Dynamic self-learning medical image method and system
A method and system for creating a dynamic self-learning medical image network system, wherein the method includes receiving, from a first node initial user interaction data pertaining to one or more user interactions with the one or more initially obtained medical images; training a deep learning algorithm based at least in part on the initial user interaction data received from the node; and transmitting an instance of the trained deep learning algorithm to the first node and/or to one or more additional nodes, wherein at each respective node to which the instance of the trained deep learning algorithm is transmitted, the trained deep learning algorithm is applied to respective one or more subsequently obtained medical images in order to obtain a result.
1 . A learning database for creating digital patterns found in a breast, comprising computer executable instructions stored in a non-transitory memory, for execution by a processor, forming:
a data storage space associated with a central brain server and in communication with a deep learning algorithm, the data storage space configured to:
store a set of basic training data for the deep learning algorithm, the training data comprising breast images having digital patterns indicative of a plurality of abnormal and normal objects in the breast;
obtain user interaction data from each of a plurality of distributed nodes, wherein the user interaction data comprises:
observed user interactions with breast images at the plurality of distributed nodes, and
digital patterns or features detected based on the observed user interactions with the breast images;
store the user interaction data; and
continuously update the basic training data for each digital pattern of the digital patterns indicative of a plurality of abnormal and normal objects in the breast according to the user interaction data to yield updated training data.
2 . The learning database of claim 1 , wherein the deep learning algorithm generates data models for detecting recurring patterns in medical images based on the updated training data.
3 . The learning database of claim 2 , wherein the data storage space is further configured to store and update the data models for detecting recurring patterns in medical images.
4 . The learning database of claim 2 , wherein communication between the data storage space and the deep learning algorithm includes:
the data storage space providing the updated training data to the deep learning algorithm; and
the data storage space receiving the data models for detecting recurring patterns in medical images from the deep leaning algorithm.
5 . The learning database of claim 2 , wherein the recurring patterns in medical images relate to one or more of abnormalities in a breast and variations among healthy breast tissues.
6 . The learning database of claim 5 , wherein the variations among healthy breast tissues are associated in the data storage for patient populations.
7 . The learning database of claim 2 , wherein the recurring patterns relating to one or more abnormalities include at least one of cysts, tumors, abnormal masses, spiculated masses, and calcifications.
8 . The learning database of claim 7 , wherein the recurring patterns relating to one or more abnormalities comprise feature values related to abnormalities, the feature values including at least one coordinates, grayscale values, and contrast values.
9 . The learning database of claim 1 , wherein the basic training data includes medical image data including image based data and non-image based data associated with one or more medical images.
10 . The learning database of claim 1 , wherein the updated training data is transmitted to at least one node of the plurality of distributed nodes.
11 . The learning database of claim 1 , wherein the user interaction data comprises one or more of:
(a) an actual or estimated portion of at least one of the one or more medical images that was focused upon by at least one user;
(b) at least one annotation on at least one of the one or more initially obtained medical images;
(c) a selection of one of more pixels associated with at least one of the one or more initially obtained medical images;
(d) an actual or estimated amount of time one or more users spent viewing one or more of the initially obtained medical images; and
(e) a description of a patient condition and/or written or recorded audio diagnostic findings.
12 . A method of preparing and maintaining a learning database for creating digital patterns found in a breast, the method comprising:
storing, in a data storage space associated with a central brain server and in communication with a deep learning algorithm, a set of basic training data for the deep learning algorithm, the training data comprising breast images having digital patterns indicative of a plurality of abnormal and normal objects in the breast;
obtaining user interaction data from each of a plurality of distributed nodes, wherein the user interaction data comprises:
observed user interactions with breast images at the plurality of distributed nodes, and
digital patterns or features detected based on the observed user interactions with the breast images;
storing the user interaction data; and
continuously updating the basic training data for each digital pattern of the digital patterns indicative of a plurality of abnormal and normal objects in the breast according to the user interaction data to yield updated training data.
13 . The method of claim 12 , wherein the deep learning algorithm generates data models for detecting recurring patterns in medical images based on the updated training data.
14 . The method of claim 13 , wherein the data storage space is further configured to store and update the data models for detecting recurring patterns in medical images.
15 . The method of claim 13 , wherein communication between the data storage space and the deep learning algorithm includes:
the data storage space providing the updated training data to the deep learning algorithm; and
the data storage space receiving the data models for detecting recurring patterns in medical images from the deep leaning algorithm.
16 . The method of claim 13 , wherein the recurring patterns in medical images relate to one or more of abnormalities in a breast and variations among healthy breast tissues, wherein the recurring patterns relating to one or more abnormalities include at least one of cysts, tumors, abnormal masses, spiculated masses, and calcifications and the variations among healthy breast tissues are associated in the data storage for patient populations.
17 . The method of claim 16 , wherein the recurring patterns relating to one or more abnormalities comprise feature values related to abnormalities, the feature values including at least one coordinates, grayscale values, and contrast values.
18 . The method of claim 12 , wherein the updated training data is transmitted to at least one node of the plurality of distributed nodes.
19 . The method of claim 12 , wherein the user interaction data comprises one or more of:
(a) an actual or estimated portion of at least one of the one or more medical images that was focused upon by at least one user;
(b) at least one annotation on at least one of the one or more initially obtained medical images;
(c) a selection of one of more pixels associated with at least one of the one or more initially obtained medical images;
(d) an actual or estimated amount of time one or more users spent viewing one or more of the initially obtained medical images; and
(e) a description of a patient condition and/or written or recorded audio diagnostic findings.
20 . A system for preparing and maintaining a learning database for creating digital patterns found in a breast, the system comprising:
one or more processors;
a non-transitory memory in communication with the one or more processors and including instructions that, when executed by the processor, cause the processor to:
store, in a data storage space associated with a central brain server and in communication with a deep learning algorithm, a set of basic training data for the deep learning algorithm, the training data comprising breast images having digital patterns indicative of a plurality of abnormal and normal objects in the breast;
obtain user interaction data from each of a plurality of distributed nodes, wherein the user interaction data comprises:
observed user interactions with breast images at the plurality of distributed nodes, and
digital patterns or features detected based on the observed user interactions with the breast images;
store the user interaction data; and
continuously update the basic training data for each digital pattern of the digital patterns indicative of a plurality of abnormal and normal objects in the breast according to the user interaction data to yield updated training data.