Ujian palitan Pap adalah ujian yang dijalankan untuk mengesan barah pangkal rahim. Bagaimanapun, ujian Pap mempunyai beberapa kekurangan terutamanya tersilap diagnosis kerana kesilapan manusia dan teknikal. Sebelum ini, teknik pemproses imej digital yang dinamakan algoritma ‘seeded region growing features extraction’ (SGRFE) telah berjaya digunakan sebagai teknik segmentasi untuk mengekstrak empat ciri sel pangkal rahim tunggal iaitu: saiz nukleus, saiz sitoplasma, paras kelabu nukleus dan paras kelabu sitoplasma. Terdapat spesimen yang dipalitkan tidak sekata ke atas slaid kaca dan kemungkinan akan wujud palitan sel tunggal, sel-sel bertindih atau sekumpulan sel. Semasa pengaplikasian algoritma ‘seed based region growing’ (SBRG) untuk sel pangkal rahim yang bertindih, proses pensegmenan akan dilaksanakan sama ada terhadap satu sel atau kedua-dua sel bersama. Masalah ini telah mendorongkan beberapa kajian untuk membezakan sel pangkal rahim yang bertindih untuk mendapatkan saiz dan paras kelabu bagi nukleus dan sitoplasma untuk setiap sel dan menentukan kawasan pertindihan. Kajian ini mencadangkan satu teknik pemprosesan imej digital untuk mengurangkan masalah tersebut. Teknik yang dicadangkan dibahagi kepada beberapa peringkat. Pertama, konsep warna ruang diaplikasikan untuk mengekstak imej asal palitan Pap kepada tiga paparan iaitu merah, hijau dan biru dengan keamatan warna yang berbeza. Kemudian, algoritma SBRG digunakan untuk mensegmen kawasan yang dikehendaki berdasarkan paras kelabu sel. Teknik warna palsu diaplikasikan keatas kawasan yang telah disegmen untuk membezakan kawasan nukleus, sitoplasma dan laterbelakang. Algoritma ‘pseudo color features extraction’ (PCFE) digunakan untuk mengekstrak saiz nukleus, saiz sitoplasma, paras kelabu nukleus dan aras kelabu sitoplasma. Keputusan menunjukkan saiz dan paras kelabu nukleus dan sitoplasma bagi setiap sel servikal dan rantau pertindihan berjaya ditentukan. Teknik pemprosesan yang dicadangkan telah berjaya meningkatkan segmentasi untuk sel servikal yang bertindih. Oleh itu, imej paduan itu akan menjadi lebih berguna untuk analisis lanjut.
_____________________________________________________________________________________
The Pap smear is generally used as a screening test to screen cervical precancerous or cancer. However, Pap test has some limitations. Thus, the screening result produced could be miss-diagnosed due to the human and technical errors. Previously, seeded region growing features extraction (SRGFE) algorithm has successfully been used as a segmentation technique to extract four features of a single cervical cell; size of nucleus, size of cytoplasm, grey level of nucleus and grey level of cytoplasm. But during Pap smear slide preparation, the cervical cell may spread unevenly and the image could be smeared as a single cell, overlapping cells or a group of cells. When applying the seed based region growing (SBRG) algorithm for overlapping cervical cells, it either extracts only one cell or both cells together. This problem has encouraged several studies to distinguish the overlapping cervical cell to obtain the size of nucleus and cytoplasm of each cell and determine the overlapping region. This study proposes a digital image processing technique to reduce the aforementioned problems. The proposed technique is divided into several stages. Firstly, the color space concept is applied to extract the original image of Pap smear into red plane, green plane and blue plane then the SBRG is applied to segment the region of interest of the cells. Pseudo color technique will then be applied to the demarcated region to determine each part of the cells; nucleus, cytoplasm and background. Pseudo color features extraction (PCFE) algorithm is applied to extract the size of nucleus, size of cytoplasm, grey level of nucleus and grey level of cytoplasm. The results
xiii
show that the size and grey level of nucleus and cytoplasm of each cervical cell and overlapping region are successfully determined. The proposed technique is capable to distinguish each cervical cell from overlapping cervical cells image. Therefore, the resultant image will be more useful for further analysis.
Implementation of image processing technique for overlapping cervical cell_Ooi Tze Theng_E3_2009_875004516_00003095337_NI