Camera-Based Hemoglobin Detection

Abstract

A method for detecting a molecule in a sample includes applying a first filter to an image at a first wavelength. The method also includes applying a second filter to the image at a second wavelength. The method also includes applying a third filter to the image at a third wavelength. The first, second and third wavelengths are within a predetermined wavelength range, and the first, second and third wavelengths are different from one another. The method also includes detecting a spectral signature of the molecule in the sample in the image after the first, second and third filters have been applied to the image. The method also includes determining whether the molecule is present in the sample based at least in part on the detected spectral signature.

Description

[Technical Field]

[0001] The present disclosure relates generally to systems and methods for detecting molecules in a sample, and more particularly, to systems and methods for camera-based (e.g., spectral-based) hemoglobin detection in stool samples. [Background technology]

[0002] Colonoscopy is an endoscopic procedure commonly used to screen for colorectal cancer or detect other abnormalities in the colon and rectum. During a colonoscopy, a long, flexible tube (i.e., a colonoscope) is inserted into the rectum. A small video camera is attached to the end of the tube, allowing the physician to view the inside of the entire colon. If necessary, polyps or other types of abnormal tissue can be removed through the scope during a colonoscopy. Tissue samples (e.g., biopsies) can also be taken during a colonoscopy. However, alternative systems and methods for convenient and non-invasive colon cancer screening would be beneficial. Summary of the Invention [Problem to be solved by the invention]

[0003] According to one aspect of the present disclosure, a method for detecting molecules in a sample is disclosed. The method includes applying a first filter at a first wavelength to an image. The method also includes applying a second filter at a second wavelength to the image. The method also includes applying a third filter at a third wavelength to the image. The first, second, and third wavelengths are within a predetermined wavelength range, and the first, second, and third wavelengths are different from one another. The method also includes detecting a spectral signature of a molecule in the sample in the image after the first, second, and third filters have been applied to the image. The method also includes determining whether the molecule is present in the sample based at least in part on the detected spectral signature.

[0004] A method for detecting hemoglobin in a stool sample is also disclosed. The method includes capturing an image of the stool sample with a camera. The method also includes applying a first bandpass filter at a first wavelength to the image as the image is captured. The method also includes applying a second bandpass filter at a second wavelength to the image as the image is captured. The method also includes applying a third bandpass filter at a third wavelength to the image as the image is captured. The first, second, and third wavelengths are between 450 nm and 690 nm. The third wavelength is between the first and second wavelengths. The method also includes detecting a spectral signature of hemoglobin in the stool sample in the image after the first, second, and third bandpass filters have been applied to the image. The spectral signature includes absorption characteristics. Detecting the spectral signature includes performing continuum removal on the spectral signature between 450 nm and 690 nm using linear interpolation to remove the gradient from the spectral signature while preserving the absorption features. Performing continuum removal includes determining a first product of a weight and a value of the absorption feature at a first wavelength, determining a second product of a complement of the weight and a value of the absorption feature at a second wavelength, and determining a sum of the first and second products. Detecting the spectral signature also includes determining a band ratio of the absorption features for each pixel in the image. The band ratio comprises a ratio of the sum to a value of the absorption feature at a third wavelength. The method also includes aggregating the band ratios for the pixels in the image. The method also includes determining that hemoglobin is present in the stool sample based at least in part on the aggregation of the band ratios.

[0005] A system for detecting molecules in a sample is also disclosed. The system includes a camera configured to capture an image of the sample. The system also includes a computing system. The computing system is configured to detect spectral signatures of molecules in the sample in the image after first, second, and third filters have been applied to the image. The computing system is also configured to determine whether the molecule is present in the sample based at least in part on the detected spectral signatures. [Brief explanation of the drawings]

[0006] [Figure 1] FIG. 1 is a schematic diagram of a system for identifying molecules in a sample, according to one embodiment. [Figure 2] Figure 2A shows an image of a stool sample containing hemoglobin, and Figure 2B shows an image of a stool-only (i.e., no hemoglobin) sample, according to one embodiment. [Figure 3] 1 is a graph showing stool samples containing hemoglobin versus stool-only samples, according to one embodiment. [Figure 4] 1 is a graph showing dose response in hemoglobin detection with increasing hemoglobin concentration in stool, according to one embodiment. [Figure 5] 1 is a graph showing reflectance versus wavelength of hemoglobin at different concentrations, according to one embodiment. [Figure 6] 1 is a graph illustrating detection of a spectral signature in an image according to one embodiment. [Figure 7] 1 shows a flowchart of a method for detecting molecules in a sample, according to one embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0007] The presently disclosed subject matter will now be described more fully hereinafter with reference to the accompanying drawings, in which some, but not all embodiments of the disclosure are shown. Like numbers refer to like elements throughout. The presently disclosed subject matter may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will satisfy applicable legal requirements. Indeed, many variations and other embodiments of the presently disclosed subject matter described herein will come to mind to one skilled in the art to which the presently disclosed subject matter pertains having the benefit of the teachings presented in the foregoing descriptions and the associated drawings. Therefore, it is to be understood that the presently disclosed subject matter is not limited to the particular embodiments disclosed, and that variations and other embodiments are intended to be included within the scope of the appended claims.

[0008] FIG. 1 is a schematic diagram of a system 100 for identifying molecules in a sample, according to one embodiment. The molecules may be or include hemoglobin, bilirubin, calprotectin, albumin, fatty acids, hydrogen sulfide, etc. The sample may be or include stool, urine, saliva, another biological specimen, or a combination thereof. FIG. 2A shows an image 200A of a stool sample 210A containing hemoglobin, according to one embodiment, and FIG. 2B shows an image 200B of a stool-only (i.e., hemoglobin-free) sample 210B, according to one embodiment.

[0009] System 100 may include camera 110, computing system 120, and inspection device 130. In one embodiment, camera 110, computing system 120, inspection device 130, or a combination thereof, may be co-located in a single device. For example, at least a portion of system 100 may include, be a part of, or connect to a smartphone, tablet, laptop, or the like. Camera 110 may be configured to capture one or more images (e.g., images 200A, 200B) of a sample (e.g., samples 210A, 210B), which may or may not have molecules therein. As part of capturing images 200A, 200B, one or more filters (e.g., three shown: 112, 114, and 116) may be applied. In one embodiment, filters 112, 114, and 116 may be applied to the lens of camera 110. For example, filters 112, 114, and 116 may be or include thin film filters that cover the lens of camera 110. The membrane filters may be manually changed so that three images (one for each filter) are captured for a single sample. In another embodiment, the filters 112, 114, 116 may be integrated into (e.g., directly into) the CCD focal plane of the camera 110. In yet another embodiment, a Bayer pattern filter array may be used. In yet another embodiment, the filters 112, 114, 116 may be applied to the images 200A, 200B by the computing system 120.

[0010] The computing system 120 may be configured to analyze the images 200A, 200B to detect the presence and / or amount of a molecule (e.g., hemoglobin) in a sample (e.g., stool). More specifically, the computing system 120 may be configured to detect a molecule's unique spectral signature to distinguish between samples containing the molecule and samples that do not contain the molecule. FIG. 3 is a graph showing a stool sample 210A containing hemoglobin versus a stool-only sample 210B, according to one embodiment. In one example, as the amount of a molecule in a sample increases, the detectability of the molecule also increases in a dose-response manner. This is shown in FIG. 4, which is a graph showing the dose-response in hemoglobin detection with increasing hemoglobin concentration in stool, according to one embodiment. More specifically, ratios of specific wavelength features and quantitative comparisons of these resulting values ​​across multiple types of calculations can help distinguish samples that contain hemoglobin from samples that do not.

[0011] Testing device 130 may be or include an immunochemical fecal testing (FIT) device or other diagnostic test / information. Testing device 130 may test for molecules (e.g., hemoglobin) in a sample (e.g., stool) before, simultaneously with, or after camera 110 and computing system 120 attempt to detect the presence and / or amount of the molecules in the sample. For example, testing device 130 may be configured to connect to computing system 120 and function as a secondary testing system for molecules after camera 110 and computing system 120 perform image-based detection.

[0012] FIG. 5 is a graph showing reflectance versus wavelength for hemoglobin at different concentrations, according to one embodiment. Each hemoglobin curve may have one or more spectral signatures (e.g., also referred to as fingerprints) 510 that can be used to detect its presence in various types of samples. The spectral signatures 510 may resemble the letter V (also referred to as a V feature) and / or the letter W (also referred to as a W feature). The spectral signatures 510 may be within a predetermined wavelength range. The predetermined wavelength range may be from about 450 nm to about 690 nm, from about 575 nm to about 625 nm, or from about 600 nm to about 650 nm.

[0013] As described in further detail below, the computing system 120 may implement a spectral processing algorithm on the images 200A, 200B to detect the presence of a molecular (e.g., hemoglobin) spectral signature 510. To accomplish this, the algorithm may utilize spectral continuum removal and / or band ratio analysis. Continuum removal may use linear interpolation to remove the slope of the spectral signature 510 while preserving one or more spectral absorption features 520A, 520B. As used herein, spectral absorption feature refers to a change in the shape of a spectral curve. Continuum removal may be performed within a predetermined wavelength range.

[0014] After continuum removal is performed, a band ratio may then be determined for one or more of the absorption features 520A, 520B to measure the ratio of the absorption features 520A, 520B, which indicates the amount of chemical associated with the absorption features 520A, 520B present in the sample. In one embodiment, the reflectance value at point 520B may represent the numerator, and the reflectance value at point 520A may be the denominator. The ratio (e.g., numerator / denominator) may be greater than or equal to 1 when positive (i.e., hemoglobin is present).

[0015] 6 is a graph illustrating detection of a spectral signature in an image, according to one embodiment. Line 610 represents a measured spectrum (e.g., from spectral signature 510). Interpolation can be used to determine spectrum 620, which is a measure of the continuum (e.g., overall shape) of measured spectrum 610.

[0016] Curve 630 can be determined by dividing spectrum 620 by spectrum 610. This is referred to as continuum removal, which effectively removes the overall shape of the measured spectrum 610 while preserving (e.g., enhancing) the spectral features (e.g., spectral absorption features 520A, 520B). From the continuum-removed spectrum 630, the spectral depth can be determined using band ratios, as shown along dashed vertical line 640.

[0017] 7 shows a flowchart of a method 700 for detecting molecules in a sample, according to one embodiment. An exemplary sequence of method 700 is shown below, although one or more steps of method 700 may be performed in a different order, combined, divided into substeps, repeated, or omitted. One or more steps of method 700 may be performed by system 100.

[0018] The method 700 may include capturing one or more images of the sample, as in 702. For example, this may include capturing an image 200A including the sample 210A. The image 200A may be captured with the camera 110.

[0019] Method 700 may also include applying one or more filters 112, 114, 116 to image 200A, as at 704. As described above, filters 112, 114, 116 may be applied by camera 110 and / or by computing system 120. Filters 112, 114, 116 may be or include bandpass filters configured to transmit a predetermined wavelength range. As described above, for hemoglobin, the predetermined wavelength range may be from about 450 nm to about 690 nm, from about 575 nm to about 625 nm, or from about 600 nm to about 650 nm.

[0020] The filters 112, 114, and 116 may each be configured to transmit a different wavelength. The first filter 112 may be configured to transmit a first wavelength, the second filter 114 may be configured to transmit a second wavelength, and the third filter 116 may be configured to transmit a third wavelength. The third wavelength may be between the first and second wavelengths. In one example, the first wavelength may be approximately 639 nm to approximately 647 nm, the second wavelength may be approximately 623 nm to approximately 631 nm, and the third wavelength may be approximately 628 nm to approximately 636 nm. In another example, the first wavelength may be approximately 643 nm, the second wavelength may be approximately 627 nm, and the third wavelength may be approximately 632 nm.

[0021] The method 700 may also include detecting a spectral signature 510 in the image 200A, as at 706. The spectral signature 510 (e.g., a V-shape and / or a W-shape) may be detected by the computing system 120. The spectral signature 510 may be detected after the filters 112, 114, 116 are applied to the image 200A. The spectral signature 510 may be unique to the molecule being detected (e.g., hemoglobin). As described above, the spectral signature 510 may include one or more absorption features 520A, 520B.

[0022] In one embodiment, detecting the spectral signature 510 may include performing continuum removal on the spectral signature 510, as in 708. The continuum removal may be performed within a predetermined wavelength range. The continuum removal may be performed using linear interpolation to remove gradients from the spectral signature 510 while preserving absorption features (e.g., absorption feature 520A). In one example, performing continuum removal may include: w * r(λ1)+(1-w) * r(λ2)...Equation 1 where w represents a weighting value, r(λ1) represents the spectral value of absorption feature 520A at a first wavelength, and r(λ2) represents the spectral value of absorption feature 520A at a second wavelength. The weighting value w may be specific to the particular molecule being detected. For example, the weighting value w may be 0.5156 for hemoglobin.

[0023] In other words, performing continuum removal may include determining a first product of the weight and the spectral value of the absorption feature 520A at a first wavelength, determining a second product of the complement of the weight and the spectral value of the absorption feature 520A at a second wavelength, and determining the sum of the first product and the second product.

[0024] Detecting the spectral signature 510 may also or alternatively include determining band ratios of the absorption features 520A in the spectral signature 510, as in 610. The band ratios may be determined after continuum removal is performed. In one example, the band ratios may include: Bandwidth ratio = (w * r(λ1)+(1-w) * r(λ2)) / r(λ3) Equation 2 where r(λ3) represents the spectral value of absorption feature 520A at the third wavelength. In other words, the numerator of the band ratio may include the sum (from Equation 1) and the denominator of the band ratio may include the spectral value of absorption feature 520A at the third wavelength.

[0025] In one embodiment, image 200A may include multiple pixels, and spectral signature 510 may be detected (e.g., band ratios may be determined) for one or more pixels. For example, spectral signature 510 may be detected (e.g., band ratios may be determined) for all pixels in image 200A.

[0026] Method 700 may also include aggregating band ratios for pixels in image 200A, as at 712. One or more techniques may be used to aggregate the band ratios. For example, one technique may include aggregating or counting the number of values ​​above a certain threshold, while another technique may include aggregating or counting values ​​over a certain spatial region. In one embodiment, one technique may be used when the concentration of the molecule in the sample is below a predetermined concentration threshold, and another technique may be used when the concentration of the molecule in the sample is above a predetermined concentration threshold. In another embodiment, multiple techniques may be combined to create a complex.

[0027] Method 700 may also include determining, as at 714, that a molecule (e.g., hemoglobin) is present in sample 210A. The determination that a molecule is present may be based, at least in part, on detecting spectral signature 510, determining a band ratio, aggregating the band ratio, or a combination thereof. Method 700 may be capable of detecting a predetermined mass of a molecule (e.g., hemoglobin) in 1 gram of sample 210A (e.g., stool + hemoglobin). The predetermined mass may be between about 5 micrograms and about 10 micrograms, or between about 10 micrograms and about 20 micrograms, which is below the threshold used in conventional FIT testing in the United States (i.e., 20 micrograms hemoglobin / gram stool).

[0028] The method 700 may also include determining the amount of molecules present in the sample 210A, as at 716. The determination of the amount of molecules present may be based, at least in part, on detecting the spectral signature 510, determining band ratios, aggregating band ratios, other mathematical approaches, or a combination thereof. In another embodiment, the determination of the amount of molecules present may be a function of band ratio scores from pixels of the sample. The use of "function" may refer to any algorithm that takes band ratio scores as input and generates a numerical value that is assigned to the sample.

[0029] Method 700 may also include confirming the presence of a molecule (e.g., hemoglobin) in sample 210A using inspection device 130, as at 718. This may also, or instead, include determining the amount of the molecule present in sample 210A using inspection device 130. This step may occur before, simultaneously with, or after one or more of steps 702-716. For example, this step may occur in response to the image-based determinations in step 714 and / or step 716.

[0030] In another embodiment, additional diagnostic information may be obtained about the patient instead of or in addition to using the testing device 130 (e.g., FIT test). For example, the patient's medical history or test data may be used to generate a composite score that includes multiple risk variables beyond the spectrum and FIT alone.

[0031] Method 700 may also include performing a colonoscopy, as at 720. The colonoscopy may be performed at least in part in response to determining that the molecule is present (at 714), determining the amount of the molecule present (at 716), determining that the molecule is present (at 718), or a combination thereof.

[0032] While the present disclosure has been described in connection with its preferred embodiments, those skilled in the art will understand that additions, deletions, modifications, and substitutions not specifically described may be made therein without departing from the spirit and scope of the present disclosure as defined in the appended claims.

Claims

1. 1. A method for detecting a molecule in a sample, comprising: applying a first filter at a first wavelength to the image; applying a second filter at a second wavelength to the image; applying a third filter to the image at a third wavelength, the first, second, and third wavelengths being within a predetermined wavelength range, and the first, second, and third wavelengths being different from one another; detecting spectral signatures of molecules in the sample in the image after the first, second and third filters have been applied to the image; and determining whether the molecule is present in the sample based at least in part on the detected spectral signature.

2. 10. The method of claim 1, wherein the molecule comprises hemoglobin, the sample comprises a stool sample, and the spectral signature comprises a V-shaped portion of a curve or a W-shaped portion of a curve on a reflectance versus wavelength graph.

3. 3. The method of claim 2, wherein determining whether the molecule is present comprises determining that the molecule is present, the method further comprising: confirming that said hemoglobin is present in said stool sample using an immunochemical fecal test (FIT).

4. 2. The method of claim 1, wherein the predetermined wavelength range is between 450 nm and 690 nm, and the third wavelength is between the first and second wavelengths.

5. 2. The method of claim 1, wherein detecting the spectral signature comprises: performing continuum removal on the spectral signature within the predetermined wavelength range; determining a band ratio of absorption features of the spectral signature after the continuum removal is performed, and determining that the molecule is present in the sample based at least in part on the band ratio.

6. 6. The method of claim 5, wherein the continuum removal is performed using linear interpolation to remove gradients from the spectral signature while preserving the absorption features.

7. 6. The method of claim 5, wherein the step of performing continuum removal comprises: determining a first product of a weight and a value of the absorption feature at the first wavelength; determining a second product of the complement of the weight and the value of the absorption feature at the second wavelength; determining a sum of the first product and the second product.

8. 8. The method of claim 7, wherein the bandwidth ratio comprises a ratio of the sum to a value of the absorption feature at the third wavelength.

9. The method of claim 5 , wherein the band ratio is determined for each pixel in the image.

10. 10. The method of claim 9, further comprising determining the amount of the molecule present in the sample based at least in part on compiling the band ratios.

11. 1. A method for detecting hemoglobin in a stool sample, comprising: capturing an image of the stool sample with a camera; applying a first bandpass filter at a first wavelength to the image as the image is captured; applying a second bandpass filter at a second wavelength to the image as the image is captured; applying a third bandpass filter at a third wavelength to the image as the image is captured, the first, second, and third wavelengths being between 450 nm and 690 nm, and the third wavelength being between the first and second wavelengths; detecting a spectral signature of the hemoglobin in the stool sample in the image after the first, second, and third bandpass filters have been applied to the image, the spectral signature comprising an absorption characteristic, and detecting the spectral signature includes: performing a continuum removal on the spectral signature between 450 nm and 690 nm using linear interpolation to remove a gradient from the spectral signature while preserving the absorption features, wherein performing the continuum removal comprises: determining a first product of a weight and a value of the absorption feature at the first wavelength; determining a second product of the complement of the weight and the value of the absorption feature at the second wavelength; determining a sum of the first product and the second product; determining a bandwidth ratio of the absorption feature for each pixel in the image, the bandwidth ratio comprising a ratio of the sum to a value of the absorption feature at the third wavelength; aggregating the bandwidth ratios for pixels in the image; determining that the hemoglobin is present in the stool sample based at least in part on compiling the band ratios.

12. 12. The method of claim 11, wherein the first wavelength is from about 639 nm to about 647 nm, the second wavelength is from about 623 nm to about 631 nm, and the third wavelength is from about 628 nm to about 636 nm.

13. 12. The method of claim 11, wherein the spectral signature comprises a V-shaped portion of a curve or a W-shaped portion of a curve on a reflectance versus wavelength graph.

14. 12. The method of claim 11, further comprising determining the amount of hemoglobin present in the stool sample based at least in part on compiling the band ratios.

15. 12. The method of claim 11, further comprising, after determining that the hemoglobin is present in the stool sample, confirming that the hemoglobin is present in the stool sample using an immunochemical fecal test (FIT).

16. 1. A system for detecting molecules in a sample, comprising: a camera configured to capture an image of the sample; 1. A computing system comprising: detecting spectral signatures of molecules in the sample in the image after the first, second and third filters have been applied to the image; and a computing system configured to determine whether the molecule is present in the sample based at least in part on the detected spectral signature.

17. 17. The system of claim 16, wherein the first, second, and third filters are applied to the camera lens before the image is captured, and the wavelength of the third filter is between the wavelengths of the first and second filters.

18. 17. The system of claim 16, wherein the first, second, and third filters are applied by the computing system after the image is captured, and the wavelength of the third filter is between the wavelengths of the first and second filters.

19. 17. The system of claim 16, wherein detecting the spectral signature comprises: performing continuum removal on the spectral signature within a predetermined wavelength range; determining a band ratio of absorption features of the spectral signature after the continuum removal is performed, wherein the molecule is determined to be present in the sample based at least in part on the band ratio.

20. 17. The system of claim 16, wherein the molecule comprises hemoglobin, the sample comprises a stool sample, and the computing system further comprises an immunochemical fecal test (FIT) configured to confirm the presence of the hemoglobin in the stool sample after determining that the hemoglobin is present in the stool sample.

Images