Spectral histopathology (SHP) using glass slides is a promising method for cancer detection. In order to achieve clinical application, it is very important to find the effect of glass type on detection results. The project has studied the influence of glass type on infrared spectra, tissue histopathological classification and cancer detection by two experimental methods. The backgrounds of experiment 1 and 2 were selected tissue-free areas (blank glass, glue & coverslip) and blank glass, respectively. In addition, glue removal by Matlab is an essential data process due to the existence of glue and coverslip in the H&E stained tissue sample. Therefore, finding the most suitable experimental method and glue removal method are also the project's important aims. Due to the partly opaque, the infrared transmission window for glass slides is 2000-3800 cm-1. Only hydroxyl and its related groups are shown in the infrared spectra of the blank glass slide. The comparison of spectra on 12 blank glass slides indicates that most glass slides cannot be discriminated by infrared spectra except for glass D. A series of prostate tissue sections from the same BPH patient are mound on the 12 different glass slides. Stroma and epithelium are used for studying tissue classification because they are important compositions and have a high proportion of prostate tissue. The results show that the tissue classification on the same glass slide has very high accuracy (above 98.40%) for both experiments. In addition, tissue classification accuracies for most classifiers tested on different glass slides are above 90.57%. Good performance means that the type of glass slide has a minor influence on tissue classification before removing the glue. Only 6 glass slides are commonly used for tissue work in 12 glass slides. A series of adjacent tissue slices are mounted on the 6 glass slides, which are from 4 patients (2 BPH & 2 CaP). The accuracies of cancer detection on the same glass slide for experiments 1 and 2 are 97.00% and 92.00%, respectively. However, the worse result on different glass slides indicates that the type of glass slide has an impact on cancer detection. The purpose of glue removal is correct the bands in 3400-3450 cm-1 and 3400-3600 cm-1, which are related to glue and coverslip, respectively. However, after removing the glue, tissue classification and cancer detection accuracy on the same glass slides are reduced. It means that glue has a contribution to tissue classification and cancer detection. Experiment 1 always had higher accuracy than experiment 2. The type of glass slide has a slight influence on tissue classification but has a huge effect on cancer detection. Therefore, the type of glass slide must be consistent during cancer detection by SHP, and experiment 1 is the most suitable method.
Date of Award | 1 Aug 2023 |
---|
Original language | English |
---|
Awarding Institution | - The University of Manchester
|
---|
Supervisor | Peter Gardner (Supervisor) & Bernard Treves Brown (Supervisor) |
---|
- glue removal
- Infrared spectral histopathology (SHP)
- glass slides
- cancer detection
- tissue classification
Infrared spectroscopic analysis of prostate tissue: Evaluation of different glass substrate and their influence on cancer detection, with a view for clinical translation
Zheng, Y. (Author). 1 Aug 2023
Student thesis: Phd