Dual-modality probe for prostate cancer detection by combining Raman spectroscopy and tactile resonance technology
Abstract: Prostate adenocarcinoma, or prostate cancer (PCa), is the most common type of cancer and the leading cause of cancer-related deaths for men in Sweden and Europe. Radical prostatectomy (RP) is the most common curative treatment for PCa. This is a surgical procedure where the entire prostate is removed. The surgeons strive to minimise postoperative side-effects, and will use nerve-sparing techniques if possible. The final decision on how much tissue to remove around the prostate is taken during surgery. After the prostate is resected it is histopathologically analysed. The risk of PCa recurrence is elevated when cancer cells are found in the surgical margin. Currently, there is no viable method to detect and localise PCa tumours near the resection margin during surgery. The general aim of this thesis was to develop a medical instrument with the ability to detect PCa tumours near the tissue surface.Datasets from the resonance sensor and from the Raman spectroscopy obtained on porcine model tissue were compared. Principal component analysis (PCA) was used to reduce the Raman spectroscopic dataset, and groups of tissue content were formed by a hierarchical cluster analysis (HCA) of the PCA results. The correlation of the two datasets was evaluated by a model using the PCA results for describing the stiffness of the groups.A support vector machine was evaluated as a method to combine the datasets of the two modalities. The method was used for classifying three types of porcine prostate tissues and to discern healthy vs. cancerous human prostate tissue. The cross-validation accuracy for the SVM classification was 87% for the porcine prostate and 77% for the human prostate tissue types for highly homogeneous tissue samples (>83%). Several important aspects regarding design and methods to be used for combining the two modalities into one probe were investigated. The effects of using different amounts of rubber latex for combining a TRM sensor an a fibre optic Raman probe substitute were determined. A description of the heat produced by the laser at the fibre optic tip was established, and the temperature dependence of Δf was investigated. Methods and conditions, e.g. instrument settings and light shields, for performing Raman spectroscopy with ambient light present were investigated.The dual-modality probe prototype was used to perform the first measurements on porcine model tissue. Methods for using the Raman spectroscopic modality of the combined probe with ambient light present were evaluated. The TRM modality could discern tissue with significantly different stiffness. Raman spectroscopy could be performed with ambient fluorescent light present. The first measurements on fresh human prostate tissue using the dual-modality probe were performed. Prostates with different average stiffnesses could be compared by calculating stiffness contrast. The TRM modality could discern tissue with significantly different stiffness contrast (p < 0.05). The background fluorescence in the Raman spectra of fresh human prostate tissue was higher than expected. The high-wavenumber region, 2700 cm−1 to 3200 cm−1, of the Raman spectra could be used to discern tissue characteristics using HCA of the PCA results. In this thesis, methods and design considerations for the development of a dualmodality probe, combining Raman spectroscopy and tactile resonance sensor technology, were identified and evaluated. In addition, methods for how to use the dual-modality probe for tissue classification and for combining of the datasets of the two modalities, were studied. The resonance sensor swiftly evaluates the tissue stiffness, like the palpation with a finger. The Raman spectroscopy would be applied when malignancy is suspected and adds detailed knowledge of the molecular content. This makes the dual-modality probe a promising instrument for use during RP.
CLICK HERE TO DOWNLOAD THE WHOLE DISSERTATION. (in PDF format)