Systems and methods for generating synthesized medical images for training a machine learning based network. An input medical image in a first modality is received comprising a nodule region for each of one or more nodules, a remaining region and an annotation for each of the nodules. A synthesized medical image in a second modality is generated from the input medical image comprising the annotation for each of the nodules. A synthesized nodule image of each of the nodule regions and synthesized remaining image of the remaining region are generated in the second modality. It is determined whether a particular nodule is visible in the synthesized medical image based on the synthesized nodule image for the particular nodule and the synthesized remaining image. If at least one nodule is not visible in the synthesized medical image, the annotation for the not visible nodule is removed from the synthesized nodule image.

An object detection device that detects a specific object included in an input image includes a first candidate region specifying unit that specifies a first candidate region in which an object candidate is included from a first input image obtained by imaging a subject in a first posture, a second candidate region specifying unit that specifies a second candidate region in which an object candidate is included from a second input image obtained by imaging the subject in a second posture different from the first posture, a deformation displacement field generation unit that generates a deformation displacement field between the first input image and the second input image, a coordinate transformation unit that transforms a coordinate of the second candidate region to a coordinate of the first posture based on the deformation displacement field, an association unit that associates the first candidate region with the transformed second candidate region that is close to the first candidate region, and a same object determination unit that determines that the object candidates included in the candidate regions associated with each other by the association unit are the same object and are the specific object.

Disclosed are methods and systems for processing medical image data. The method comprising inputting, with one or more processors of one or more computation devices, medical image data into a model for nodule detection; calculating, for at least one nodule detected by the model for nodule detection, a nodule histogram of all voxel intensities of said nodule; determining, from each nodule histogram, a nodule classification among a plurality of nodule classifications for the at least one nodule.

A learning device includes at least one processor. The processor acquires a medical image to be detected, acquires apparatus identification information for identifying an imaging apparatus that has captured the medical image to be detected, selects any one of a plurality of lesion detection models, which detect a lesion from the medical image, on the basis of the apparatus identification information, and detects the lesion from the medical image to be detected, using the selected lesion detection model.

Systems and methods for automatic assessment of a lesion are provided. One or more input medical images of a vessel of a patient is received. A lesion is defined in the one or more input medical images. A region of interest around the lesion is defined in the one or more input medical images. Radiomic features are extracted from the region of interest. An assessment of the lesion is determined using a machine learning based classifier network based on the radiomic features. The assessment of the lesion is output.

There are provided an image processing device, an image processing method, and a program that can accurately select normal images suitable for learning without visually selecting learning data by a human.

An image processing device includes a processor, and the processor acquires an examination video picked up by an endoscope apparatus, acquires instructional information of the endoscope apparatus in the pickup of the examination video, specifies a learning frame section from a plurality of frames forming the examination video on the basis of the instructional information, and outputs a frame group of the learning frame section as first learning data.

Systems, apparatuses, and methods for more efficiently and effectively determining the accuracy with which a human evaluates a set of data, as this may reduce the error in the assessment of a model's performance. This can be helpful in situations where human inputs are used to confirm the output of a machine learning generated classification and in situations where it is desired to evaluate the accuracy of data that may have been labeled or annotated by a human curator. This may assist in reducing the need for new validation/test data when evaluating a new model. The system and methods described can be used to evaluate the accuracy of a trained Machine Learning (ML) model, and as a result, allow a comparison between models based on different ML algorithms.

In one embodiment, a method includes, receiving a digital pathology image of a tissue sample and subdividing the digital pathology image into a plural in of patches. For each patch of the plurality of patches, the method includes identify an image feature detected in the patch and generating one or more labels corresponding to the image feature identified in the patch using a machine-learning model. The method includes determining, based on the generated labels, a heterogeneity metric for the tissue sample. The method includes generating an assessment of the tissue sample based on the heterogeneity metric.

A computer-implemented method for detecting tumor subclones may include receiving one or more digital images into a digital storage device, the one or more digital images including images of a tumor of a patient, detecting one or more neoplasms in the one or more received digital images for each patient, extracting one or more visual features from each detected neoplasm, determining a hierarchy dendrogram based on the detected one or more neoplasms and the extracted one or more visual features for each detected neoplasm, determining one or more leaf nodes based on the determined hierarchy dendrogram, and determining whether there are two or more neoplasms among the detected one or more neoplasms that originated independently.

A method for identifying a mutational signature may include receiving one or more digital images into electronic storage for at least one patient, identifying one or more neoplasms in each received digital image, extracting one or more visual features from each identified neoplasm, and applying a trained machine learning system to identify a mutational signature ratio vector for the one or more extracted visual features.