Artificial Intelligence-Based Prediction of Contrast Medium Doses for Computed Tomography Angiography Using Optimized Clinical Parameter Sets.

OBJECTIVES: In this paper, an artificial intelligence-based algorithm for predicting the optimal contrast medium dose for computed tomography (CT) angiography of the aorta is presented and evaluated in a clinical study. The prediction of the contrast dose reduction is modelled as a classification problem using the image contrast as the main feature. METHODS: This classification is performed by random decision forests (RDF) and k-nearest-neighbor methods (KNN). For the selection of optimal parameter subsets all possible combinations of the 22 clinical parameters (age, blood pressure, etc.) are considered using the classification accuracy and precision of the KNN classifier and RDF as quality criteria. Subsequently, the results of the evaluation were optimized by means of feature transformation using regression neural networks (RNN). These were used for a direct classification based on regressed Hounsfield units as well as preprocessing for a subsequent KNN classification. RESULTS: For feature selection, an RDF model achieved the highest accuracy of 84.42% and a KNN model achieved the best precision of 86.21%. The most important parameters include age, height, and hemoglobin. The feature transformation using an RNN considerably exceeded these values with an accuracy of 90.00% and a precision of 97.62% using all 22 parameters as input. However, also the feasibility of the parameter sets in routine clinical practice has to be considered, because some of the 22 parameters are not measured in routine clinical practice and additional measurement time of 15 to 20 minutes per patient is needed. Using the standard feature set available in clinical routine the best accuracy of 86.67% and precision of 93.18% was achieved by the RNN. CONCLUSION: We developed a reliable hybrid system that helps radiologists determine the optimal contrast dose for CT angiography based on patient-specific parameters.

Proposal of Semantic Annotation for German Metadata Using Bidirectional Recurrent Neural Networks.

The distributed nature of our digital healthcare and the rapid emergence of new data sources prevents a compelling overview and the joint use of new data. Data integration, e.g., with metadata and semantic annotations, is expected to overcome this challenge. In this paper, we present an approach to predict UMLS codes to given German metadata using recurrent neural networks. The augmentation of the training dataset using the Medical Subject Headings (MeSH), particularly the German translations, also improved the model accuracy. The model demonstrates robust performance with 75% accuracy and aims to show that increasingly sophisticated machine learning tools can already play a significant role in data integration.

Image registration and appearance adaptation in non-correspondent image regions for new MS lesions detection.

Manual detection of newly formed lesions in multiple sclerosis is an important but tedious and difficult task. Several approaches for automating the detection of new lesions have recently been proposed, but they tend to either overestimate the actual amount of new lesions or to miss many lesions. In this paper, an image registration convolutional neural network (CNN) that adapts the baseline image to the follow-up image by spatial deformations and simulation of new lesions is proposed. Simultaneously, segmentations of new lesions are generated, which are shown to reliably estimate the real new lesion load and to separate stable and progressive patients. Several applications of the proposed network emerge: image registration, detection and segmentation of new lesions, and modeling of new MS lesions. The modeled lesions offer the possibility to investigate the intensity profile of new lesions.

A systematic comparison of generative models for medical images.

PURPOSE: This work aims for a systematic comparison of popular shape and appearance models. Here, two statistical and four deep-learning-based shape and appearance models are compared and evaluated in terms of their expressiveness described by their generalization ability and specificity as well as further properties like input data format, interpretability and latent space distribution and dimension. METHODS: Classical shape models and their locality-based extension are considered next to autoencoders, variational autoencoders, diffeomorphic autoencoders and generative adversarial networks. The approaches are evaluated in terms of generalization ability, specificity and likeness depending on the amount of training data. Furthermore, various latent space metrics are presented in order to capture further major characteristics of the models. RESULTS: The experimental setup showed that locality statistical shape models yield best results in terms of generalization ability for 2D and 3D shape modeling. However, the deep learning approaches show strongly improved specificity. In the case of simultaneous shape and appearance modeling, the neural networks are able to generate more realistic and diverse appearances. A major drawback of the deep-learning models is, however, their impaired interpretability and ambiguity of the latent space. CONCLUSIONS: It can be concluded that for applications not requiring particularly good specificity, shape modeling can be reliably established with locality-based statistical shape models, especially when it comes to 3D shapes. However, deep learning approaches are more worthwhile in terms of appearance modeling.

Memory-efficient GAN-based domain translation of high resolution 3D medical images.

Generative adversarial networks (GANs) are currently rarely applied on 3D medical images of large size, due to their immense computational demand. The present work proposes a multi-scale patch-based GAN approach for establishing unpaired domain translation by generating 3D medical image volumes of high resolution in a memory-efficient way. The key idea to enable memory-efficient image generation is to first generate a low-resolution version of the image followed by the generation of patches of constant sizes but successively growing resolutions. To avoid patch artifacts and incorporate global information, the patch generation is conditioned on patches from previous resolution scales. Those multi-scale GANs are trained to generate realistically looking images from image sketches in order to perform an unpaired domain translation. This allows to preserve the topology of the test data and generate the appearance of the training domain data. The evaluation of the domain translation scenarios is performed on brain MRIs of size 155 × 240 × 240 and thorax CTs of size up to 5123. Compared to common patch-based approaches, the multi-resolution scheme enables better image quality and prevents patch artifacts. Also, it ensures constant GPU memory demand independent from the image size, allowing for the generation of arbitrarily large images.

Unsupervised pathology detection in medical images using conditional variational autoencoders.

PURPOSE: Pathology detection in medical image data is an important but a rather complicated task. In particular, the big variability of the pathologies is a challenge to automatic detection methods and even to machine learning methods. Supervised algorithms would usually learn the appearance of a single pathological structure based on a large annotated dataset. As such data is not usually available, especially in large amounts, in this work we pursue a different unsupervised approach. METHODS: Our method is based on learning the entire variability of healthy data and detect pathologies by their differences to the learned norm. For this purpose, we use conditional variational autoencoders which learn the reconstruction and encoding distribution of healthy images and also have the ability to integrate certain prior knowledge about the data (condition). RESULTS: Our experiments on different 2D and 3D datasets show that the approach is suitable for the detection of pathologies and deliver reasonable Dice coefficients and AUCs. Also this method can estimate missing correspondences in pathological images and thus can be used as a pre-step to a registration method. Our experiments show improving registration results on pathological data when using this approach. CONCLUSIONS: Overall the presented approach is suitable for a rough pathology detection in medical images and can be successfully used as a preprocessing step to other image processing methods.