Deep Learning Analysis of In Vivo Hyperspectral Images for Automated Intraoperative Nerve Detection
Nerves are critical structures that may be difficult to recognize during surgery. Inadvertent nerve injuries can have catastrophic consequences for the patient and lead to life-long pain and a reduced quality of life. Hyperspectral imaging (HSI) is a non-invasive technique combining photography with...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-08-01
|
| Series: | Diagnostics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2075-4418/11/8/1508 |
| _version_ | 1827685459565740032 |
|---|---|
| author | Manuel Barberio Toby Collins Valentin Bencteux Richard Nkusi Eric Felli Massimo Giuseppe Viola Jacques Marescaux Alexandre Hostettler Michele Diana |
| author_facet | Manuel Barberio Toby Collins Valentin Bencteux Richard Nkusi Eric Felli Massimo Giuseppe Viola Jacques Marescaux Alexandre Hostettler Michele Diana |
| author_sort | Manuel Barberio |
| collection | DOAJ |
| description | Nerves are critical structures that may be difficult to recognize during surgery. Inadvertent nerve injuries can have catastrophic consequences for the patient and lead to life-long pain and a reduced quality of life. Hyperspectral imaging (HSI) is a non-invasive technique combining photography with spectroscopy, allowing non-invasive intraoperative biological tissue property quantification. We show, for the first time, that HSI combined with deep learning allows nerves and other tissue types to be automatically recognized in in vivo hyperspectral images. An animal model was used, and eight anesthetized pigs underwent neck midline incisions, exposing several structures (nerve, artery, vein, muscle, fat, skin). State-of-the-art machine learning models were trained to recognize these tissue types in HSI data. The best model was a convolutional neural network (CNN), achieving an overall average sensitivity of 0.91 and a specificity of 1.0, validated with leave-one-patient-out cross-validation. For the nerve, the CNN achieved an average sensitivity of 0.76 and a specificity of 0.99. In conclusion, HSI combined with a CNN model is suitable for in vivo nerve recognition. |
| first_indexed | 2024-03-10T08:52:59Z |
| format | Article |
| id | doaj.art-cd40fc5a8fd846ae8d42b3f03c5870d6 |
| institution | Directory Open Access Journal |
| issn | 2075-4418 |
| language | English |
| last_indexed | 2024-03-10T08:52:59Z |
| publishDate | 2021-08-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Diagnostics |
| spelling | doaj.art-cd40fc5a8fd846ae8d42b3f03c5870d62023-11-22T07:21:30ZengMDPI AGDiagnostics2075-44182021-08-01118150810.3390/diagnostics11081508Deep Learning Analysis of In Vivo Hyperspectral Images for Automated Intraoperative Nerve DetectionManuel Barberio0Toby Collins1Valentin Bencteux2Richard Nkusi3Eric Felli4Massimo Giuseppe Viola5Jacques Marescaux6Alexandre Hostettler7Michele Diana8Department of Research, Institute of Image-Guided Surgery, IHU-Strasbourg, 67091 Strasbourg, FranceDepartment of Research, Research Institute against Digestive Cancer, IRCAD, 67091 Strasbourg, FranceDepartment of Research, Institute of Image-Guided Surgery, IHU-Strasbourg, 67091 Strasbourg, FranceDepartment of Research, Research Institute against Digestive Cancer, IRCAD Africa, Kigali 2 KN 30 ST, RwandaDepartment of Research, Institute of Image-Guided Surgery, IHU-Strasbourg, 67091 Strasbourg, FranceDepartment of Surgery, Ospedale Card. G. Panico, 73039 Tricase, ItalyDepartment of Research, Research Institute against Digestive Cancer, IRCAD, 67091 Strasbourg, FranceDepartment of Research, Research Institute against Digestive Cancer, IRCAD, 67091 Strasbourg, FranceDepartment of Research, Research Institute against Digestive Cancer, IRCAD, 67091 Strasbourg, FranceNerves are critical structures that may be difficult to recognize during surgery. Inadvertent nerve injuries can have catastrophic consequences for the patient and lead to life-long pain and a reduced quality of life. Hyperspectral imaging (HSI) is a non-invasive technique combining photography with spectroscopy, allowing non-invasive intraoperative biological tissue property quantification. We show, for the first time, that HSI combined with deep learning allows nerves and other tissue types to be automatically recognized in in vivo hyperspectral images. An animal model was used, and eight anesthetized pigs underwent neck midline incisions, exposing several structures (nerve, artery, vein, muscle, fat, skin). State-of-the-art machine learning models were trained to recognize these tissue types in HSI data. The best model was a convolutional neural network (CNN), achieving an overall average sensitivity of 0.91 and a specificity of 1.0, validated with leave-one-patient-out cross-validation. For the nerve, the CNN achieved an average sensitivity of 0.76 and a specificity of 0.99. In conclusion, HSI combined with a CNN model is suitable for in vivo nerve recognition.https://www.mdpi.com/2075-4418/11/8/1508hyperspectral imagingartificial intelligencetissue recognitionintraoperative navigation tooloptical imagingdeep learning |
| spellingShingle | Manuel Barberio Toby Collins Valentin Bencteux Richard Nkusi Eric Felli Massimo Giuseppe Viola Jacques Marescaux Alexandre Hostettler Michele Diana Deep Learning Analysis of In Vivo Hyperspectral Images for Automated Intraoperative Nerve Detection Diagnostics hyperspectral imaging artificial intelligence tissue recognition intraoperative navigation tool optical imaging deep learning |
| title | Deep Learning Analysis of In Vivo Hyperspectral Images for Automated Intraoperative Nerve Detection |
| title_full | Deep Learning Analysis of In Vivo Hyperspectral Images for Automated Intraoperative Nerve Detection |
| title_fullStr | Deep Learning Analysis of In Vivo Hyperspectral Images for Automated Intraoperative Nerve Detection |
| title_full_unstemmed | Deep Learning Analysis of In Vivo Hyperspectral Images for Automated Intraoperative Nerve Detection |
| title_short | Deep Learning Analysis of In Vivo Hyperspectral Images for Automated Intraoperative Nerve Detection |
| title_sort | deep learning analysis of in vivo hyperspectral images for automated intraoperative nerve detection |
| topic | hyperspectral imaging artificial intelligence tissue recognition intraoperative navigation tool optical imaging deep learning |
| url | https://www.mdpi.com/2075-4418/11/8/1508 |
| work_keys_str_mv | AT manuelbarberio deeplearninganalysisofinvivohyperspectralimagesforautomatedintraoperativenervedetection AT tobycollins deeplearninganalysisofinvivohyperspectralimagesforautomatedintraoperativenervedetection AT valentinbencteux deeplearninganalysisofinvivohyperspectralimagesforautomatedintraoperativenervedetection AT richardnkusi deeplearninganalysisofinvivohyperspectralimagesforautomatedintraoperativenervedetection AT ericfelli deeplearninganalysisofinvivohyperspectralimagesforautomatedintraoperativenervedetection AT massimogiuseppeviola deeplearninganalysisofinvivohyperspectralimagesforautomatedintraoperativenervedetection AT jacquesmarescaux deeplearninganalysisofinvivohyperspectralimagesforautomatedintraoperativenervedetection AT alexandrehostettler deeplearninganalysisofinvivohyperspectralimagesforautomatedintraoperativenervedetection AT michelediana deeplearninganalysisofinvivohyperspectralimagesforautomatedintraoperativenervedetection |