Similarity metrics for surgical process models.

Thomas Neumuth 1, * Frank Loebe 2, 3 Pierre Jannin 4, 5
* Corresponding author
4 VisAGeS - Vision, Action et Gestion d'informations en Santé
INSERM - Institut National de la Santé et de la Recherche Médicale : U746, Inria Rennes – Bretagne Atlantique , IRISA-D5 - SIGNAUX ET IMAGES NUMÉRIQUES, ROBOTIQUE
Abstract : OBJECTIVE: The objective of this work is to introduce a set of similarity metrics for comparing surgical process models (SPMs). SPMs are progression models of surgical interventions that support quantitative analyses of surgical activities, supporting systems engineering or process optimization. METHODS AND MATERIALS: Five different similarity metrics are presented and proven. These metrics deal with several dimensions of process compliance in surgery, including granularity, content, time, order, and frequency of surgical activities. The metrics were experimentally validated using 20 clinical data sets each for cataract interventions, craniotomy interventions, and supratentorial tumor resections. The clinical data sets were controllably modified in simulations, which were iterated ten times, resulting in a total of 600 simulated data sets. The simulated data sets were subsequently compared to the original data sets to empirically assess the predictive validity of the metrics. RESULTS: We show that the results of the metrics for the surgical process models correlate significantly (p<0.001) with the induced modifications and that all metrics meet predictive validity. The clinical use of the metrics was exemplarily, as demonstrated by assessment of the learning curves of observers during surgical process model acquisition. CONCLUSION: Measuring similarity between surgical processes is a complex task. However, metrics for computing the similarity between surgical process models are needed in many uses in the field of medical engineering. These metrics are essential whenever two SPMs need to be compared, such as during the evaluation of technical systems, the education of observers, or the determination of surgical strategies. These metrics are key figures that provide a solid base for medical decisions, such as during validation of sensor systems for use in operating rooms in the future.
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-00905410
Contributor : Morgane Le Corre <>
Submitted on : Monday, November 18, 2013 - 11:24:49 AM
Last modification on : Monday, March 4, 2019 - 2:08:21 PM

Identifiers

Citation

Thomas Neumuth, Frank Loebe, Pierre Jannin. Similarity metrics for surgical process models.. Artificial Intelligence in Medicine, Elsevier, 2012, 54 (1), pp.15-27. ⟨10.1016/j.artmed.2011.10.001⟩. ⟨hal-00905410⟩

Share

Metrics

Record views

1328