Computer classification of biological tissue

1. A method for classification of a biological tissue using a computing system, the method comprising:

obtaining a set of raw time-series images of an examination area of the biological tissue, wherein the set of raw time-series images corresponds to a discrete time period in which portions of the biological tissue undergo a transient optical response responsive to topical application of a pathology differentiating agent to the biological tissue;

obtaining time data associated with the set of raw time-series images;

obtaining a set of subject characteristics, wherein each subject characteristic corresponds to a subject from which the biological tissue originates, and wherein each subject characteristic relates to at least one of medical history information, risk factor information, or clinical test result information;

inputting the time data, the set of subject characteristics, and at least one of the set of raw time-series images or a modified set of time-series images into a trained neural network that characterizes a likelihood that the biological tissue corresponds to a disease state, wherein the modified set of time-series images is created by applying one or more transformations to each raw time-series image including scaling or rotating to create the modified set of time-series images;

obtaining one or more classifications of the biological tissue from the trained neural network; and

outputting a cancer assessment based on the one or more classifications output by the trained neural network, wherein the cancer assessment indicates one of a precancerous disease state, a cancerous disease state, or a non-cancerous state.

2. The method of claim 1 , wherein:

at least one image of the set of raw time-series images is captured at a start of the discrete time period, prior to the transient optical response occurring; and

at least some of the set of raw time-series images are captured at intervals of a predetermined duration during the topical application of the pathology differentiating agent.

3. The method of claim 1 , wherein:

the biological tissue comprises a cervix uteri;

the examination area is exposed to optical radiation during the discrete time period;

the pathology differentiating agent comprises an acid; and

said obtaining the set of raw time-series images comprises capturing a plurality of optical images of the examination area of the biological tissue using an image collection module, the set of raw time-series images being derived from the plurality of optical images.

4. The method of claim 1 , wherein one or more of:

each of the modified set of raw time-series images is derived from a respective initial image transformed so as to provide alignment of the examination area within the modified set of raw time-series images,

each of the modified set of raw time-series images is derived from a respective initial image processed to remove one or more artefacts,

wherein the biological tissue comprises a cervix uteri, and

wherein the method further comprises processing the modified set of raw time-series images to identify a portion of the modified set of raw time-series images corresponding with the cervix uteri.

5. The method of claim 1 , wherein each of the set of raw time-series images is defined by a respective set of pixels, each of the sets of pixels having a same pixel arrangement, the method further comprising:

obtaining, map data, the map data comprising a respective analysis index for each pixel of the pixel arrangement, the analysis indices being derived from the set of raw time-series images; and

inputting the map data into the trained neural network, wherein the trained neural network characterizes the likelihood that the biological tissue corresponds to a disease state based on the map data.

6. The method of claim 1 , further comprising:

processing the set of raw time-series images to identify at least one morphological characteristic and/or at least one extracted feature; and

inputting an indication of the at least one morphological characteristic and/or the at least one extracted feature into the trained neural network, wherein the trained neural network characterizes the likelihood that the biological tissue corresponds to a disease state based on the indication.

7. The method of claim 1 , wherein the trained neural network comprises at least one of a convolutional neural network, a fully-connected neural network, or a recurrent neural network; and

the trained neural network is multi-modal.

8. The method of claim 1 , wherein the one or more subject characteristics comprises the risk factor information, the medical history information, and the clinical test result information.

9. The method of claim 8 , wherein the risk factor information comprises at least one of an age of the subject, a smoker status of the subject, a prior HPV vaccination status of the subject, information on use of condom during intercourse for the subject, or a parity for the subject; and

wherein the clinical test result information comprises at least one of a prior cytology result, a prior HPV test result, a prior HPV typing test result, a prior cervical treatment information, or a prior history of screening for and/or diagnosis of cervical cancers or pre-cancers.

10. The method of claim 1 , further comprising:

allocating one of the one or more classifications to an entire portion of the biological tissue based on a different trained neural network.

11. The method of claim 1 , wherein the one or more classifications indicate a presence of at least one morphological characteristic.

12. The method of claim 1 , wherein the trained neural network was previously trained based on another set of raw time-series images, a respective allocated classification for each of a plurality of other biological tissues, and a user-determined or database classification.

13. A computing system operative for classification of a tissue, comprising a processor configured to:

obtain a set of raw time-series images of an examination area of a biological tissue, wherein the set of raw time-series images corresponds to a discrete time period in which portions of the biological tissue undergo a transient optical response responsive to topical application of a pathology differentiating agent to the examination area of the biological tissue;

obtain time data associated with the set of raw time-series images;

apply one or more transformations to each raw time-series image including scaling or rotating to create a modified set of time-series images;

obtain a set of subject characteristics, wherein each subject characteristic corresponds to a subject from which the biological tissue originates, and wherein each subject characteristic relates to at least one of medical history information, risk factor information, or clinical test result information; and

input the time data, the set of subject characteristics, and the modified set of time-series images into a trained neural network configured to characterizes a likelihood that the biological tissue corresponds to a disease state;

obtain one or more classifications of the biological tissue from the trained neural network; and

output a cancer assessment based on the one or more classifications output by the trained neural network, wherein the cancer assessment indicates one of a precancerous disease state, a cancerous disease state, or a non-cancerous state.

14. The computing system of claim 13 , further comprising:

an image collection module configured to capture the set of raw time-series images.

15. The computing system of claim 14 , wherein:

the image collection module is located remotely from a processor on which the trained neural network is operated.

16. The method of claim 1 , further comprising:

determining a measure of diffuse reflectance associated with the set of raw time-series images; and

inputting the measure of diffuse reflectance into the trained neural network, wherein the trained neural network characterizes the likelihood that the biological tissue corresponds to a disease state based on the measure of diffuse reflectance.

17. The method of claim 1 , wherein the trained neural network is a first trained neural network, and wherein said outputting the cancer assessment is performed using a second trained neural network that is different from the first trained neural network.

18. The method of claim 1 , wherein the at least one of the set of raw time-series images or the modified set of time-series images comprises both the set of raw time-series images and the modified set of time-series images.

19. The computing system of claim 13 , wherein the processor is configured to:

determine a measure of diffuse reflectance associated with the set of raw time-series images; and

input the measure of diffuse reflectance into the trained neural network, wherein the trained neural network characterizes the likelihood that the biological tissue corresponds to a disease state based on the measure of diffuse reflectance.

20. The method of claim 1 , further comprising applying one or more transformations to each raw time-series image including scaling or rotating to create a modified set of time-series images.