Fluorescence in situ hybridization (FISH) is a test that “maps” the genetic material in a person’s cells. This test can be used to visualize specific genes or portions of genes. FISH testing is done on breast cancer tissue removed during biopsy to see if the cells have extra copies of the HER2 gene. The more copies of the HER2 gene that are present, the more HER2 receptors the cells have. These HER2 receptors receive signals that stimulate the growth of breast cancer cells. The FISH test results will tell you that the cancer is either 'positive' or 'negative' (a result sometimes reported as 'zero') for HER2.
The FISH test is not as common as the Immunohistochemistry test for HER2 detection. However, the FISH test gives more accurate results than the IHC test. Due to the high cost of the FISH test, many laboratories first apply the IHC test. If the IHC test fails to produce a 'positive' or 'negative' result for HER2 in the involved tissue, the FISH test is applied.
FISH technique provides promising molecular imaging biomarkers to precisely and dependably detect and diagnose disorders which are sign of cancers. Since contemporary manual FISH signal analysis is low-effective and inconsistent, it is an attractive research area to develop automated FISH image scanning systems and Computeraided Diagnosis (CAD) schemes. The gene expression of epidermal growth factor receptors 2 (HER-2) genes is highly related to clinical results in patients with probable breast cancer. Although FISH technology outperforms other methods, yet it has so many drawbacks. The precise diagnosis of the HER2 gene amplification is so important that the adjuvant or neoadjuvant therapy determinations are based on the scores that are given using these tests. The status of HER2 is assessed using different methods which are immunohistochemistry (ICH) or FISH methods. The FISH technique is very useful to identify the nucleotides sequence which is common in certain genetic disorders. Amplification of the HER2, which is associated with aggressive tumour behaviour, occurs in approximately quarter of breast cancers. Amplification of HER2 has been proved to be related to women with invasive breast cancer and negative prognostic variables such us Estrogen Receptor (ER) negative status. Therefore, HER2 gene status which is supposed as the ratio of the average number of the copy HER2 gene to the average number of copy Chromosome 17 (Chrl7) per cell is important in some cases. HER2 gene status is classified using the accepted rules. Different alterations in HER2 occur at the DNA and protein level and both in situ hybridization and immunohistochemistry are accurate methods to assess these alterations. The resolution of HER2 effects the clinical decisionmaking at the prognosis and treatment levels. The information about HER2 status of patients is also important to decide the use of therapy anti-HER2.
Breast cancer is the leading cause of cancer deaths in women which is a thoughtful menace to the health of the people. The most prevalent breast cancer types are commonly recognized using HER2-positive practice. The HER2-positive subtype is characterized by the expression of HER2, a transmembrane tyrosine kinase receptor of the family of human growthfactor receptors. As patients with HER2-positive breast cancer gain considerable benefit from special treatment with HER2 targeted therapies like monoclonal antibodies trastuzumab, pertuzumab and T-DM1 or the tyrosinekinase inhibitor lapatinib. Traditional approaches on FISH analysis are performed manually by clinician. This lets the results are highly dependent to human eye. Also FISH test colors constitutes of dark blue and black regions, it is reasonable that human eye will fail to distinguish between colors. Therefore, the success of computer vision algorithms compared to human eye in analyzing gene expression rate in FISH images will be discussed in this study. Different large FISH images were chosen for this study from pathology laboratory from Acibadem Maslak hospital. The proposed CAD scheme first applies preprocessing median and gaussian filters. An adaptive thresholding method followed by a watershed segmentation algorithm is employed to segment cells of possible interest. Furthermore, analyzable cells are found and nonanalyzable cells due to cell overlapping or image staining debris are discarded. The scheme then splits the detected analyzable region of interest into two red and green color spaces which is also followed by application of a scanning algorithm to detect the CEP17 (green) and HER2/neu (red) FISH signals separately. The results of the CAD-guided proposed tool would lead to a more efficient approach in determining HER2 status of probable patients.
Figure 1 shows summary results of a FISH image that counts within each cell HER2 sample status score.
Specifically, the proposed scheme segmented and detected a total of 113 cells depicted on the image.
In these 113 cells a total of 231 red and 51 green FISH signals were separately detected and counted.
In figure 2, as it is shown the method is implemented on another FISH image and the detected "red" and "green"
FISH signals were 234 and 36 respectively.The average HER2 amplification ratio for figure 1 was 0.22 (<2.2) and
this cancer case was classified as a HER2 negative. In figure 2 as this ratio was 0.15 (<2.2), this case was
classified as a HER2 negative also. Because of considerable cell overlapping and other stain related debris in
a number of images, no analyzable cells were segmented and detected in some regions in sample images. The sample
images which are not noisy and have analyzable cells could be detected with accuracy over 99 percent in the acquired dataset.
Using the digital HER2 FISH image database we proposed a system which automatically detects HER2 amplification status using FISH images.
Furthermore, we developed an image processing and viewing tools that include image viewing, preprocessing, and feature based
classification which is shown to user by a graphical user interface. An example of the procedure of FISH image in shown in figure 1.
The first task of the scheme is to find and detect analyzable interphase cells. This task has different steps that are as follows:
I) The scheme first splits RGB color space to three gray scale images. Besides two median and gaussian filters are applied. As shown in figure 1, morphology filters are also employed to remove small objects and fill inner parts.
II)The scheme uses an auto threshold percentile method whereas to binarize the image.
III) The scheme then uses binary image to build Euclidean Distance Map (EDM). The computed distance map and the eroded points are local maxima points which are seed points for watershed segmentation method.
IV) Once an acceptable region and cell is detected, it is selected from the original image and moved into a new stack image buffer and all of the stacks are saved as a montage image that helps pathologist and users view the results of the region of interest in an appropriate way.
Different processes of FISH image analysis