Development and Evaluation of a Rib Statistical Shape Model for Thoracic Surgery

Patients with advanced cancer undergoing chest wall resection may require reconstruction. Currently, rib prostheses are created by segmenting computed tomography images, which is time-consuming and labour intensive. The aim was to optimise the production of digital rib models based on a patient's age, weight, height and gender. A statistical shape model of human ribs was created and used to synthetise rib models, which were compared to the ones produced by segmentation and mirroring. The segmentation took
$11.56\pm 1.60$
min compared to
$0.027 \pm 0.009\ \min$
using the new technique. The average mesh error between the mirroring technique and segmentation was
$0.58\pm 0.25$
mm (right ribs), and
$0.87\pm 0.18$
mm (left ribs), compared to
$1.37\pm 0.66$
mm (
$\mathrm{p} < 0.0001$
) and 1.68
$\pm 0.77$
mm (
$\mathrm{p} < 0.05$
), respectively, for the new technique. The new technique is promising for the efficiency and ease-of-use in the clinical environment. Clinical Relevance— This is an optimised 3D modelling method providing clinicians with a time-efficient technique to create patient-specific rib prostheses, without any expertise or software knowledge required.