Abstract
Accidental injury to autonomic nerves during colorectal surgery poses a considerable risk and can lead to urinary, defecatory, and sexual dysfunction, substantially reducing the postoperative quality of life. This study investigates the potential of Hyperspectral Imaging (HSI) for nerve identification under realistic intraoperative conditions. We acquired laparoscopic HSI data from 13 patients undergoing colorectal surgery using
an HSI system that captured 100 spectral bands ranging from 500 nm to 995 nm. We performed pixel-level tissue classification using classical ma-
chine learning methods and compared the results with classifications on conventional RGB images. The 3D-CNN achieved the best overall performance across the dataset with a macro F1-score of 70.47 ± 11.69 and nerve-specific precision and recall of 69.21 ± 25.68 and 70.70 ± 15.55, respectively. Spectral analysis and feature importance assessment suggest that wavelengths below 600 nm are most discriminative for nerve identification. These results show that HSI data contain additional nerve-specific spectral information beyond conventional white-light imaging and may support safety-aware surgical assistance.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/SafeSurg_007.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: https://openreview.net/forum?id=zSFb9nV1nr
BibTex
@InProceedings{HarFar_Hyperspectral_MICCAISAT2026,
author = { Harb, Farid AND Wagner, Lars AND Schneider, Dennis N. AND Geiger, Alexander AND Berlet, Maximilian AND Neumann, Philipp-Alexander AND Wilhelm, Dirk},
title = { { Hyperspectral Recognition of Pelvic Autonomic Nerves for Safety-Aware Colorectal Surgery } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS pending},
month = {pending},
page = {pending}
}