Abstract: Various examples are provided related to dynamic brain parcellation. In one example, a method for functional task prediction with dynamic supervoxel parcellation includes preprocessing activation data obtained from brains of multiple subjects to generate one or more dynamic parcellated supervoxel maps of the brain, the activation data associated with a functional task, and determining an anatomical location of the functional task in the brain of another subject based upon classification of supervoxels of the one or more dynamic parcellated supervoxel maps. In another example, a system includes at least one computing device that can preprocess activation data to generate one or more dynamic parcellated supervoxel maps of the brain, the activation data associated with a functional task, and determine an anatomical location of the functional task based upon classification of supervoxels of the one or more dynamic parcellated supervoxel maps.

Abstract: Various examples are provided related to predicting perfusion images from non-contrast scans. In one example, a method for predicting perfusion images includes generating perfusion maps of an organ of a subject from non-contrast computed tomography (NCCT) slices of the organ; processing the perfusion maps based upon weights determined by a Physicians-in-the-Loop (PILO) module; and generating synthetic computed tomography perfusion (CTP) maps from the processed perfusion maps, the synthetic CTP maps generated by deep learning-based multimodal image translation. In another example, a system includes at least one computing device that can generate prefusion maps of an organ from NCCT slices; process the perfusion maps based upon weights determined by a PILO module; and generate synthetic CTP maps from the processed perfusion maps using deep learning-based multimodal image translation. The CTP maps can be rendered for display to a user.

Abstract: Various examples are provided related to machine learning for Parkinson's Disease diagnosis, systems, and methods. In one example, a method for identification of onset or presence of Parkinson's disease (PD) includes receiving a retinal image that has been acquired by an image acquisition system; processing the acquired retinal image using one or more trained machine learning models to classify one or more retinal features contained in the acquired retinal image; and predicting, by the processing circuitry, whether the retinal image is indicative of an onset or presence of PD in the human subject based on the classification. A machine learning system can perform trained machine learning. Machine learning models can trained on stored retinal images obtained from a group of subjects who have previously been diagnosed as having PD and a group of subjects who have not previously been diagnosed as having PD.

Abstract: Various examples are provided related to reconstructing images such as, e.g., medical images from low-dose image scans. Adversarial learning such as, e.g., a Cyclic Simulation and Denoising (CSD) framework can be used to address challenges of complicated mixed noise in real low-dose scans. The CSD framework can include a simulator model that can extract low-dose noise and features (e.g., tissue features) from separate image spaces into a unified feature space and a denoiser model that can learn how to remove noise and restore features, simultaneously. Both the simulator model and the denoiser model can regularize each other in a cyclic manner to optimize network learning effectively. The CSD framework in combination with phantom scans can embrace the realistic low-dose noise and features into a unified learning environment to address the challenge of real low-dose image restoration.

Abstract: A method is provided for precision dosing of electrical stimulation of the brain. The method includes determining a location of each voxel of a plurality of voxels in a reference frame of an electro-stimulation device including a plurality of electrodes positioned on a head of a subject. The method also includes obtaining measurements that indicate a tissue type at each voxel inside the head of the subject based on an imaging device. The method also includes determining, with a processor, a value of one or more parameters of the electro-stimulation device based on the tissue type measurements at each voxel such that the electro-stimulation device is configured to generate a value of one or more parameters of an electric field at each voxel inside the head of the subject to improve the treatment outcome of the subject.

Abstract: A machine learning system and method are disclosed that enable full automation of the process of analyzing retinal fundus images to predict Alzheimer's disease, thereby obviating the need for manual labeling of retinal features while also improving prediction accuracy. A machine learning system and method are disclosed that classify retinal features and predict, based on the classified retinal features, the onset or presence of Alzheimer's disease in a human subject. The system comprises a processor configured to perform one or more machine learning models and a memory device in communication with the processor. The machine learning model(s) is trained to process retinal fundus images acquired by an image acquisition system to classify retinal features contained in the images and to predict, based on the classified retinal features, whether the images are indicative of the presence or onset of Alzheimer's disease.

Abstract: Embodiments of the present disclosure are directed to training a neural network for ocular cup (OC) or ocular disc (OD) detection. One such method comprises initiating training of a first network to learn detection of OC/OD regions within a labeled source sample from a source domain; sharing training weights of the first network with a second network; initiating training of the second network to learn detection of OC/OD regions within an unlabeled sample from a target domain; transferring average training weights of the second network to a third network; initiating training of the third network to learn detection of OC/OD regions within an unlabeled sample from the target domain; computing a mean square error loss between the third network and the second network for a same target sample; and adjusting training weights of the second network based on the mean square error loss computation.

Abstract: Various examples related to CT imaging using multimodal CT image super-resolution are provided. In one example, a method includes generating an enhanced super-resolution generative adversarial network (ESRGAN) by training a generative adversarial network (GAN) with a plurality of CT image modalities (e.g., non-contrast CT, CT Perfusion, CT Angiography, CT with contrast-enhanced, etc.) and generating an enhanced CT image by applying the ESRGAN to a low resolution CT image. In another example, a system includes at least one computing device and program instructions stored in memory and executable in the at least one computing device that, when executed, cause the at least one computing device to generate an ESRGAN by training a GAN with a plurality of CT image modalities and generate an enhanced CT image by applying the ESRGAN to a low resolution CT image.

Abstract: Various embodiments for image denoising using a convolutional neural network (CCN) are described. A system may include at least one computing device and program instructions stored in memory and executable in the at least one computing device that, when executed, direct the at least one computing device to implement a genetic algorithm (GA) routine that identifies and optimizes a plurality of hyperparameters for use in denoising an image using the convolutional neural network. An image may be denoised using the convolutional neural network, where the image is denoised using the hyperparameters identified and optimized in the genetic algorithm routine.

Abstract: Various embodiments for image denoising using a convolutional neural network (CCN) are described. A system may include at least one computing device and program instructions stored in memory and executable in the at least one computing device that, when executed, direct the at least one computing device to implement a genetic algorithm (GA) routine that identifies and optimizes a plurality of hyperparameters for use in denoising an image using the convolutional neural network. An image may be denoised using the convolutional neural network, where the image is denoised using the hyperparameters identified and optimized in the genetic algorithm routine.

Abstract: A mobile device (160) for medical image analysis is disclosed. The mobile device (160) includes a display (162), a communication module (218), a memory (204) configured to store processor-executable instructions (224) and a processor (202) in communication with the display (162), the communication module (218) and the memory (204). The processor (202) being configured to execute the processor-executable instructions (224) to implement a compression routine to generate a compressed representation of a medical image stored in the memory (204), transmit the compressed representation to a remote device (110) via the communication module (218), receive segmented results from the remote device (110), wherein the segmented results are derived from a reconstruction of the compressed representation generated at the remote device (110), and present, via the display (162), a segmented medical image based on the received segmented results.