Systems and methods for detecting potential malignancies

A method for improving malignancy detection in individuals comprising:

• receiving, in real-time, spatially arranged image data associated with a plurality of images from an imaging device;
• generating a plurality of processed images by processing the spatially arranged image data through one or more data structures storing one or more portions of executable logic included in a first machine learning model trained to identify one or more pixels of the spatially arranged image data representing a location of a potential malignancy in an individual and a second machine learning model trained to detect a surgical tool in the image as the spatially arranged image data is received from the imaging device;
• determining, in real-time, that one or more processed images of the plurality of processed images satisfy a data quality metric;
• based on determining that the one or more processed images satisfy the data quality metric, causing, in real-time, a display unit to display the one or more the processed images, wherein the one or more processed images include one or more graphical overlays;
• detecting, by the second machine learning model, that the surgical tool is present in at least one of the processed images that satisfy the data quality metric while the surgical tool is actually present in a surgical space associated with the potential malignancy;
• and
• in response to the detecting,
• altering the one or more graphical overlays from the at least one of the processed images to avoid hindering a view of the surgical tool in the at least one of the processed images, wherein the altering of the one or more graphical overlays enables the graphical overlay to be removed while the surgical tool is being detected, and
• displaying an indicator when the one or more graphical overlays are displayed for less than a threshold amount of time;
• wherein the one or more graphical overlays are displayed when the surgical tool is no longer present in the at least one of the processed images and the one or more pixels of the spatially arranged image data representing the location of the potential malignancy are present;
• wherein each of the one or more graphical overlays isolates one or more image locations;
• and
• wherein each of the one or more image locations represents at least one pixel representing a location of a potential malignancy in the individual.

A system, comprising:

• a video capture card configured to receive, from an endoscopic processing unit, spatially arranged image data captured by an endoscopic imaging device of the endoscopic processing unit;
• a data processing system comprising:
• a display unit;
• a computer-readable memory comprising computer-executable instructions;
• and
• at least one processor executing executable logic of a machine learning system including a first machine learning model trained to identify visual representations of malignancies in an image and a second machine learning model trained to detect a surgical tool in the image, wherein when the at least one processor is executing the computer-executable instructions, the at least one processor carries out operations comprising:
• receiving, in real-time, spatially arranged image data associated with a plurality of images from an imaging device;
• generating a plurality of processed images by processing the spatially arranged image data through one or more data structures storing one or more portions of executable logic included in the first machine learning model to identify one or more pixels of the spatially arranged image data representing a location of a potential malignancy in an individual as the spatially arranged image data is received from the imaging device;
• determining, in real-time, that one or more processed images of the plurality of processed images satisfy a data quality metric;
• based on determining that the one or more processed images satisfy the data quality metric, causing, in real-time, a display unit to display the one or more processed images, wherein the one or more processed images include one or more graphical overlays generated by the at least one processor;
• detecting, by the second machine learning model, that the surgical tool is present in at least one of the processed images that satisfy the data quality metric while the surgical tool is actually present in a surgical space associated with the potential malignancy;
• and
• in response to the detecting,
• altering the one or more graphical overlays from the at least one of the processed images to avoid hindering a view of the surgical tool in the at least one of the processed images, wherein the altering of the graphical overlay enables the one or more graphical overlays to be removed while the surgical tool is being detected, and
• displaying an indicator when the one or more graphical overlays are displayed for less than a threshold amount of time;
• wherein the one or more graphical overlays are displayed when the surgical tool is no longer present in the at least one of the processed images and the one or more pixels of the spatially arranged image data representing the location of the potential malignancy are present;
• wherein each of the one or more graphical overlays isolates one or more image locations;
• and
• wherein each of the one or more image locations represents at least one pixel representing a location of a potential malignancy in the individual;
• and
• a display switch configured to be communicatively coupled to the display unit, the data processing system, and the endoscopic processing unit, the display switch being further configured to allow displaying of either the plurality of processed images or a plurality of original images corresponding to the spatially arranged image data.