Hamail Ayaz

PhD Student / PEM Student Researcher


Research Topic: Deep Learning for early diagnosis and classification of Tumors from radiographic images

Hamail Ayaz.png

Hamail Ayaz is currently carrying out his PhD research in the Faculty of Engineering at IT Sligo in collaboration with GMIT, Sligo University Hospital, Connolly Hospital and Adnan Menderes University, Turkey.

Hamail completed his master's in Computer Engineering titled “Classification of Meat using Hyperspectral Imaging” and published two articles:

  1. Myoglobin-Based Classification of Minced Meat Using Hyperspectral Imaging.

  2. Hyperspectral Imaging for Minced Meat Classification Using Nonlinear Deep Features.


Prior to this, he studied in COMSATS Islamabad, where he completed his BS in Computer Science. During his past research on spectrometry and hyperspectral imaging, he has published different articles on food quality control with machine learning and deep learning techniques (see links below).


Hamail started his PhD in January 2021 and is expected to complete it by 2025. He has been awarded the IT Sligo CUA Bursary and will work on the project titled “Deep Learning for early diagnosis and classification of Tumors from radiographic images” under the supervision of Dr Saritha Unnikrishnan and Co-supervised by Dr David Tormey from IT Sligo and Dr Ian McLoughlin from GMIT.

Hamail's areas of expertise include Image Processing, Computer Vision, AI, Biomedical Imaging, Data analysis and Machine and Deep learning

Research Project:

The manual analysis of radiographic images techniques by neuro-radiologists is prone to human error and subjectivity. Additionally, Computer-Aided Diagnosis (CAD) systems have been used for the analysis of radiographic images with Machine Learning (ML) and Deep Learning (DL) techniques, which are limited to data availability, high processing units, complex systems, data segmentation and pre-processing.


The project aims to develop an efficient AI-assisted CAD system with the potential to overcome the above-mentioned limitations through hybrid architecture concepts as shown in Figure 1 and use of deep autoencoders/unsupervised learning techniques.

The Hybrid model using AlexNet and Squee

Figure 1: The Hybrid model using AlexNet and SqueezeNet architecture; CL is convolution layer; and PL is pooling layer

In a nutshell, an efficient CAD system will classify selected head and Neck tumour grades with minimal computational complexity. The primary objectives of this project are:


  1. Literature review and identification of knowledge gap in the area of Head and Neck tumour categories, image segmentation techniques and ML/DL classification algorithms. 

  2. Identification of MRI/CT image biomarkers of selected tumour categories through image segmentation followed by ML-based classification. 

  3. Deep Learning-based classification of the identified tumour categories and comparison of the ML and DL-based approaches. 

  4. Clinical endpoint review of the classification results in consultation with neuro-radiologists.

  5. PhD thesis preparation, and publications


Hamail has recently submitted a proceeding in MICAD 2021, and the article has been accepted in the Springer book series on Lecture Notes in Electrical Engineering (LNEE) (ISSN: 1876-1100).  


Hamail is a member of the European COST action GLiMR, (https://glimr.eu/) and is involved in coordinating the identification and quantification of advanced MRI biomarkers for the application in the field of glioma.



Hamail is an active researcher with several publications: https://scholar.google.com/citations?user=D8PJOWAAAAAJ&hl=en