A Comprehensive Analysis: Automated Ovarian Tissue Detection Using Type P63 Pathology Color Images
Manual microscopic ovarian reproductive tissue analysis is a general routine examination process in the laboratory. This process requires longer processing time and prone to errors. Among all existing scanning devices ultrasound is commonly used but not o
- PDF / 3,692,531 Bytes
- 14 Pages / 439.37 x 666.14 pts Page_size
- 53 Downloads / 214 Views
Edith Cowan University, Joondalup, WA, Australia {t.sazzad,l.armstrong}@ecu.edu.au 2 Don Bosco Institute of Technology, Mumbai, India [email protected]
Abstract. Manual microscopic ovarian reproductive tissue analysis is a general routine examination process in the laboratory. This process requires longer processing time and prone to errors. Among all existing scanning devices ultrasound is commonly used but not optimal as it process grayscale images which do not provide satisfactory results. Computer based approaches could be a viable option as it can minimize the labor cost, effort and time. Additionally smaller tissues can be easily analyzed. In this paper a comprehensive analysis has been carried out and a new modified approach has been presented using type P63 histopathology ovarian tissues color images with different magnifications. Comparison of various existing automated approaches with manual identification results by experts indicates excellent performance of the proposed automated approach. Keywords: Histopathology · Color digitized microscopic image · Image artifacts · Mean shift · Region fusion · Cluster · Ovarian reproductive tissues
1
Introduction
Modern digital scanners including ultrasound scanner, MRI (magnetic resonance imaging), CT (computerized tomography), PET (position emission tomography) are commonly used diagnosis modalities to analyze tissues in the pathology laboratory [1]. These modern digital scanners can quickly perform laboratory routine task and requires less human effort. Ultrasound is considered as commonly used option among all modalities in the pathology laboratory as it is cheaper, portable and less risky to patients [1]. However; it can only process grayscale images and suitable for large tissue analysis such as cancer tissues [2]. Smaller tissues are relatively hard to analyze using ultrasound scanners [2]. Ovarian tissues are smaller in size in compare to cancer tissues [1], [9]. Therefore; pathology experts prefer manual microscopic approach [9]. Research work of [3] mentioned that manual microscopic biopsy slide analysis is considered as “gold standard” for human ovarian tissue analysis. However; this method requires longer processing time and has observation variability issues [4]. Computer based approach could be more viable to overcome the issues associated with manual microscopic approach [4, 5]. Existing research study indicates that most computerized approaches are semiautomated rather automated and most analysis have been carried out on ovarian © Springer International Publishing Switzerland 2016 P. Perner (Ed.): MLDM 2016, LNAI 9729, pp. 714–727, 2016. DOI: 10.1007/978-3-319-41920-6_54
A Comprehensive Analysis: Automated Ovarian Tissue Detection
715
cancer cells or tumor cells rather than ovarian reproductive tissues [6]. Cancer cells or tumor cells analysis approaches are not suitable due to the fact that they have different shape, size and color in compare to ovarian reproductive tissue [6]. At present, only a few research works have been carried out on ovarian
Data Loading...
