Cell Feature-Based Automatic Circulating Tumor Cell Detection

Abstract

An automated method for detecting circulating tumor cells in a microscopic image of a blood sample includes receiving, by a computer, a plurality of low-resolution images, each low resolution image providing a representation of the blood sample with one of a plurality of stains applied. The computer determines a threshold value for each of the plurality of stains based on the low resolution images and identifies a list of potential cells based on the threshold values. A gating process is performed on the list of potential circulating tumor cells to identify one or more likely or highly likely circulating tumor cells. The computer presents the subset of the low-resolution images in a verification interface comprising one or more components allowing a user to confirm that a respective low-resolution image included in the subset of the low-resolution images includes one or more circulating tumor cells.

We claim: 
     
         1 . An automated method for detecting circulating tumor cells in a microscopic image of a blood sample, the method comprising:
 receiving, by a computer, a plurality of low-resolution images, each low resolution image providing a representation of the blood sample with one of a plurality of stains applied;   determining, by the computer, a threshold value for each of the plurality of stains based on the low resolution images;   identifying, by the computer, a list of potential cells based on the threshold values;   performing, by the computer, a gating process on the list of potential circulating tumor cells to identify one or more likely or highly likely circulating tumor cells; and   presenting, by the computer, the subset of the low-resolution images in a verification interface comprising one or more components allowing a user to confirm that a respective low-resolution image included in the subset of the low-resolution images includes one or more circulating tumor cells.   
     
     
         2 . The method of  claim 1 , further comprising:
 identifying, by the computer, user-confirmed low-resolution images via the verification interface;   transferring, by the computer, coordinates associated with the user-confirmed low-resolution images to a microscope;   receiving, by the computer, one or more high resolution images acquired at the coordinates; and   presenting, by the computer, the high resolution images in a new verification interface.   
     
     
         3 . The method of  claim 2 , wherein the new verification interface comprises one or more new components allowing the user to confirm that a respective high-resolution image included in the presented high-resolution images includes one or more circulating tumor cells. 
     
     
         4 . The method of  claim 1 , wherein receiving the plurality of low-resolution images comprises:
 receiving, by the computer, low-resolution image tiles from a microscope, each low-resolution image tile providing a tile representation of the blood sample with one of the plurality of stains applied; and   combining, by the computer, the plurality of low resolution image tiles to create the plurality of low resolution images.   
     
     
         5 . The method of  claim 2 , wherein receiving the one or more high-resolution images comprises:
 receiving, by the computer, high-resolution image tiles from the microscope, each high-resolution image tile providing a tile representation of the blood sample with one of the plurality of stains applied; and   selecting, by the computer, a region of interest from each of the plurality of high-resolution image tiles based on the coordinates to create the plurality of high-resolution images.   
     
     
         6 . The method of  claim 1 , wherein determining the threshold value for each of the plurality of stains based on the low resolution images comprises:
 creating a histogram of intensity values for each low resolution image; and   automatically selecting pixels within a predetermined percentile of the histogram as thresholded values.   
     
     
         7 . The method of  claim 6 , wherein the predetermined percentile is two percent. 
     
     
         8 . The method of  claim 6 , further comprising:
 adjusting, by the computer, one or more of the thresholded values based on user input.   
     
     
         9 . The method of  claim 1 , wherein the gating process comprises:
 determining a circularity value for each potential circulating tumor cell;   for each potential circulating tumor cell, determining a median response for each of the plurality of staining channels;   determining a score for each potential circulating tumor cell based on its respective circularity value and respective median responses for each of the plurality of stains; and   identifying the one or more likely or highly likely circulating tumor cells from the potential circulating tumor cells based the determined scores.   
     
     
         10 . A computer-implemented method for detecting circulating tumor cells in a blood sample, the method comprising:
 receiving, by a computer, a plurality of images, each image providing a representation of the blood sample with one of a plurality of stains applied;   determining, by the computer, a threshold value for each of the plurality of stains applied to the blood sample;   identifying, by the computer, a list of potential circulating tumor cells in the images;   creating, by the computer, a composite mask image using the images and the threshold values;   determining, by the computer, connected components included in the composite mask image;   applying, by the computer, a filter to the connected components to yield filtered components;   selecting, by the computer, a subset of the filtered components from the filtered components; and   applying, by the computer, a heuristic to identify likely or highly likely circulating tumor cells from the subset of filtered components.   
     
     
         11 . The method of  claim 10 , wherein the filter is a micrometer range filter configured filter connected components that are not within a predetermined range. 
     
     
         12 . The method of  claim 11 , wherein the predetermined range is 25 square micrometers to 13,500 square micrometers. 
     
     
         13 . The method of  claim 10 , wherein selecting the subset of the filtered components from the filtered components comprises:
 sorting, by the computer, the filtered components based on criteria related to the staining values to yield sorted filtered components; and   selecting, by the computer, a predetermined number of top sorted filtered components as the subset of filtered components.   
     
     
         14 . The method of  claim 13 , wherein the filtered components are sorted based on a ratio of mean CK to CD45. 
     
     
         15 . The method of  claim 13 , wherein the predetermined number is 50,000. 
     
     
         16 . The method of  claim 10 , wherein the heuristic is based on first measurement values corresponding to the plurality of stains applied to the blood sample and a second measurement value corresponding to circularity of each of the subset of connected components. 
     
     
         17 . A system for detecting circulating tumor cells in a blood sample, the system comprising:
 a storage medium configured to store a plurality of low-resolution images, each low resolution image providing a representation of the blood sample with one of a plurality of stains applied;   a detection processor operably coupled to the storage medium and configured to:
 determine a threshold value for each of the plurality of stains based on the low resolution images, 
 identify a list of potential cells based on the threshold values, and 
 perform a gating process on the list of potential circulating tumor cells to identify one or more likely or highly likely circulating tumor cells; and 
   a display configured to present the subset of the low-resolution images in a verification interface comprising one or more components allowing a user to confirm that a respective low-resolution image included in the subset of the low-resolution images includes one or more circulating tumor cells.   
     
     
         18 . The system of  claim 17 , further comprising a microscope configured to acquire the plurality of low-resolution images. 
     
     
         19 . The system of  claim 18 , wherein the detection processor is further configured to:
 identify user-confirmed low-resolution images via the verification interface;   transfer coordinates associated with the user-confirmed low-resolution images to a microscope; and   receive one or more high resolution images acquired at the coordinates.   
     
     
         20 . The system of  claim 19 , wherein the display is further configured to present the high resolution images in a reconfirmation interface.