Signet Ring Cell Detection from Histological Images
Muhammad Faheem Saleem, Department of Computer Science, University of Engineering and Technology, Taxila, Pakistan.
Marriam Nawaz, Department of Software Engineering, University of Engineering and Technology, Taxila, Pakistan.
Tahira Nazir, Department of Computing, Riphah International University, Islamabad, Pakistan.
Awais Mehmood, Department of Computer Science, University of Engineering and Technology, Taxila, Pakistan.
Momina Masood, Department of Computer Science, University of Engineering and Technology, Taxila, Pakistan.
Corresponding Author:
Tahira Nazir (tahira.nazir@riphah.edu.pk)
Abstract:
Signet Ring Cell (SRC) carcinoma is one of the most serious types of cancer, which is one of the major causes of deaths worldwide. Diagnosis of SRCs at early stages is a challenging task due to the smaller size and color variations. Early detection of SRCs from histological images requires human experts which makes the process costly. So, there is a need for an automated system which can detect SRCs at early stages from histological images. To cope with the existing problem, we have presented the Deep learning (DL) based method Namely Fast-RCNN for SRCs detection at early stages. Initially, we generated the bounding box (bbox) annotations which is essential for training of our method. The feature extraction is performed through ResNet-50, and then localized the SRCs by using Fast-RCNN. Our proprietary dataset is used for experimentation and evaluation of proposed approach and achieved 86.9% accuracy. We have performed the comparative analysis and attained better results than other methods.
Keywords:
Deep Learning; Signet Ring Cell; Histological Images; Cancer Detection