Computational Imaging Group Computational Imaging Lab

The Computational Imaging Lab performs AI research in medical imaging, advancing diagnostics and treatment across multiple organs and modalities. Our work focuses on classification, segmentation, image translation, visual question answering, and image captioning. Sample work includes:

Brain Imaging

  • Image Translation: Paired-unpaired unsupervised attention-guided GAN produce improved MR-CT image synthesis, validated by radiologists and advanced loss functions. [Publications 1, 2, 3 & 4]
  • Segmentation: Deepfake-based synthetic data enhances brain tumor segmentation, outperforming traditional methods. [Publications 1 & 2]
  • Histopathology: Fractal analysis quantifies tumor microvascular networks and histological features, improving diagnostic accuracy. [Publications 1, 2, 3, 4, 5 & 6]
  • Visual Question Answering: Large language model-based framework aligns multimodal datasets in minutes, surpassing six VQA models in aiding radiologists. [Publications 1]

Lung Imaging

  • Classification: Noise-robust texture measures, such as wavelet packet and fractal dimensions, outperform conventional methods. [Publications 1 & 2]
  • Deepfake Detection: Benchmarking of DCNNs highlights ResNet50V2 and DenseNet169 as optimal for detecting deepfakes. [Publications 1]
  • Fractal Analysis: Contrast-enhanced CT-based fractal methods achieve improved accuracy in predicting tumor aggression and correlate with PET findings. [Publications 1 & 2]

Prostate Imaging

  • Segmentation: Automated k-means clustering and U-Net systems enhance prostate tissue gland segmentation and Gleason grading accuracy. [Publications 1, 2, 3 & 4]

Gastroenteropancreatic Imaging

  • Classification: Semi-automated Ki-67 index quantification demonstrates high agreement with manual grading, streamlining pathologist workflows. [Publications 1]

Heart Imaging

  • Segmentation: Fractional Brownian motion-based methods accurately segment the left ventricle in 3D echocardiography. [Publications 1 & 2]

Liver Imaging

  • Fractal Analysis: Nakagami-based fractal descriptors improve tissue heterogeneity characterization and therapy response prediction. [Publications 1, 2, 3 & 4]

Learning and Image Captioning

  • Self-Taught Learning: Semi-supervised methods identify learning styles, enabling personalized educational tools. [Publications 1 & 2]
  • Image Captioning: Transformer-based ensemble models achieve high-quality captions for datasets like Flickr8K and Flickr30K. [Publications 1]

Security and Anomaly Detection

  • Deepfake Detection: Vision Transformers with knowledge distillation improve detection of facial manipulations in deepfake videos. [Publications 1]
  • Federated Learning: Secure frameworks enable collaboration for detecting cyber threats in imaging data. [Publications 1]
  • Anomaly Detection: Population-based algorithms identify anomalies in imaging and computer vision tasks. [Publications: 1, 2, 3]

Current and past members

PhD students

  • Fatima Haimour (Unsupervised Medical Image Translation for Lesion Semantic Segmentation) [Publications: 1, 2]
  • Israa Al Badarneh (An Attention-Based Transformer Model for Arabic Image Captioning) [Publications: 1]

MSc students

  • Alaa AlZgoul (Facial Expression Manipulation Detection in Deepfake Videos through Knowledge Distillation of Vision Transformers)

Visiting researchers

  • Faraz Janan (Senior lecturer – Imperial College London, UK)
  • Daniella Crisan (Associate Professor – Romanian-American University, Romania)

Research assistants

  • Roua Al-Emaryeen [Publications: 1, 2, 3]
  • Sara Al-Nahhas [Publications: 1, 2, 3]

Past PG students

  • Abdullah Zaqebah (May 2022, PhD thesis: An Enhanced Population-based Nature-inspired Technique for Anomaly Detection Applications) [Publications: 1, 2, 3] – Now Assistant Professor at Al-Ahliyya Amman University.
  • Alaa Abu Srhan (Dec 2021, PhD thesis: Generative Adversarial Networks Modelling with Transfer Learning for Medical Image Analysis) [Publications: 1, 2]  – Now lecturer at Hashemite University.
  • Waref Al Manaseer (Aug 2021, PhD thesis: Arabic language diacritization using restricted Boltzmann machine) [Publications: 1] – Now Assistant Professor at Applied Science Private University.
  • Ala Wrikat (Aug 2024, MSc thesis: Cross-Banks Horizontal Federated Learning Framework for Cyber Threat Intelligence) – Now at Central Bank of Jordan.
  • Haya Mustafa (Jun 2023, MSc thesis: A Lightweight Computational Neural Network Model for Detecting and Classifying Leukemia) – Now at Ministry of Higher Education and Scientific Research.
  • Mohammad Alqawasmi (Aug 2022, MSc thesis: Detecting Face Tampering in Videos using DeepFake Forensics) – Now developer at OPTIMIZA.
  • Eman Nasser (Aug 2021, MSc thesis: An optimised feature selection for fishing websites detection) – Now IT Manager at Dar-Alhai Business Group.
  • Feda Beqain (July 2020, MSc thesis: Histopathological web diagnosis system for prostate cancer) [Publications: 1] – Now developer at Ministry of Information and Communication Technology.
  • Mu’nis Qasaymeh (Dec 2020, MSc thesis: Breast cancer grading for unlabelled data using crowdscouring) – Now developer at Orange Jordan.
  • Hani Ayyoub (Dec 2020, MSc thesis: Personlaized learning style identification using data mining and machine intelligence) [Publications: 1] – Now senior programmer at e-learning unit at the University of Jordan.
  • Jamileh Azzam (Aug 2022, MSc thesis: A Multi-factor Authentication Framework for the Cloud)
  • Safaa Al-Haj Saleh (May 3013, MSc thesis: Histopathological prostate tissue gland segmentation for automated diagnosis) [Publications: 1, 2] – Lab Supervisor at Hashemite University.
  • Nada Misk (Dec 2014, MSc thesis: Automatic detection of unusual crowd behavior in real-time video surveillance systems) – PhD student at Istanbul University.
  • Nicolas Gallego-Ortiz (May 2011, MSc thesis: Algorithm for mass detection in mammography based on local statistical texture model) – PhD student at University of Bern.