CS Other Presentations

Department of Computer Science - University of Cyprus

Besides Colloquiums, the Department of Computer Science at the University of Cyprus also holds Other Presentations (Research Seminars, PhD Defenses, Short Term Courses, Demonstrations, etc.). These presentations are given by scientists who aim to present preliminary results of their research work and/or other technical material. Other Presentations serve as a forum for educating Computer Science students and related announcements are disseminated to the Department of Computer Science (i.e., the csall list):

RSS Directions

Presentations Coordinator: Demetris Zeinalipour

PhD Defense: AM-FM Medical Image Analysis, Kyriacos Constantinou (University of Cyprus, Cyprus), Tuesday, December 13, 2022, 11:00-12:00 EET.

The Department of Computer Science at the University of Cyprus cordially invites you to the PhD Defense entitled:

AM-FM Medical Image Analysis

Abstract:
Amplitude Modulation - Frequency Modulation (AM-FM) models provide effective representations through physically meaningful descriptors of complex non-stationary structures that can differentiate between the different lesions and normal structure. The overall objective of this research work is to in-troduce new, more powerful, more robust and more discriminatory AM-FM models and methods, and to apply them in medical image analysis. More specifically the research contributions made during the course of this dissertation work are the following. A sparse multiscale AM-FM representation accompa-nied by robust feature extraction was developed which is based on families of Gabor filterbanks. The use of multiple families of Gabor filters leads to significantly better representations than using just one filterbank. The methodology includes decompositions and reconstructions based on scalable quality, multiscale and directional representations. Special emphasis was placed on the stability of the recon-structions. Furthermore, a new model of ultrasound image analysis of atherosclerotic carotid plaque to assess risk of stroke in asymptomatic patients was developed based on multiscale AM-FM analysis. A new methodology of sparse multiscale AM-FM analysis was developed using Gabor filterbanks. AM-FM component selection criteria and parametrization were introduced to derive discriminatory AM-FM fea-tures in predicting asymptomatic vs symptomatic plaque images. The new methodology achieved better results than texture analysis features. In addition, the combination of AM-FM features sets with clinical features proved to be helpful in increasing prediction accuracy. The proposed methodologies provide a new paradigm of AM-FM analysis that opens new horizons in the analysis of medical images towards differentiating between normal and abnormal tissue.

Short Bio:
Kyriacos Constantinou is a Ph.D. candidate at the Department of Computer Science under the supervision of Prof. Constantinos Pattichis. He is a member of eHealth lab of the Department of Computer Science and the Biomedical Engineering Research Centre of the University of Cyprus. His research interests include image processing and especially medical image analysis. He holds a Master's degree in Computer Science from the University of Cyprus and a Bachelor’s degree from the Hellenic Army Academy, Greece.

	Other Presentations Web: https://www.cs.ucy.ac.cy/colloquium/presentations.php
	Colloquia Web: https://www.cs.ucy.ac.cy/colloquium/
	Calendar: https://www.cs.ucy.ac.cy/colloquium/schedule/cs.ucy.pres.2022.constantinou.ics

Seminar: Convolutional neural networks with Hessian-free optimization for detecting Alzheimer’s disease in brain MRI, Dr. Maria Constantinou (University of Cyprus, Cyprus), Wednesday, November 23, 2022, 16:30-17:30 EET.

The Department of Computer Science at the University of Cyprus cordially invites you to the Seminar entitled:

Convolutional neural networks with Hessian-free optimization for detecting Alzheimer’s disease in brain MRI

Abstract:
Brain magnetic resonance imaging (MRI) is routinely used to diagnose and monitor neurological disorders. To automate this process, a number of machine learning algorithms have been proposed, with convolutional neural network (CNN) classifiers being very promising because they are effective for extracting patterns from large image datasets and solve the commonly-encountered problems of natural images. However, a main challenge of CNN models is the long time required for training. Most applications of CNN models for MRI classification use first-order algorithms for optimization, such as stochastic gradient descent (SGD) and adaptive momentum (Adam). We investigated using Hessian-free optimization (HFO), a second-order algorithm, to accelerate the training of CNN models for brain MRI classification. We trained CNN models to classify brain MRI with either SGD, Adam or HFO using T1-weighted brain MRI labelled as normal cognitive or Alzheimer’s disease obtained from the Alzheimer’s Disease Neuroimaging Initiative database (adni.loni.usc.edu). Overall, HFO converged at least two to four times faster than Adam and SGD. The performance of the models optimized with HFO was comparable or better than those optimized with SGD or Adam. In addition, HFO was more robust to hyperparameter changes than SGD and Adam. The results are according to theory and act as a proof of concept that second-order algorithms have an advantage in terms of training speed over first-order methods in training CNN models for MRI data classification.

Short Bio:
Maria Constantinou is a postdoctoral researcher and special teaching scientist at the Department of Computer Science at the University of Cyprus (UCY). She received the BSc(Hons) in Neuroscience with Industrial/Professional Experience (2012) and the PhD in Neuroscience (2017) from The University of Manchester. Before joining UCY, she was a postdoctoral researcher at Istituto Italiano di Tecnologia (IIT) in Italy (2017-2020). Her research interests focus on developing methodologies to detect aberrant neural activities which could be used as biomarkers of brain pathology and assist in developing interventions that will improve patients’ health outcomes. Her current research focuses on developing machine learning applications for analysing neuroimaging data to identify mild cognitive impairment and dementia.

Note:
This is a dissemination seminar of the project MRI-brain, funded by the ONISILOS Postdoctoral Research Programme 2018 of the University of Cyprus.

	Other Presentations Web: https://www.cs.ucy.ac.cy/colloquium/presentations.php
	Colloquia Web: https://www.cs.ucy.ac.cy/colloquium/
	Calendar: https://www.cs.ucy.ac.cy/colloquium/schedule/cs.ucy.pres.2022.constantinou.ics