Assay reading method
A computer implemented method for reading a test region of an assay, the method comprising: (i) providing digital image data of a first assay; (ii) inputting the digital image data into a trained convolutional neural network configured to output a first probability, based on the input digital image data, that a first region of pixels of the digital image data corresponds to a first test region of the assay; (iii) if the first probability is at or above a first predetermined threshold, accepting the first region of pixels as a first region of interest associated with the first test region; and (iv) estimating an intensity value of a portion of the first test region in the first region of interest.
1. A computer implemented method for reading a test region of an assay, the method comprising:
(i) receiving digital image data of a first assay, the digital image data comprising at least one region of pixels corresponding to a first test region of a first assay and at least one region of pixels corresponding to a housing of the first assay;
(ii) inputting the digital image data into a trained convolutional neural network configured to output, for the input digital image data:
a first region of pixels of the input digital image data corresponding to the first test region of the assay, and a first probability that the first region of pixels corresponds to the first test region; and
a second region of pixels of the input digital image data corresponding to the housing of the first assay, and a second probability that the second region of pixels corresponds to the housing,
(iii) if the first probability is at or above a first predetermined threshold, reading the first test region by estimating an intensity value of at least a portion of the first test region in the first region of pixels,
wherein estimating the intensity value comprises performing thresholding on the pixels of the first region of pixels and the second region of pixels, wherein said thresholding comprising replacing any pixel value in the input digital image data below a predetermined threshold value to a first new value, and any pixel value above the threshold to a second new value;
or otherwise discarding the first region of pixels; and
(iv) when the second probability is at or above a second threshold value, determining that the first test region is associated with the first assay by determining that the first region of pixels is within the second region of pixels.
2. The computer implemented method according to claim 1 ,
wherein the trained convolutional neural network is further configured to output a third region of pixels of the input digital image data corresponding to a second test region of the assay, and a third probability that the third region of pixels corresponds to the second test region; and
wherein the method further comprises:
if the second probability is at or above the first predetermined threshold, reading the second test region by estimating an intensity value of at least a portion of the second test region in the second region of pixels, or otherwise discarding the third region of pixels.
3. The computer implemented method according to claim 1 , wherein the trained convolutional neural network uses a MobileNet architecture, and/or an SSD Single Shot MultiBox Detector architecture.
4. A data-processing apparatus comprising a processor and a memory configured to store computer instructions, wherein the processor is configured to execute the computer instructions to carry out the steps of the method of claim 1 .
5. A non-transitory computer readable storage medium storing a computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of the method of claim 2 .
6. A system for reading an assay, the system comprising:
a first assay; and
a portable user device comprising an image capture device and the non-transitory computer readable storage medium of claim 5 .
7. The system according to claim 6 , comprising a second assay.
8. The system according to claim 7 , wherein the first assay and the second assay respectively comprise a first lateral flow test device and a second lateral flow test device.
9. The system according to claim 8 ,
wherein the first test region is a first lateral flow test strip of the first lateral flow test device; and
wherein the portion of the first test region is a test line of the first lateral flow test strip.
10. The system according to claim 9 ,
wherein the second test region is a second lateral flow test strip of the second lateral flow test device; and
wherein the portion of the second test region is a test line of the second lateral flow test strip.
11. The system according to claim 10 , wherein the first test region of the first lateral flow test device is oriented at a skew angle relative to the second test region of the second lateral flow test device.
12. The system according to claim 6 , wherein the image capture device is configured to capture the digital image data as video data.
13. The computer implemented method according to claim 1 , further comprising training a convolutional neural network on a training data set comprising a plurality of images of lateral flow test devices to obtain the trained convolutional neural network.
14. The computer implemented method of claim 13 , wherein the training data set further comprises images of lateral flow test devices at different orientations and positions in a field of view of an image capture device.
15. The computer implemented method of claim 13 , wherein the training data comprises images where lateral flow devices are placed at any rotational misalignment from a vertical edge of a field of view of an image capture device.
16. The computer implemented method of claim 13 , wherein the training data comprises images where lateral flow devices are placed at a rotational misalignment up to 10 degrees, or up to 45 degrees from a vertical edge of a field of view of an image capture device.
17. The method of claim 1 , wherein the first new value is white and wherein the second new value is black.