Explainable Anatomical Shape Analysis Through Deep Hierarchical Generative Models

Abstract

Quantification of anatomical shape changes still relies on scalar global indexes which are largely insensitive to regional or asymmetric modifications. Accurate assessment of pathology-driven anatomical remodeling is a crucial step for the diagnosis and treatment of heart conditions. Deep learning approaches have recently achieved wide success in the analysis of medical images, but they lack interpretability in the feature extraction and decision processes. In this work, we propose a new interpretable deep learning model for shape analysis. In particular, we exploit deep generative networks to model a population of anatomical segmentations through a hierarchy of conditional latent variables. At the highest level of this hierarchy, a two-dimensional latent space is simultaneously optimised to discriminate distinct clinical conditions, enabling the direct visualisation of the classification space. Moreover, the anatomical variability encoded by this discriminative latent space can be visualised in the segmentation space thanks to the generative properties of the model, making the classification task transparent. This approach yielded high accuracy in the categorisation of healthy and remodelled hearts when tested on unseen segmentations from our own multi-centre dataset as well as in an external validation set. More importantly, it enabled the visualisation in three-dimensions of the most discriminative anatomical features between the two conditions. The proposed approach scales effectively to large populations, facilitating highthroughput analysis of normal anatomy and pathology in largescale studies of volumetric imaging.

Explainable Deep Learning Models in Medical Image Analysis

“…An example of the latter was presented in [ 68 ] using the latent space of the features of a variational autoencoder for classification and segmentation of the brain MRI of Alzheimer’s patients. The classification was performed in a two-dimensional latent space using an multi layer perceptron (MLP).…”

Deep Graph Generators: A Survey

“…Hence, it is of great importance to make graph generation methods more interpretable. As of now, deep generative methods in areas such as image [161]- [163] and text [164]- [166] have slowly moved towards being more interpretable. However, only a few attempts [55], [56], [102] have recently been made in graph generation, making model interpretability a notable future research prospect.…”

Deep Learning in Biomedical Optics

“…To this end, there are two general approaches that are actively being researched in the field [282]. The first is to develop an interpretable computational structure instead of DNNs [283,284], so that the predictions are made based on the crafted logic in the DL model. The second approach is to provide post hoc model prediction interpretation, such as attention mechanism [42,285] and uncertainty quantification [20,286,287], while keeping the same DNN structure.…”