Systems and methods for processing images for image matching
A computer-implemented method for processing electronic medical images, the method including receiving a plurality of electronic medical images of a medical specimen associated with a single patient. The plurality of electronic medical images may be inputted into to a trained machine learning system, the trained machine learning system being trained to compare each of the plurality of electronic medical images to each other to determine whether each pair of the electronic medical images matches within a predetermined similarity threshold. The trained machine learning system may output whether each pair of the electronic medical images matches within a predetermined similarity threshold. The output may be stored.
1 . A computer-implemented method for processing electronic medical images, comprising:
receiving a plurality of electronic medical images associated with a medical sample of a first patient;
providing the plurality of electronic medical images to a trained machine learning system, the trained machine learning system being trained with vectors extracted from each of the plurality of electronic medical images, the vectors concatenated to form unified vectors for each of the plurality of electronic medical images, wherein the trained machine learning system compares each of the plurality of electronic medical images to each other to determine whether each pair of the electronic medical images matches within a predetermined similarity threshold based on the training;
receiving, from the trained machine learning system, an output indicating whether each pair of the plurality of electronic medical images matches within the predetermined similarity threshold, wherein a match within the predetermined similarity threshold indicates that a pair of electronic medical images is associated with the medical sample of the first patient; and
outputting and/or storing the output indicating whether each pair of the electronic medical images matches within the predetermined similarity threshold.
2 . The method of claim 1 , wherein providing the plurality of electronic medical images to a trained machine learning system further comprises:
determining one or more vectors associated with each of the plurality of electronic medical images; and
providing the one or more vectors associated with each of the plurality of electronic medical images to the trained machine learning system.
3 . The method of claim 1 , wherein training of the trained machine learning system comprises:
generating sets of matching and non-matching slides as training data; and
training the machine learning system utilizing binary classification loss.
4 . The method of claim 1 , wherein training of the trained machine learning system comprises:
generating sets of training data wherein each set included an anchor image, positive image, and negative image; and
training the machine learning system utilizing triplet loss training techniques.
5 . The method of claim 1 , wherein training of the trained machine learning system comprises:
generating sets of matching and non-matching slides as training data;
extracting vectors from each of the plurality of electronic medical images;
concatenating the vectors from each of the plurality of electronic medical images to form unified vectors for each image; and
training the machine learning system utilizing pairwise neural network classifications.
6 . The method of claim 1 , wherein providing the plurality of electronic medical images to the trained machine learning system further comprises:
receiving a plurality of historical electronic medical images corresponding to one or more individuals.
7 . The method of claim 1 , wherein outputting and/or storing the output indicating whether each pair of the electronic medical images matches within the predetermined similarity threshold further comprises:
flagging any medical image outside of the predetermined similarity threshold for a human technician to perform review.
8 . A system for processing electronic digital medical images, the system comprising:
at least one memory storing instructions; and
at least one processor configured to execute the instructions to perform operations comprising:
receiving a plurality of electronic medical images associated with a medical sample of a first patient;
providing the plurality of electronic medical images to a trained machine learning system, the trained machine learning system being trained with vectors extracted from each of the plurality of electronic medical images, the vectors concatenated to form unified vectors for each of the plurality of electronic medical images, wherein the trained machine learning system compares each of the plurality of electronic medical images to each other to determine whether each pair of the electronic medical images matches within a predetermined similarity threshold based on the training;
receiving, from the trained machine learning system, an output indicating whether each pair of the plurality of electronic medical images matches within the predetermined similarity threshold, wherein a match within the predetermined similarity threshold indicates that a pair of electronic medical images is associated with the medical sample of the first patient; and
outputting and/or storing the output indicating whether each pair of the electronic medical images matches within the predetermined similarity threshold.
9 . The system of claim 8 , wherein providing the plurality of electronic medical images to the trained machine learning system further comprises:
determining one or more vectors associated with each of the plurality of electronic medical images; and
providing the one or more vectors associated with each of the plurality of electronic medical images to the trained machine learning system.
10 . The system of claim 8 , wherein training of the trained machine learning system comprises:
generating sets of matching and non-matching slides as training data; and
training the machine learning system utilizing binary classification loss.
11 . The system of claim 8 , wherein training of the trained machine learning system comprises:
generating sets of training data wherein each set included an anchor image, positive image, and negative image; and
training the machine learning system utilizing triplet loss training techniques.
12 . The system of claim 8 , wherein training of the trained machine learning system comprises:
generating sets of matching and non-matching slides as training data;
extracting vectors from each of the plurality of electronic medical images;
concatenating the vectors from each of the plurality of electronic medical images to form unified vectors for each image; and
training the machine learning system utilizing pairwise neural network classifications.
13 . The system of claim 8 , wherein providing the plurality of electronic medical images to the trained machine learning system further comprises:
receiving a plurality of historical electronic medical images corresponding to one or more individuals.
14 . The system of claim 8 , wherein outputting and/or storing the output indicating whether each pair of the electronic medical images matches within the predetermined similarity threshold further comprises:
flagging any medical image outside of the predetermined similarity threshold for a human technician to perform review.
15 . A non-transitory computer-readable medium storing instructions that, when executed by a processor, perform operations processing electronic digital medical images, the operations comprising:
receiving a plurality of electronic medical images associated with a medical sample of a first patient;
providing the plurality of electronic medical images to a trained machine learning system, the trained machine learning system being trained with vectors extracted from each of the plurality of electronic medical images, the vectors concatenated to form unified vectors for each of the plurality of electronic medical images, wherein the trained machine learning system compares each of the plurality of electronic medical images to each other to determine whether each pair of the electronic medical images matches within a predetermined similarity threshold based on the training;
receiving, from the trained machine learning system, an output indicating whether each pair of the plurality of electronic medical images matches within the predetermined similarity threshold, wherein a match within the predetermined similarity threshold indicates that a pair of electronic medical images is associated with the medical sample of the first patient; and
outputting and/or storing the output indicating whether each pair of the electronic medical images matches within the predetermined similarity threshold.
16 . The computer-readable medium of claim 15 , wherein providing the plurality of electronic medical images to the trained machine learning system further comprises:
determining one or more vectors associated with each of the plurality of electronic medical images; and
providing the one or more vectors associated with each of the plurality of electronic medical images to the trained machine learning system.
17 . The computer-readable medium of claim 15 , wherein providing the plurality of electronic medical images to the trained machine learning system further comprises:
receiving a plurality of historical electronic medical images corresponding to one or more individuals.
18 . The computer-readable medium of claim 15 , wherein outputting and/or storing the output indicating whether each pair of the electronic medical images matches within the predetermined similarity threshold further comprises:
flagging any medical image outside of the predetermined similarity threshold for a human technician to perform review.