Visually Integrated Clinical Cooperation - algorithmic concepts, implementation and evaluation

The workflow in modern hospitals entails that the medical treatment of a patient is distributed between several physicians and nurses. This leads to an intensive cooperation, which takes place under particular time pressure and requires an efficient conveyance of the relevant patient-related medical data to colleagues. This requirement is difficult to achieve with traditional data representation approaches, which usually do not consider the specifics of the cooperative work. This thesis introduces a concept of Visually Integrated Clinical Cooperation (VICC) to support the information transfer in cooperative tasks in a hospital. Developed in the context of an interdisciplinary research project, it exploits findings of a sociological field study in a hospital and feedback from the discussions with physicians. Accordingly, the VICC concept aims at an multivariate visualization on a mobile device that provides a synopsis of the relevant data and is intuitively comprehensible for medical personnel. The core components of the proposed concept are anatomically integrated in-place visualization and iconic glyphs.
The anatomically integrated visualization uses a 3D human avatar as spatial representation of visually encoded medical data with an (inherent) anatomical reference. This component comprises a set of formal requirements and procedures for this kind of visual encoding as well as a prototypical implementation for the diagnosis of spinal disc herniation, evaluated by neurosurgeons.
The VICC concept also includes a personalization option of the generic 3D human avatar, using a 3D reconstruction of a patient’s body from range data. To allow it, a method for robust camera pose tracking for the spatially and temporally low-resolution range data, given in mobile applications, is proposed. It combines a geometry-based pose estimation by means of the iterative closest point (ICP) algorithm with inertial tracking, using an extended Kalman filter (EKF). In particular, it uses the extrapolated ICP pose estimates as virtual measurements and the output of the EKF as initial guess for the next ICP-based pose estimation.
The iconic glyph approach allows for representation of a patient’s data without anatomical reference. It aims to combine visual metaphors, inherent for icons, and the glyph capability for multivariate visualization. Technically, it is based on a parametric representation that utilizes diffusion curves, enriched with new degrees of freedom in arc-length parametrization, which allows for automated, controllable manipulation of the icon contours’ geometry and the related colour attributes. Besides the generic concept, an implementation for a specific design, based on periodic, wavelike contour modifications along with a perception and quantization model for these kinds of visual variables are proposed. The practicality of the approach is demonstrated by examples for visualization of weather forecast uncertainty, COVID-19 statistic trends and intracranial pressure.