Automated parasite analysis system

1. A parasite analysis system comprising:

a material reservoir configured to store a biological sample;

an imaging cell controllably connected to the material reservoir and configured to receive a portion of the biological sample from the material reservoir;

a camera configured to capture a chronological set of images of the portion of the biological sample in the imaging cell; and

an image analysis system configured to:

identify candidate patches in at least a subset of the chronological set of images that are candidates for depicting a parasite;

filter the candidate patches based on a comparison of locations of the candidates patches to locations of candidate patches in images immediately before and/or after the candidate patches;

generate, based on the filtered candidate patches, parasite counts for the subset of the chronological set of images, each parasite count indicating a number of parasites detected in a corresponding image of the chronological set of images; and

generate, based on the parasite counts, an estimate of a number of parasites in the portion of the biological sample, wherein the estimate of the number of parasites in the portion of the biological sample is based on the locations of the candidate patches that depict a parasite.

2. The parasite analysis system of claim 1 , wherein the image analysis system being configured to identify candidate patches comprises the image analysis system being configured to:

identify patches in at least some images of the chronological set of images as the candidate patches based on foreground pixels included in the patches, wherein each candidate patch is a portion of one of the images having a location within the corresponding image; and

apply a classifier to the candidate patches to identify candidate patches that depict a parasite.

3. The parasite analysis system of claim 2 , wherein the image analysis system being configured to identify candidate patches further comprises the image analysis system being configured to:

identify foreground pixels in at least some of images of the chronological set of images, the foreground pixels being pixels whose values change significantly between images in the set, wherein the candidate patches are identified based on foreground pixels included in the patches.

4. The parasite analysis system of claim 2 , wherein, to identify patches in at least some images of the chronological set of images as candidate patches, the image analysis system is configured to:

place a window of predetermined size at a plurality of positions in an image, the predetermined size based on an expected size of a parasite;

calculate, for each position of the window, a foreground pixel score based on a number of foreground pixels within the window;

identify a portion of the image within the window when the window is at a given position as a candidate patch responsive to the foreground pixel score for the given position exceeding a threshold; and

set at least some pixels within the window when the window is at the given position as background pixels responsive to the foreground pixel score for the given position exceeding the threshold.

5. The parasite analysis system of claim 2 , wherein, to identify candidate patches that depict a parasite, the classifier is configured to:

calculate a histogram of gradients for a candidate patch in a first image of the chronological set of images;

identify a comparison portion of a second image in the chronological set of images, a location of the comparison portion in the second image corresponding to a location of the candidate patch in the first image;

calculate a histogram of gradients for the comparison portion; and

determine whether the candidate patch depicts a parasite based on a comparison between the histogram of gradients for the candidate patch and the histogram of gradients for the comparison portion.

6. The parasite analysis system of claim 1 , wherein, to generate the estimate of the number of parasites in the portion of the biological sample, the image analysis system is configured to:

determine a parasite count for each of a plurality of images of the chronological set of images; and

generate the estimate of the number of parasites by calculating a median of the parasite counts.

7. The parasite analysis system of claim 1 , further configured to:

generate an estimate of a number of parasites in each of a plurality of additional portions of the biological sample; and

generate an estimate for a total number of parasites in the biological sample based on the estimates of the number of parasites in the portion and the plurality of additional portions.

8. The parasite analysis system of claim 1 , further comprising:

a piston or syringe; and

a three-way valve connecting the material reservoir, imaging cell, and piston or syringe, wherein, in a first configuration, the portion of the biological sample can flow from the material reservoir to the piston or syringe and, in a second configuration, the portion of the biological sample can flow from the piston or syringe to the imaging cell.

9. The parasite analysis system of claim 8 , further comprising a controller configured to:

generate a first control signal that causes the three-way valve to be in the first configuration;

generate a second control signal that causes the piston or syringe to suck the portion of the biological sample into the piston or syringe;

generate a third control signal that causes the three-way valve to be in the second configuration; and

generate a fourth control signal that causes the piston or syringe to push the portion of the biological sample into the imaging cell.

10. The parasite analysis system of claim 1 , wherein the material reservoir is pressurized and the imaging cell is controllably connected to the material reservoir by an input valve, the system further comprising:

a controller configured to open the input valve to enable the portion of the biological sample to flow from the material reservoir into the imaging cell.

11. The parasite analysis system of claim 10 , further comprising an output valve configured to control flow of the portion of the biological sample from the imaging cell into a waste depository, wherein the controller opens and closes the input valve and the output valve simultaneously.

12. The parasite analysis system of claim 10 , wherein the controller is configured to open the input valve, wait for a predetermined period of time in a range from one hundred milliseconds to one second while the portion of the biological sample flows into the imaging cell, and close the input valve after the predetermined period of time.

13. The parasite analysis system of claim 12 , wherein the camera is further configured to capture the chronological set of images in a period beginning in a range from two seconds to three seconds after the input valve is closed.

14. A method for analyzing a sample, the method comprising:

receiving a chronological set of images of a portion of the sample;

identifying candidate patches in at least a subset of the chronological set of images that are candidates for depicting a parasite;

filtering the candidate patches based on a comparison of locations of the candidates patches to locations of candidate patches in images immediately before and/or after the candidate patches;

generating, based on the filtered candidate patches, parasite counts for the subset of the chronological set of images, each parasite count indicating a number of parasites detected in a corresponding image of the chronological set of images; and

generating, based on the parasite counts, an estimate of a number of parasites in the portion of the sample, wherein the estimate of the number of parasites in the portion of the sample is based on the locations of the candidate patches that depict a parasite.

15. The method of claim 14 , further comprising:

identifying foreground pixels in at least some of images of the chronological set of images, the foreground pixels being pixels whose values change significantly between images in the set, wherein the candidate patches are identified based on foreground pixels included in the patches.

16. The method of claim 14 , wherein identifying the candidate patches comprises:

placing a window of predetermined size at a plurality of positions in an image, the predetermined size based on an expected size of a parasite;

calculating, for each position of the window, a foreground pixel score based on a number of foreground pixels within the window;

identifying a portion of the image within the window when the window is at a given position as a candidate patch responsive to the foreground pixel score for the given position exceeding a threshold; and

setting at least some pixels within the window when the window is at the given position as background pixels responsive to the foreground pixel score for the given position exceeding the threshold.

17. The method of claim 14 , wherein filtering the candidate patches comprises:

calculating a histogram of gradients for a candidate patch in a first image of the chronological set of images;

identifying a comparison portion of a second image in the chronological set of images, a location of the comparison portion in the second image corresponding to a location of the candidate patch in the first image;

calculating a histogram of gradients for the comparison portion; and

determining whether the candidate patch depicts a parasite based on a comparison between the histogram of gradients for the candidate patch and the histogram of gradients for the comparison portion.

18. The method of claim 14 , wherein generating the estimate of the number of parasites in the portion of the sample comprises:

determining a parasite count for each of a plurality of images of the chronological set of images; and

generating the estimate of the number of parasites by calculating a median of the parasite counts.

19. The method of claim 14 , further comprising:

generating an estimate of a number of parasites in each of a plurality of additional portions of the sample; and

generating an estimate for a total number of parasites in the sample based on the estimates of the number of parasites in the portion and the plurality of additional portions.

20. A non-transitory, computer-readable medium comprising computer program code that, when executed by a computing system, causes the computing system to perform operations including:

receiving a chronological set of images of a portion of a sample;

identifying candidate patches in at least a subset of the chronological set of images that are candidates for depicting a parasite;

filtering the candidate patches based on a comparison of locations of the candidates patches to locations of candidate patches in images immediately before and/or after the candidate patches;

generating, based on the filtered candidate patches, parasite counts for the subset of the chronological set of images, each parasite count indicating a number of parasites detected in a corresponding image of the chronological set of images; and

generating, based on the parasite counts, an estimate of a number of parasites in the portion of the sample, wherein the estimate of the number of parasites in the portion of the sample is based on the locations of the candidate patches that depict a parasite.