Methods and systems for identifying a region of interest in breast tissue utilize artificial intelligence to confirm that a target lesion identified during imaging the breast tissue using a first imaging modality (e.g. x-ray imaging) has been identified using a second imaging modality (e.g. ultrasound imaging). A computing system operating a lesion matching engine utilizes a machine learning classifier algorithm trained on cases of x-ray images and corresponding ultrasound images in which lesions were identified for further analysis. The lesion matching engine analyzes a target lesion identified with x-ray imaging and a potential lesion identified with ultrasound imaging to determine a likelihood that the target lesion is the same as the potential lesion. A confidence level indicator for the lesion match is presented on a display of a computing device to aid a healthcare provider in locating a lesion in breast tissue.

An image processing system includes an interface to which an annotation result on a learning image captured inside a living body is input and a processor including hardware. The processor acquires metadata including difficulty information indicating difficulty of the annotation of the learning image itself, determines reliability information indicating reliability of the annotation result based on the metadata, and outputs a dataset in which the learning image, the annotation result, and the reliability information are associated with each other, as data used in generating a trained model used in inference based on deep learning on an inference target image captured inside a living body.

An analysis method for breast image and an electronic apparatus using the same are provided. The method includes the following steps. A breast image scanned by an ultrasound wave is obtained. Based on rectangular features of the breast image, a region of interest including an aberrant region in the breast image is obtained by applying a detection model. The aberrant region is further acquired from the region of interest, and a plurality of feature parameters of the aberrant region are extracted for a property analysis of the aberrant region.

Provided is a method and apparatus for providing information needed for the diagnosis of lymph node metastasis of a thyroid cancer, and the method includes the steps of: acquiring medical images produced correspondingly to the continuous volumes of a body region including the neck; detecting at least one or more lymph nodes from the medical images through a first network function learned, the lymph nodes including at least one or more lymph nodes having higher lymph node metastasis risks than a given reference value; dividing the neck tissue around the thyroid into a plurality of compartments on the medical images through a second network function learned, based on the anatomical characteristics of the neck tissue; and matching diagnostic information including the information of the detected lymph nodes and the plurality of compartments with the medical images and displaying the diagnostic information on the medical images.

Methods, apparatus, and other embodiments predict response to pemetrexed based chemotherapy. One example apparatus includes an image acquisition circuit that acquires a radiological image of a region of tissue demonstrating NSCLC that includes a region of interest (ROI) defining a tumoral volume, a peritumoral volume definition circuit that defines a peritumoral volume based on the boundary of the ROI and a distance, a feature extraction circuit that extracts a set of discriminative tumoral features from the tumoral volume, and a set of discriminative peritumoral features from the peritumoral volume, and a classification circuit that classifies the ROI as a responder or a non-responder using a machine learning classifier based, at least in part, on the set of discriminative tumoral features and the set of discriminative peritumoral features.

State of the art deep network based Universal Lesion Detection (ULD) techniques inherently depend on large number of datasets for training the systems. Moreover, these system are specifically trained for lesion detection in organs of a Region of interest (RoI) of a body. Thus, requires retraining when the RoI varies. Embodiments herein disclose a method and system for domain knowledge augmented multi-head attention based robust universal lesion detection. The method utilizes minimal number of Computer Tomography (CT) scan slices to extract maximum information using organ agnostic HU windows and a convolution augmented attention module for a computationally efficient ULD with enhanced prediction performance.

Disclosed is a system and a method for determining a brock score. A CT scan image may be resampled into a plurality of slices using a bilinear interpolation. A nodule may be detected on one or more of the plurality of slices. A region of interest associated with the nodule may be identified using an image processing technique. Further, a nodule segmentation may be performed to remove an area surrounding the region of interest. Subsequently, a plurality of characteristics associated with the nodule may be identified automatically using a deep learning model. Finally, a brock score for the patient may be determined based on the plurality of characteristics and demographic data of the patient.

Computational techniques are applied to video images of polyps to extract features and patterns from different perspectives of a polyp. The extracted features and patterns are synthesized using registration techniques to remove artifacts and noise, thereby generating improved images for the polyp. The generated images of each polyp can be used for training and testing purposes, where a machine learning system separates two types of polyps.

A method of forming a probability map is disclosed. According to one embodiment, a method may include: (1) obtaining multiple measures of multiple imaging parameters for every stop of a moving window on an image, wherein two neighboring ones of the stops of the moving window are partially overlapped with each other; (2) obtaining first probabilities of an event for the stops of the moving window by matching the measures of the imaging parameters to a classifier; and (3) obtaining second probabilities of the event for multiple voxels of a probability map based on information associated with the first probabilities.

The present disclosure relates to a method for training a classifier. The method includes: acquiring an original image; determining a candidate target by segmenting the original image based on at least two segmentation models; determining a universal set of features by extracting features from the candidate target; determining a reference subset of features by selecting features from the universal set of features; and determining a classifier by performing classifier training based on the reference subset of features.