Apparatus and method for automatic detection of pathogens

Patent metadata

Field	Value
Publication number	US-9522396-B2
Application number	US-201113338291-A
Country	US
Kind code	B2
Filing date	Dec 28, 2011
Priority date	Dec 29, 2010
Publication date	Dec 20, 2016
Grant date	Dec 20, 2016

Abstract

The invention discloses an apparatus and method for automatic detection of pathogens within a sample. The apparatus and method are especially useful for high-throughput and/or low-cost detection of parasites with minimal need for trained personnel. The invention entails automated microscopic data acquisition, and performing image processing of images captured from a sample using classification algorithms. Visual classification features of structures are extracted from the image and compared to visual classification features associated with known pathogens. A determination is reached whether a pathogen is present in the sample; and if present, the pathogen may be identified to a pathogen species. Diagnosis is rapid and does not require medically trained personnel.

First claim

The invention claimed is:

1. A method of automatic detection of plasmodium within a sample comprising:
obtaining, for each of a plurality of fields, at least three digital images of a sample stained with a combination of acridine orange and Hoechst for differential staining of DNA and RNA, wherein the images are obtained under at least two different lighting conditions, including a brightfield lighting condition and a fluorescent lighting condition, wherein at least a first image is obtained under a lighting condition causing fluorescence of DNA in the sample, and at least a second image is obtained under a lighting condition causing fluorescence of RNA in the sample;
performing image processing of the images using at least one classification algorithm, said image processing including identifying bodies showing fluorescent staining for DNA in the at least a first digital image and showing fluorescent staining for RNA in the at least a second digital image, extracting one or more visual classification features of the identified bodies, and using at least one machine-learning algorithm to determine a presence or absence of plasmodium ; and
outputting an indication of said determination.

2. The method of claim 1 , wherein said sample is selected from the group consisting of a blood smear (thin or thick), a fecal smear, a lympathic smear, a cerebrospinal fluid smear, and a tissue biopsy.

3. The method of claim 1 , wherein the digital images are obtained at a single focus depth.

4. The method of claim 1 , wherein the digital images are obtained at a plurality of magnifications.

5. The method of claim 1 , wherein the digital images are obtained using one or more optical filters or one or more dichroic beamsplitters/mirrors for exciting or detecting fluorescence.

6. The method of claim 1 , wherein the digital images are obtained using a plurality of microscopy methods.

7. The method of claim 6 , wherein said plurality of microscopy methods comprise different fluorescence excitations and/or emissions, thereby yielding multiparametric image data.

8. The method of claim 1 , wherein said obtaining digital images includes, by a processor that interacts with and activates mechanical and optical components of a digital microscopy system, performing at least one of the following:
selecting at least one light source from among a plurality of light sources, and activating said light source;
selecting an exposure time of a light source;
selecting an exposure time for a digital camera.

9. The method of claim 1 , wherein said indication is indicative of at least one of the following: the presence or absence of the plasmodium ; the species of the plasmodium ; the number or concentration of the plasmodium detected; the life stage of the plasmodium ; a finding of anemia; a finding of an unusual white blood cell count; and information on the quality of the sample.

10. The method of claim 1 , wherein said outputting further includes outputting one or more images of suspected plasmodium.

11. The method of claim 1 , wherein said one or more visual classification features are selected from the group consisting of: motion, size, shape, coloring, contrast, location in respect to additional biological structures, presence of internal structures, presence of extracellular structures, the aspect ratio, the optical density, florescence at predetermined wavelengths, optical birefringence, clustering behavior, intensity, gradient, distance from human cells to suspected plasmodium , and pattern matching.

12. The method of claim 1 , wherein said image processing includes, for at least one image, searching for at least one patch which is likely to contain a target and marking it as a candidate.

13. The method of claim 12 , wherein said image processing further includes, for at least one image, at least one of the following:
calculating the likelihood that said at least one candidate contains a target;
processing said candidates for finding if more than one candidate belongs to the same target, and where found, candidates of the same target are clustered together;
tracking at least one candidate, in relation to said at least one cluster, that may belong to said cluster, and where said tracked candidate belongs, adding said tracked candidate to said cluster;
determining and classifying the likelihood that said at least one cluster contains a target; and
determining if said image contains a target, based on said classification, and thus determining if said sample contains a plasmodium.

14. The method of claim 12 , wherein said searching for at least one patch which is likely to contain a target is performed using at least one of the following: pattern matching, model matching, high florescence segmenting and clustering, and multi-frame tracking.

15. The method of claim 1 , wherein said image processing is performed using at least one of the following modules: a single frame classification cascade module; a multi-frame candidate construction module; a multi-frame candidate tracking module; a multi-frame candidate classification module; a sample classification module; a motion field construction module; an image verification module; a camera control model; and a masking module.

16. The method of claim 1 , wherein said image processing is performed in an aggregated processing manner.

17. The method of claim 1 , wherein said image processing is performed on-the-fly.

18. The method of claim 1 , wherein said image processing includes identifying and discarding at least one poor image out of said digital images.

19. The method of claim 11 , wherein the visual classification features which are used in the image processing, for any given digital image, depend at least in part on the magnification of the digital image.

20. The method of claim 1 wherein obtaining said digital images of the sample comprises using an automated microscopy system comprising at least one light source, at least one lens, and at least one digital camera; said digital images are obtained prior to image processing.

Assignees

• Bachelet Ido
• Pollak Joseph Joel
• Levner Daniel
• Bilu Yonatan
• Yorav-Raphael Noam
• S D Sight Diagnostics Ltd

Inventors

• Bachelet Ido
• Pollak Joseph Joel
• Levner Daniel
• Bilu Yonatan
• Yorav-Raphael Noam

Classifications

• B01L2300/0864

  comprising only one inlet and multiple receiving wells, e.g. for separation, splitting · CPC title

• G01N15/1463

  Physics · mapped topic

• G01N21/6458

  Fluorescence microscopy (fluorescence microscopes per se G02B21/0076 and G02B21/16) · CPC title

• G01N2021/6421

  Measuring at two or more wavelengths · CPC title

• B01L2300/0816

  Cards, e.g. flat sample carriers usually with flow in two horizontal directions · CPC title