An artificial intelligence (AI) platform, method and program product for detecting and sizing a lesion in real time during a clinical procedure. An AI platform is disclosed that includes: a trained classifier that includes a deep learning model trained to detect lesions and reference objects in image data; a real time video analysis system that receives a video feed during a clinical procedure, uses the trained classifier to determine if a video frame from the video feed has both a lesion and a reference object, calculates an actual size of the lesion based on a pixel size of both the lesion and the reference object, and outputs an indication that the lesion was detected and the actual size of the lesion.

Devices and methods for preparing RNA and DNA from single cells are disclosed. In particular, the invention relates to a method of simultaneously extracting RNA and DNA from single cells and separating the nucleic acids electrophoretically. An electric field is used to lyse a single target cell, such that the plasma membrane is selectively disrupted without lysing the nuclear membrane. Cytoplasmic RNA is separated from the nucleus by performing isotachophoresis (ITP) in the presence of a sieving matrix that preferentially reduces the mobility of the nucleus. During ITP, the cytoplasmic RNA accumulates at an ITP interface between leading and trailing electrolytes and can later be extracted free of nuclear DNA. The method can be performed in a microfluidic device that fully automates all steps of the process.

A system, composition and method detect diseases using a method for identifying spectral biomarkers and patterns from stool samples. In one embodiment, the system, composition and method may provide a non-invasive method for detecting colorectal cancer and precancerous polyps comprises subjecting stool samples from cancerous and non-cancerous subjects to hyperspectral spectroscopy and wherein differences in spectra indicates cancer, or assesses risk of development thereof. The system, composition and method may also include a method for identifying spectral biomarkers and patterns from stool samples from cancerous, precancerous and inflammatory bowel disease subjects.

An image processing device includes a processor comprising hardware, wherein the processor is configured to execute: acquiring intraluminal images; generating, for each of the intraluminal images, lesion information by estimating a visual point with respect to a lesion region extracted from each of the intraluminal images and analyzing a three-dimensional structure of the lesion, the lesion information indicating any of a top portion, a rising portion, and a marginal protruding portion in the lesion region; and extracting, based on the lesion information, a target image satisfying a prescribed condition from the intraluminal images.

A method and apparatus for estimating or measuring a physical area or physical volume of an object of interest in one or more images captured using an endoscope are disclosed. According to the present method, one or more structured-light images and one or more regular images captured using an imaging apparatus are received. An object of interest in the regular images is determined. Distance information associated with the object of interest with respect to the imaging apparatus is derived from the structured-light images. The physical area size or physical volume size of the object of interest is determined based on the regular images and the distance information. The imaging apparatus can be a capsule endoscope or an insertion endoscope.

The present disclosure relates to computer-implemented systems and methods for training and using generative adversarial networks to detect abnormalities in images of a human organ. In one implementation, a method is provided for training a neural network system, the method may include applying a perception branch of an object detection network to frames of a first subset of a plurality of videos to produce a first plurality of detections of abnormalities. Further, the method may include using the first plurality of detections and frames from a second subset of the plurality of videos to train a generator network to generate a plurality of artificial representations of polyps, and training an adversarial branch of the discriminator network to differentiate between artificial representations of the abnormalities and true representations of abnormalities. Additionally, the method may include retraining the perception branch based on difference indicators between the artificial representations of abnormalities and true representations of abnormalities included in frames of the second subset of plurality of videos and a second plurality of detections.

Methods, systems, and computer readable media for deriving a three-dimensional (3D) surface from colonoscopic video are disclosed. According to one method for deriving a 3D surface from colonoscopic video, the method comprises: performing video frame preprocessing to identify a plurality of keyframes of a colonoscopic video, wherein the video frame preprocessing includes informative frame selection and keyframe selection; generating, using a recurrent neural network and direct sparse odometry, camera poses and depth maps for the keyframes; and fusing, using SurfelMeshing and the camera poses, the depth maps into a three-dimensional (3D) surface of a colon portion, wherein the 3D surface indicates at least one region of the colon portion that was not visualized.

An image processing system connected to an endoscope and processing in real-time endoscopic images to identify suspicious tissues such as polyps or cancer. The system applies preprocessing tools to clean the received images and then applies in parallel a plurality of detectors both conventional detectors and models of supervised machine learning-based detectors. A post processing is also applied in order select the regions which are most probable to be suspicious among the detected regions. Frames identified as showing suspicious tissues can be marked on an output video display. Optionally, the size, type and boundaries of the suspected tissue can also be identified and marked.

An image processing apparatus includes a processor that detects regions of interest for each of plural of observation images acquired by performing image pickup of an object, sequentially records the plurality of observation images as record images in either one of a first period from a first detection start at which a first region of interest starts to be detected to a first detection cessation at which detection of the first region of interest ceases and a second period from the first detection start to a second detection cessation at which detection of a second region of interest ceases, calculates a display timing at which the plurality of record images start to be reproduced, and performs processing for displaying at least one of the record images on a display screen of a display apparatus while sequentially displaying the plurality of observation images on the display screen at the display timing.

The present disclosure relates to systems and methods for processing real-time video and detecting objects in the video. In one implementation, a system is provided that includes an input port for receiving real-time video obtained from a medical image device, a first bus for transferring the received real-time video, and at least one processor configured to receive the real-time video from the first bus, perform object detection by applying a trained neural network on frames of the received real-time video, and overlay a border indicating a location of at least one detected object in the frames. The system also includes a second bus for receiving the video with the overlaid border, an output port for outputting the video with the overlaid border from the second bus to an external display, and a third bus for directly transmitting the received real-time video to the output port.