithm The position of the cell suspected of cancer cell imaging is circled to determine the position of the cell in the image; a penetrating spectrum analysis step is to use the multi-spectral analysis of the circled cancer cell position to obtain an average transmission spectrum; a cell sorting step of classifying images of suspected cancer cells into fourth-stage cancer cells according to different penetration spectra; a color gain step of increasing the color difference of the images of the fourth-stage cancer cells; a color image reproduction step, It uses principal component analysis, linear regression, and color-adaptive transformation to obtain the spectral performance of the circled cancer cells; and a cell detection that identifies different stages of cancer through the alignment of the cancer cell spectrum database. cell.