Causally-Aware Intraoperative Imputation for Overall Survival Time Prediction

Previous efforts in vision community are mostly made on learning good representations from visual patterns. Beyond this, this paper emphasizes the high-level ability of causal reasoning. We thus present a case study of solving the challenging task of Overall Survival (OS) time in primary liver cancers. Critically, the prediction of OS time at the early stage remains challenging, due to the unobvious image patterns of reflecting the OS. To this end, we propose a causal inference system by leveraging the intraoperative attributes and the correlation among them, as an intermediate supervision to bridge the gap between the images and the final OS. Particularly, we build a causal graph, and train the images to estimate the intraoperative attributes for final as prediction. We present a novel Causally-aware Intraoperative Imputation Model (CAWIM) that can sequentially predict each attribute using its parent nodes in the estimated causal graph. To determine the causal directions, we propose a splitting-voting mechanism, which votes for the direction for each pair of adjacent nodes among multiple predictions obtained via causal discovery from heterogeneity. The practicability and effectiveness of our method are demonstrated by the promising results on liver cancer dataset of 361 patients with long-term observations.

1. Introduction

The success of recent deep learning model is largely attributed to learning the good representations for visual patterns. Such representations essentially facilitate various vision task, such as recognition and synthesis [15], [25], [33]. Nevertheless, one important goal for the vision community is to model and summarize the relationships of observed variables of a system, in order to enable well predictions on similar data. Essentially, it is desirable to understand how the system is changed if one modifies these relationships under certain conditions, e.g., the effects of a treatment in healthcare. Thus this demands the high-level ability of causal reasoning beyond the previous efforts of only learning good representations for visual patterns [1], [5], [16],[21],[35], [37]. This naturally leads into our task of causal inference.