The translucent back: utilizing optical 3D scanning and thermal imaging to assess internal deformities of the spine

Background

To date, the standard approach for assessing and monitoring spinal deformities is to repeatedly use trunk radiography involving ionizing radiation. However, cumulative exposure due to repetitive radiographic imaging is thought to increase the risk of developing malignancies. For scoliosis patients, the ability to monitor possible scoliotic progression without radioactive methods would imply a disruptive change in the clinical assessment and monitoring landscape. Hence, we’re focusing on non-invasive scoliosis diagnosis and monitoring approaches, namely optical 3D scanning and thermal imaging. The former is used to digitize the dorsal trunk topography, whereas the latter is used to measure the emitted thermal radiation from underlying anatomical structures of the trunk.

Goal

The main goals of the project are to

• develop of a machine learning-based model to capture the relationship between dorsal trunk appearance and spinal curvature, and to
• investigate the suitability and possibilities of thermal imaging in the clinical assessment of scoliosis.

Method

In a clinical study conducted at the University Hospital Balgrist, dorsal trunk topography scans, thermograms and biplanar radiographs of scoliosis patients are collected. This data is then used to train, test and validate models that estimate the spinal curvature given a specific back topography. This data is further fused with thermal data to ultimately achieve more accurate estimations of the spinal curvature. Ultimately, we aim to use our approach to monitor scoliosis patients non-invasively and predict disease progression.