Ulf Dahlstrand
Research project participant
Novel Diagnostic Tools for Skin and Periorbital Cancer - Exploring Photoacoustic Imaging and Diffuse Reflectance Spectroscopy
Author
Summary, in English
The eyelids are susceptible to a number of skin cancers which are challenging to excise radically without sacrificing excessive healthy tissue. The way in which a tumor is delineated preoperatively has not changed significantly over the past century. The aims of the work presented in this thesis were to investigate two novel non-invasive techniques for diagnosing and delineating skin tumors.
Extended-wavelength diffuse reflectance spectroscopy (EWDRS) was evaluated to determine its ability to differentiate between and classify different skin and tissue types in an in vivo pig model, with the aid of machine learning methods.
The recordings were used to train a support vector machine, and it was possible to perform classifications with an overall accuracy of over 98%. The ability of EWDRS to identify the borders of pigmented skin lesions in an in vivo pig model was also evaluated. Using a thin probe, it was possible to detect the border with a median discrepancy of 70 μm, compared to the border found on histological examination.
Photoacoustic imaging (PAI), a biomedical imaging modality that combines laser irradiation and ultrasound, was used to examine basal cell carcinomas (BCCs) and human eyelids ex vivo. Typical photoacoustic spectra were observed for BCCs as well as for the different layers of the healthy eyelid, and these structures could be visualized in three-dimensional images. A case was described in which PAI showed that the pentagonal excision of an eyelid BCC was non-radical, as was later confirmed by histological examination.
In conclusion, both EWDRS and PAI are capable of differentiating between different kinds of tissue and, following further development and studies, could potentially be used to diagnose and delineate skin and eyelid tumors prior to surgical excision.
Extended-wavelength diffuse reflectance spectroscopy (EWDRS) was evaluated to determine its ability to differentiate between and classify different skin and tissue types in an in vivo pig model, with the aid of machine learning methods.
The recordings were used to train a support vector machine, and it was possible to perform classifications with an overall accuracy of over 98%. The ability of EWDRS to identify the borders of pigmented skin lesions in an in vivo pig model was also evaluated. Using a thin probe, it was possible to detect the border with a median discrepancy of 70 μm, compared to the border found on histological examination.
Photoacoustic imaging (PAI), a biomedical imaging modality that combines laser irradiation and ultrasound, was used to examine basal cell carcinomas (BCCs) and human eyelids ex vivo. Typical photoacoustic spectra were observed for BCCs as well as for the different layers of the healthy eyelid, and these structures could be visualized in three-dimensional images. A case was described in which PAI showed that the pentagonal excision of an eyelid BCC was non-radical, as was later confirmed by histological examination.
In conclusion, both EWDRS and PAI are capable of differentiating between different kinds of tissue and, following further development and studies, could potentially be used to diagnose and delineate skin and eyelid tumors prior to surgical excision.
Department/s
- Ophthalmology, Lund
- Ophthalmology Imaging Research Group
Publishing year
2020
Language
English
Publication/Series
Lund University, Faculty of Medicine Doctoral Dissertation Series
Issue
2020:31
Full text
Document type
Dissertation
Publisher
Lund University, Faculty of Medicine
Topic
- Ophthalmology
Status
Published
Research group
- Ophthalmology Imaging Research Group
Supervisor
- Malin Malmsjö
- Jonas Blohmé
- Karl Engelsberg
- Nina Reistad
ISBN/ISSN/Other
- ISSN: 1652-8220
- ISBN: 978-91-7619-891-9
Defence date
20 March 2020
Defence time
13:00
Defence place
Konferensrummet, Ögonklinik A, Kioskgatan 1, plan 2, Skånes universitetssjukhus i Lund
Opponent
- Marcelo Ayala (Associate Professor)