# Artificial intelligence for personalized management of vestibular schwannoma: a multidisciplinary clinical implementation study

## TL;DR

A new AI tool helps doctors manage vestibular schwannoma by automating tumor measurements and improving team meeting efficiency.

## Contribution

A novel computer-assisted reporting tool for vestibular schwannoma management using deep learning for tumor segmentation.

## Key findings

- Automated reports were acceptable for 72% of cases, with high Dice scores for tumor segmentation.
- Computer-assisted reports initially increased preparation time but improved MDTM efficiency.
- The tool enables more personalized care by enhancing communication of tumor growth patterns.

## Abstract

Management of patients with vestibular schwannoma (VS) relies on precise tumor size and growth trend evaluation. We introduce and evaluate a novel computer-assisted reporting tool for clinical decision support during multidisciplinary team meetings (MDTMs) for VS patients.

Our approach exploits deep learning for tumor segmentation, automating tumor volume, and standard linear measurement extraction. We conducted 2 simulated MDTMs with the same 50 patients evaluated in both arms to compare our proposed approach against the standard process, focusing on its impact on preparation time and decision-making.

Automated reports provided acceptable information for an expert neuroradiologist in 72% of cases, while the remaining 28% required some revision with manual feature extraction. The segmentation models used in this report generation task achieved Dice scores of 0.9392 (±0.0351) for contrast-enhanced T1 and 0.9331 (±0.0354) for T2 MRI in delineating whole tumor regions. The automated computer-assisted reports that included additional tumor information initially extended the neuroradiologist’s preparation time for the MDTM (2 min 54 s [±1 min and 22 s] per case) compared to the standard preparation time (2 min 36 s (±1 min and 5 s] per case). However, the computer-assisted simulated MDTM approach significantly improved (P < .01) MDTM efficiency, with shorter discussion times per patient (1 min 15 s [±0 min and 28 s] per case) compared to standard simulated MDTM (1 min 21 s [±0 min and 44 s] per case).

An initial learning curve in interpreting new data measurements is quickly mastered and the enhanced communication of growth patterns and more comprehensive assessments ultimately provides clinicians with the tools to offer patients more personalized care.

This pilot clinical implementation study highlights the potential benefits of integrating automated measurements into clinical decision-making for VS management.