The invention relates to a method that allows a set of embryos to be ranked on the basis of ploidy potential and/or pregnancy generation potential, to aid the process of selecting embryos for transfer in an in-vitro fertilisation procedure. The method measures properties or characteristics of the entire blastocyst; extracts characteristics by identifying different cell types, mainly blastocyst structures and patterns, without extracting characteristics of the first cell divisions and the behaviour thereof over time; and predicts the prognosis of pregnancy and/or ploidy (result of genetic study and successful implantation), using micrographs standardised for the management thereof and by means of sequential preprocessing and machine learning algorithms implemented in a computer in order to rank the potential of a set of embryos, to obtain a successful, live, full-term pregnancy.

An ultrasonic device and a method and system for transforming the display of a three-dimensional ultrasonic image thereof. The method comprises: first acquiring original ultrasonic three-dimensional body data including a tested object (S10); detecting the orientation of the tested object from the original ultrasonic three-dimensional body data according to image features of the tested object (S20); then comparing the orientation of the tested object with a desired orientation, so as to obtain a rotation transformation parameter in a three-dimensional coordinate system (S30); next, rotating and transforming the original ultrasonic three-dimensional body data according to the rotation transformation parameter, so as to obtain the transformed three-dimensional body data (S40); and finally, outputting the transformed three-dimensional body data (S50). Therefore, the ultrasonic three-dimensional body data of the tested object can automatically rotate to the desired orientation without requiring manual adjustment, thereby improving the efficiency thereof.

The present invention extends to methods, systems, and computer program products for estimating oocyte quality. A machine learning algorithm accesses oocyte training data for a mammalian species (e.g., humans) and trains a neural network to estimate oocyte quality for the mammalian species based on the oocyte training data. The neural network accesses a microscopic image of an oocyte and identifies oocyte features of the oocyte. Based on the identified oocyte features, the neural network estimates oocyte quality, including: (a) predicting a probability of a corresponding embryo maintaining sufficient developmental competence until a specified time after fertilization and (b) predicting another probability of the corresponding embryo reaching a specific embryonic stage after fertilization. An oocyte is selected, from among a plurality of human oocytes including the human oocyte, for a potential recipient based at least in part on the oocyte quality, including based on the probability and the other probability.

The present embodiments relate generally to systems and methods for performing a measurement on an ultrasound image displayed on a touchscreen device. The method may include: receiving, via the touchscreen device, first input coordinates corresponding to a point on the ultrasound image; using the first input coordinates as a seed for performing a contour identification process on the ultrasound image, wherein the contour identification process performs contour evolution using morphological operators to iteratively dilate from the first input coordinates; upon identification of a contour from the contour identification process, placing measurement calipers on the identified contour; and storing a value identified by the measurement calipers as the measurement.

A method includes: acquiring a training data set including pieces of training data, each of the pieces including an image of a training target, first annotation data representing a rectangular region in the image, and second annotation data; training, based on the image and the first annotation data, an object detection model specifying a rectangular region including the training target; training, based on the image and the second annotation data, a neural network; and calculating a first index value related to a relationship of a pixel number, the trained estimation model and the calculated first index value being used in a determination process that determines, based on the calculated first index value and a second index value relationship between a pixel number in an output result and an estimation result, whether or not a target in a target image is normal.

A system is described for generating diagnostic information from a video sequence of ultrasound images acquired in “blind sweeps”, i.e., without operator seeing ultrasound images as they are acquired. We disclose two different types of machine learning systems for predicting diagnostic information: a “Temporal Accumulation” system and a “3-D Modeling Component” system. These machine learning systems could be implemented in several possible ways: using just one or the other of them in any given implementation, or using both of them in combination. We also disclose a computing system which implements (a) an image selection system including at least one machine learning model trained to identify clinically suitable images from the sequence of ultrasound images and (b) an image diagnosis/measurement system including of one or more machine learning models, configured to obtain the clinically suitable images identified by the image selection system and further process such images to predict health states.

A medical image recognition method includes the following steps: establishing an image recognition model, wherein the image recognition model is generated by inputting a plurality of labeled medical image slices in a plurality of initial medical image piles into a neural network; and in response to determining that the accuracy of the image recognition model is not higher than an accuracy threshold; calculating a plurality of image change rates corresponding to each of a plurality of initial medical image slices or the initial medical image piles formed by the initial medical image slices according to the image recognition model; selecting at least one of the initial medical image piles or the initial medical image slices as a training medical image slice according to the image change rates; obtaining the target range of each training medical image slice to re-establish the image recognition model.

Systems and methods for automated imaging and evaluation of image based features are disclosed herein. Method for automated imaging and evaluation of image based features can include receiving time-lapse images of at least one human embryo contained in a multi-well culture dish that can have a plurality of micro-wells. Image based features can be automatically generated from the time-lapse images of the human embryo. The image based features, which can include a cavitation feature, can be inputted into a classifier. The classifier can automatically and directly generate a viability prediction with the classifier from the image-based features.

Automated assessment for a fetus may be applied based on imaging data obtained during medical imaging examination of the fetus, with the applying including processing imaging data corresponding to a plurality of a cross-section imaging slices corresponding to a limb of the fetus, where the processing includes for each imaging slice: automatically generating a predicted outer mask for an outer contour of the limb based on application of a first pre-trained model to imaging data corresponding to the imaging slice; and automatically generating a segmentation of fat-lean mask for the imaging slice based on application of a second pre-trained model to both of the imaging data corresponding to the imaging slice and the generated predicted output mask; and applying based on the processing of the imaging data corresponding to the plurality of a cross-section imaging slices: a fractional limb volume assessment; and a fat-lean mass assessment.

An Artificial Intelligence (AI) based computational system is used to non-invasively estimate the presence of a range of aneuploidies and mosaicism in an image of embryo prior to implantation. Aneuploidies and mosaicism with similar risks of adverse outcomes are grouped and training images are labelled with their group. Separate AI models are trained for each group using the same training dataset and the separate models are then combined, such as by using an Ensemble or Distillation approach to develop a model that can identify a wide range of aneuploidy and mosaicism risks. The AI model for a group is generated by training multiple models including binary models, hierarchical layered models and a multi-class model. In particular the hierarchical layered models are generated by assigning quality labels to images. At each layer the training set is partitioned in the best quality images and other images. The model at that layer is trained on the best quality images, and the other images are passed down to the next layer and the process repeated (so the remaining images are separated into next best quality images and other images). The final model can then be used to non-invasively identify aneuploidy and mosaicism and associated risk of adverse outcomes from an image of an embryo prior to implantation.