Faculty Profiles

Dr Khurram Ejaz

Designation : Lecturer
Specialization : Medical Image Processing

Email : khurram.ejaz@uog.edu.pk

Dr. Khurram Ejaz did his PhD in 2020 from University Technology Malaysia (UTM) Malaysia (University QS rank is 187, Faculty QS rank is 100). He did his Master of Sciences (MSCS) in computer science from University Central Punjab (UCP), Lahore Pakistan in 2010. He had served as lecturer in Federal Urdu University of Arts science and Technology (FUUAST) Islamabad Pakistan from 2008 to 2010. He is currently serving as senior lecturer in University of Gujrat Pakistan. He is also member of virtual reality lab (Vicube lab) in UTM. He is author of various index articles (Impact factor and Scopus index). His research interests are in Pattern Recognition and Computer Vision.

  • This research statement is divided into three parts, in part one, some published novelty is discussed. Secondly, current research problem and objectives are focused; lastly the methodology was proposed. Main believed is to serve humanity therefore; the scope of the work is applied. Literature review (review based/systematic literature review based /experimental based) is gone through in detail. The main aim of this research is to help surgeon before brain tumour operation. Different datasets like Sheikh Zaid Hospital Lahore Pakistan (Published at IJACSA), Harvard brain tumor repository (Published dataset) are analyzed. Brain tumor images classification is performed with the help of machine learning techniques. With these algorithms label of dataset is performed which classifies image as tumourous or non-tumourous. In research methodology, more techniques are belonging to Supervised Learning/ Un-Supervised Learning; Set of features (texture features, statistical features) are extracted, reduced and selected. Flavors of SVM (Linear Kernel, Radial based kernel, Polynomial kernel, quadratic kernel) along k fold training are finding classification accuracy of each individual feature. From 2017 till 2020, recent dataset is explored namely as Brats MICCAI brain tumor segmentation challenge; images are taken from year 2013 to year 2018. Machine Learning (ML) and Image processing techniques are helpful. The scope of this work is to segment the brain tumor with accuracy and have verified accuracy of results as compare to state of the art studies. Proposed Un-supervised approach is experimented to recognize the tumor pattern with overlap of brain tumor with high percentage. Three combinations of clusters are proposed. In this research 4-5K images are segmented and this work is also capable of classifying more than 200,000 images and suggested technique is so much intelligent to segment complex intensities.