2. CHAPTER 2: METHODOLOGY
2.6 Histological preparation, staining and techniques
2.6.2 Methodology for the Image analysis studies
2.6.2.1 Picro Sirius Red (PSR) staining
This staining method was used in the slides for the Image analysis study. Picro Sirius Red (PSR) staining is used to demonstrate collagen and non-collagenous proteins in tissue sections. Collagen stains red and non-collagenous proteins stain a light green. It is used to demonstrate conditions such as hepatic fibrosis 400. Description of the character and staining of SR is shown in Appendix A.
2.6.2.2 Tissue staining
Ninety-five biopsy sections were selected from the available 114 specimens and were stained with Picro Sirius red. These were the same group of specimens used in the Mandard study. Each biopsy was already graded according to the Mandard system. Sections of each biopsy stained with Sirius red were used for measurement of fibrosis using digital image analysis technique. This stain binds to all connective tissue, but primarily to collagen, and the quantity of bound stain has been shown to correlate well with chemically determined collagen content and morphometrically determined fibrosis, as was shown in several studies in other tissue 400 405.
123 Figure (9). Layout of the Image analysis processing sequence
2.6.2.3 Image capture and processing
Image acquisition was done by a digital grabber and camera attached to a light microscope (Olympus BX50). This microscope offers superior performance covering applications from routine investigation to sophisticated research, and would suit a wide range of observation requirements. The Universal Infinity System (UIS) provides an extended field of view, higher contrast and the sharpest images, as well as outstanding flatness throughout the entire field of view.
Extra-bright 100W halogen illumination was used, to deliver optimum performance. An analogue camera was used (Nikon, Japan). The output from the camera was converted to digital imaging by a CMS790 PAL frame grabber device (Scion Corp., MD 21701, USA). This gave a digital image, which was processed using a HP desktop computer with Intel DuoCore processor, that is equipped by a dedicated software for image analysis (NIS elements Basic Research [BR], Nikon, USA). The camera first had to be warmed to minimize background ‘noise’ due to changes in temperature. Microscopy was done using high level of illumination, with a green
124 optical filter to reduce the influence of the other non-specific objects which were stained by the Picrosirius red, but without reducing the staining of the fibrotic regions. Also, the different components of the microscope e.g. the lens, the reflective mirror, etc, as well as the slides had to be checked for their cleanliness to ensure minimum of impurities during the examinations.
Figure (10). Steps of the techniques used in analysing each hotspot in Image analysis
2.6.2.4 Thresholding
In terms of practicality, image analysis at lower magnification gives superior reproducibility, correlates well with its high magnification counterpart and is 10 - 20 times faster. Using the low (whole biopsy in one frame) magnification has several potential advantages: random loss of tissue caused by gaps between frames and double counting due to overlap does not occur. Other dark staining tissues which are
125 not collagen such as nuclei are below the level of resolution at low magnification and therefore do not falsely elevate the measured fibrosis area. This concept has been investigated and validated before in other studies 396. However, it was not possible at the time of the study to obtain lenses that can accommodate the whole slide in one frame to be assessed, so it was decided to choose 5 random sections from each slide at a 20x magnification, where there is apparent fibrosis in the interface between the apparently normal or fibrotic tissue and the tumour, and then taking the mean of the 5 chosen areas.
Images were digitized following interactive light intensity equilibration, and analysed as RGB 24-bit images. The stained section captured represented fibrosis as red and the parenchyma as blue or green for illustration purposes using the available software. Sirius red was used in order to allow accurate location of the fibrous regions. The problem with counterstaining slides for image analysis is that there must be sufficient contrast between the counterstained tissue and the stained deposits for the computer to be able to differentiate between the two quantities. To overcome this problem, an additional colour filter (green filter) was used when capturing images in order to enhance the stain/counterstain contrast. A stabilised light source was used to ensure consistency in the data acquisition. In order to remove any other spurious lighting variations background subtraction was performed on all images. This process involved “training” the software to recognise the background spaces in the form of empty areas, or completely different colouration on the image, and subtract them from the calculation of the final average. After interactive thresholding, the image was converted into a binary image. The 2-dimensional patterns were measured by direct pixel counting on the
126 binary images. The total area of the section was the sum of all microscopic field including parenchyma and fibrosis.
In terms of image intensity, the fibrous areas present with a slide should appear acquiring a red colour than the surrounding tissue and counterstained cells bodies. The computer requires an intensity threshold in order to be able to differentiate between these red areas and the surrounding areas. It is possible to use automatic methods of setting this threshold in a completely objective way; however, the irregular distribution of the stained deposits visible here meant that a semi-automatic system of intensity thresholding had to be adopted. The setting of the intensity threshold was performed by presenting a number of randomly selected microscope fields (around 75) to the software after the agreement between two pathologists as to the intensity of the stain that would properly represent the fibrosis area, and then training the software to pick up only these intensities. The values for the different fields were then averaged to obtain an intensity threshold for that section, which is an automatic process included in the software used. In an attempt to overcome any potential operator bias in the thresholding process, all the thresholds used in this study were set by one operator.
With this image analysis system it was possible to use a low magnification view of the tissue sample to manually define an area to be scanned in detail. Using the image intensity threshold selected earlier the computer then scanned the defined area by each assessor at the same magnification and identified 5 of the areas which represented the high degree of fibrotic changes within that slide. The total areas of tissue scanned were automatically averaged by the software, and recorded on a spreadsheet template.
127 2.6.2.5 Inter-observer and intra-observer method for assessment of fibrosis using image analysis technique
Inter-observer reproducibility of fibrosis percentage measurement with digital image analysis was assessed among three pathologists with an interest and experience in the use and concept of image analysis. Each one was blinded to the Mandard staging results of the slides in test, and each performed the complete protocol for fresh digitization of slides and thresholded images to obtain fibrosis percentage in a sub- cohort of biopsies consisting of the 95 biopsies.
The technique was reviewed and checked with the researcher, and it was confirmed that the same apparatus i.e. the microscope, the camera, the grabber, and the software, and the settings were kept in the same operating condition for every assessor, as well as stabilising the parameters used for each. The assessments took place in St. Mary’s cellular pathology laboratories.
Each assessor was given the slides to assess independently, without a prior knowledge of the clinical data or the corresponding Mandard grading. Each assessor repeated the same test twice independently, with an interval of 2 - 4 weeks between examinations.
2.6.3 Methodology of immunohistochemical staining and