Machine learning model for analyzing pathology data from metastatic sites

Abstract

Described herein are systems and methods of determining primary sites from biomedical images. A computing system may identify a first biomedical image of a first sample from one of a primary site or a secondary site associated with a condition in a first subject. The computing system may apply the first biomedical image to a site prediction model comprising a plurality of weights to determine the primary site for the condition. The computing system may store an association between the first biomedical image and the primary site determined using the site prediction model.

What is claimed is: 
     
         1 . A method, comprising:
 identifying, by a computing system, a first biomedical image of a first sample obtained from one of a plurality of sites associated with a condition in a first subject, the plurality of sites including (i) a first organ from which the condition originated and (ii) a second organ to which the condition spread from the first organ;   determining, by the computing system, a site corresponding to the first organ for the first subject by applying the first biomedical image to a machine learning (ML) model, wherein the ML model is trained using a plurality of examples, each example of the plurality of examples including (i) a respective second biomedical image from one of the plurality of sites associated with the condition in a corresponding second subject and (ii) a respective identification of the first organ for the corresponding second subject; and   storing, by the computing system, for the first subject, an association between the first biomedical image and the determined site.   
     
     
         2 . The method of  claim 1 , further comprising providing, by the computing system, an output based on the determined site having the association with the first biomedical image. 
     
     
         3 . The method of  claim 1 , further comprising presenting, by the computing system, information identifying the determine site and at least one of the first sample, the first biomedical image, or the first subject. 
     
     
         4 . The method of  claim 1 , wherein determining further comprises identifying a plurality of candidate sites for the first organ from which the condition originated for the first subject. 
     
     
         5 . The method of  claim 1 , wherein determining further comprises generating a confidence score indicating a degree of likelihood that the site is the first organ from which the condition originated for the first subject. 
     
     
         6 . The method of  claim 1 , wherein determining further comprises determining, for each condition of a plurality of conditions, the site for the first organ from which the condition originated for the subject. 
     
     
         7 . The method of  claim 1 , wherein identifying further comprises obtaining the first biomedical image of the first sample via a histological image preparer. 
     
     
         8 . A method, comprising:
 identifying, by a computing system, a training dataset including a plurality of examples, each example of the plurality of examples including: (i) a respective biomedical image from one of a plurality of sites associated with a condition in a corresponding subject, the plurality of sites including (a) a first organ from which the condition originated and (b) a second organ to which the condition spread from the first organ; and (ii) a respective identification of a first organ from which the condition originated for the corresponding subject;   determining, by the computing system, for each example of the plurality of examples, a second site corresponding to the first organ for the corresponding subject by applying the first biomedical image to a machine learning (ML) model comprising a plurality of weights;   comparing, by the computing system, for each example of the plurality of examples, the second site determined by applying the ML model with the first site of the respective identification in the training dataset; and   updating, by the computing system, at least one of the plurality of weights of the ML model based on the comparison between the second site with the first site of the respective identification in each example of the plurality of examples.   
     
     
         9 . The method of  claim 8 , further comprising reapplying, by the computing system, the respective biomedical image of at least one example of the plurality of examples to the ML model, responsive to determining that a loss metric for the at least one example exceeds a threshold. 
     
     
         10 . The method of  claim 8 , wherein determining further comprises generating, for each example of the plurality of examples, a confidence score indicating a degree of likelihood that the second site is the first organ from which the condition originated for the respective subject. 
     
     
         11 . The method of  claim 8 , wherein comparing further comprises determining at least one loss metric indicating a degree of deviation between the second site determined by applying the ML model and the first site of the respective identification in the training dataset. 
     
     
         12 . The method of  claim 8 , wherein updating further comprises updating at least one of the plurality of weights of the ML model in accordance with a classification loss between the second site determined by applying the ML model and the first site of the respective identification in the training dataset. 
     
     
         13 . The method of  claim 8 , wherein each example of the plurality of examples in the training dataset further comprises a respective identification of a third site corresponding to the second organ to which the condition spread for the corresponding subject. 
     
     
         14 . The method of  claim 8 , further comprising storing, by the computing system, using one or more data structures, the plurality of weights of the ML model to apply to an acquired biomedical image from a sample of a subject to determine the first organ from which the condition originated for the subject. 
     
     
         15 . A system, comprising:
 a computing system having one or more processors coupled with memory, configured to:
 identify a first biomedical image of a first sample obtained from one of a plurality of sites associated with a condition in a first subject, the plurality of sites including (i) a first organ from which the condition originated and (ii) a second organ to which the condition spread from the first organ; 
 determine a site corresponding to the first organ for the first subject by applying the first biomedical image to a machine learning (ML) model, wherein the ML model is trained using a plurality of examples, each example of the plurality of examples including (i) a respective second biomedical image from one of the plurality of sites associated with the condition in a corresponding second subject and (ii) a respective identification of the first organ for the corresponding second subject; and 
 store, for the first subject, an association between the first biomedical image and the determined site. 
   
     
     
         16 . The system of  claim 15 , wherein the computing system is further configured to provide an output based on the determined site having the association with the first biomedical image. 
     
     
         17 . The system of  claim 15 , wherein the computing system is further configured to present information identifying the determine site and at least one of the first sample, the first biomedical image, or the first subject. 
     
     
         18 . The system of  claim 15 , wherein the computing system is further configured to identify, by applying the first biomedical image to the ML model, a plurality of candidate sites for the first organ from which the condition originated for the first subject. 
     
     
         19 . The system of  claim 15 , wherein the computing system is further configured to generate, by applying the first biomedical image to the ML model, a confidence score indicating a degree of likelihood that the site is the first organ from which the condition originated for the first subject. 
     
     
         20 . The system of  claim 15 , wherein the computing system is further configured to determine, by applying the first biomedical image to the ML model, for each condition of a plurality of conditions, the site for the first organ from which the condition originated for the subject.