Sel-sel dendrit (DCs) sangat berkesan dalam mendorong tindak balas imun yang kini merupakan tumpuan utama dalam imunoterapi kanser untuk merawat kanser. Mengenal pasti dan mengira DC adalah penting untuk merangsang atau memanipulasi DC dalam mendorong tindak balas imun untuk imunoterapi. Sistem pengimejan perubatan dibangunkan untuk membantu manusia untuk mengenal pasti dan mengira DC dari berates-ratus sel dan memastikan daya maju dan kesihatan DC yang tidak dapat dijamin dengan menggunakan teknologi semasa. MATLAB digunakan untuk menghasilkan algoritma system pengimejan. Imej-imej perubatan telah disediakan oleh Cancer Research Malaysia. Pertama, imej sel darah telah diproses menggunakan teknik peningkatan imej berdasarkan histogram untuk meningkatkan kualiti imej. Kemudian, segmentasi imej telah digunakan untuk segmen DC berpotensi dalam imej. Seterusnya, proses pengekstrakan ciri digunakan untuk mendapatkan ciri-ciri penting DC berpotensi. 6 ciri-ciri geometri dan 20 ciri-ciri statistik diekstrak daripada DC berpotensi tersebut. Ciri-ciri ini telah digunakan untuk mengklasifikasikan DC dan membezakannya daripada sel-sel lain dengan menggunakan Artificial Neural Network (ANN). Rangkaian ANN dengan seni bina MLP digunakan dalam kajian ini dan mencapai keseluruhan klasifikasi ketepatan terbaik 90.1%. Bilangan DC kemudiannya dikira dan direkodkan. Sistem ini telah diproseskan untuk dijadikan aplikasi Graphic User Interface (GUI). Kesimpulannya, projek ini membantu dalam imunoterapi kanser dengan mengautomasikan klasifikasi dan proses pengiraan DC di kalangan beratus-ratus atau beribu-ribu sel darah tanpa merosakkan DC. Sistem pengimejan akhir dijangka membantu manusia dalam proses mengenal pasti DC dalam kadar dan ketepatan yang lebih tinggi.
_______________________________________________________________________________________________________
Dendritic cells (DCs) are very efficient in inducing immune responses which are now a major focus in cancer immunotherapy to treat cancer. Identifying and counting of DCs are essential in order to stimulate or manipulate it in inducing immune responses for cancer immunotherapy to treat cancer. A medical imaging system was developed to assist human eye to identify and count DCs from hundreds of cells and at the same time assure viability and health of DCs which is not guaranteed by using current technology. MATLAB was used to develop image processing algorithm of the system. The medical images were provided by Cancer Research Malaysia. First, the image of blood cells was processed using histogram based image enhancement technique to improve its quality. Then, image segmentation was employed to the resultant image to segment potential DCs in the image. Next, feature extraction process was applied to the segmented potential DCs in order to extract significant features of potential DCs. 6 geometrical and 20 statistical features were extracted from the segmented potential DCs. These features was then used to classify DCs and distinguish it from other cells by using Artificial Neural Network (ANN). ANN network with MLP architecture was employed in this study to classify DCs and other cells and achieved the best overall classification accuracy of 90.1%. Number of DCs were then counted and recorded. The system was developed into a Graphic User Interface (GUI) application. The application is able to automate all the DCs identification process. In conclusion, this project helps in cancer immunotherapy by automating the classification and counting process of DCs among hundreds or even thousands of blood cells without damaging the DCs. The final imaging system is expected to assist human inspection in the process of identifying DCs in a much higher rate and accuracy.