Abstract: A system and method for automated domain adaptation determination for a deep learning model trained on a training dataset of a source domain and to be applied on a target dataset of a target domain. When labels are available for DT which provide context to the data for the DL model to learn from it, the DL model is trained on the labeled data. With labels available for DT, an evaluation of N supervised domain adaptation methods is automatically launched, including an iterative testing process for testing the N SDA methods on a first test dataset of DT, and automatically selecting the best-performing SDA method for the DA of the trained DL model with respect to DT. The first test dataset comprises at least part of the labeled data of DT. When no labels are available for DT an unsupervised domain adaptation is automatically selected for handling the target dataset.

Abstract: A part of a biological object is imaged with a reduced Image Acquisition Protocol (IAP), obtained by sub-sampling a set of imaging parameters of a rich IAP. The imaging parameters control an acquisition of images by an imaging apparatus. The reduced IAP is used to acquire one or several images of the part of the biological object. The one or several images are input into a trained machine learning (ML) which, for each inputted image, outputs one or several predicted images. The ML model has been trained on input training data and output training data. The input training data correspond to a first set of images obtained via the reduced IAP, and the output training data correspond to a corresponding second set of images obtained via the rich IAP. The one or several predicted images are provided via an interface.

Abstract: A computer-implemented method and system for automated segmentation of anatomical structures of a biological object, include acquiring an MRI image of the object constructed from a set of slices of the object, dividing the set of slices into overlapping groups of consecutive slices, and feeding each overlapping group of consecutive slices as input into a neural network for outputting a labelled map for each inputted slice. For each slice belonging to several overlapping groups, determining for each voxel a final label from specific labels assigned to the voxel by the neural network when considering the labelled maps outputted for the considered slice and assigning to each voxel the final label previously determined for the considered voxel and outputting a final segmentation map of final labels assigned to the voxels of the considered slice. A final 3D segmented image of the object is created from previously obtained final segmentation maps.

Abstract: A system and a method measure volumetric changes of brain structures. The method includes initializing an intensity value of all voxels of a 3D voxel dataset representing the brain of a subject to an initial value preferentially equal to 0. For all voxels that belong to a segmented brain structure for which reference data of a longitudinal reference model exists, automatically executing the following steps: calculating a deviation of a volume change for the segmented brain structure from the longitudinal reference model, normalizing the deviation to obtain a quantitative value of the volume change on a same scale for voxel's belonging to different brain structures; and setting the intensity value of the voxels to the previously obtained quantitative value Q. The voxels of the 3D voxel dataset are displayed in a form of a longitudinal deviation map.

Abstract: A computer-implemented method and system for automated segmentation of anatomical structures of a biological object, include acquiring an MRI image of the object constructed from a set of slices of the object, dividing the set of slices into overlapping groups of consecutive slices, and feeding each overlapping group of consecutive slices as input into a neural network for outputting a labelled map for each inputted slice. For each slice belonging to several overlapping groups, determining for each voxel a final label from specific labels assigned to the voxel by the neural network when considering the labelled maps outputted for the considered slice and assigning to each voxel the final label previously determined for the considered voxel and outputting a final segmentation map of final labels assigned to the voxels of the considered slice. A final 3D segmented image of the object is created from previously obtained final segmentation maps.

Abstract: A system and a method monitor a biological process. The method includes obtaining an abnormal tissue mask from an abnormal tissue segmentation of an image of an object containing tissue to be analyzed, the image being acquired at a time t0 being a reference time point. Other images of the object are registered onto the abnormal tissue mask, the other images being acquired at other time points. Image contrasts of the other images are normalized with respect to the contrasts of the image acquired at the reference time point. The normalized images are subtracted for each available contrast in order to obtain difference images. A joint difference image is created by summing the previously obtained difference images. A biological process progression map is created by overlapping the abnormal tissue mask obtained and the joint difference image after applying a pre-defined threshold.

Abstract: A system and a method measure volumetric changes of brain structures. The method includes initializing an intensity value of all voxels of a 3D voxel dataset representing the brain of a subject to an initial value preferentially equal to 0. For all voxels that belong to a segmented brain structure for which reference data of a longitudinal reference model exists, automatically executing the following steps: calculating a deviation of a volume change for the segmented brain structure from the longitudinal reference model, normalizing the deviation to obtain a quantitative value of the volume change on a same scale for voxel's belonging to different brain structures; and setting the intensity value of the voxels to the previously obtained quantitative value Q. The voxels of the 3D voxel dataset are displayed in a form of a longitudinal deviation map.

Abstract: A system and a method monitor a biological process. The method includes obtaining an abnormal tissue mask from an abnormal tissue segmentation of an image of an object containing tissue to be analyzed, the image being acquired at a time t0 being a reference time point. Other images of the object are registered onto the abnormal tissue mask, the other images being acquired at other time points. Image contrasts of the other images are normalized with respect to the contrasts of the image acquired at the reference time point. The normalized images are subtracted for each available contrast in order to obtain difference images. A joint difference image is created by summing the previously obtained difference images. A biological process progression map is created by overlapping the abnormal tissue mask obtained and the joint difference image after applying a pre-defined threshold.

Abstract: In a magnetic resonance method and apparatus, deficiencies in conventional masking in quantitative susceptibility mapping (QSM) are addressed by the inclusion of an additional step in the conventional QSM post-processing pipeline. In this additional step, atlas-based segmentation techniques, which have been developed for morphological applications such as T1w MPRAGE are used in order to provide the mask. This mask is then fed to the remainder of the QSM post-processing pipeline.

Abstract: In a magnetic resonance method and apparatus, deficiencies in conventional masking in quantitative susceptibility mapping (QSM) are addressed by the inclusion of an additional step in the conventional QSM post-processing pipeline. In this additional step, atlas-based segmentation techniques, which have been developed for morphological applications such as T1w MPRAGE are used in order to provide the mask. This mask is then fed to the remainder of the QSM post-processing pipeline.