Research

What we work on

Our research spans 2D/3D microscopic imaging, extended depth of focus, shape from focus, autofocusing, cellular analysis, drug–drug and drug–target interactions, drug repurposing and mechanism-of-action prediction.

01

Extended Depth of Focus

One sharp image from a focal stack

A light microscope sees only a thin slice in focus. We fuse multi-focus image stacks into a single, fully focused image with pyramid and Swin vision transformers, attention auto-encoders and learned focus measures — and we publish the multi-focus datasets we build.

Projects

  • TÜBİTAKDeep Learning Approaches for Extended Depth of Focus in Light Microscopy (2023–2024)

Publications

  • 2026An effective pyramid vision transformer-based extended depth of focus technique for optimally focused microscopic imaging The Journal of Supercomputing
  • 2026A Novel Neuro-Focus Fusion Framework for Fully Focused Microscopic Imaging The 12th International Conference on Computer Technology Applications (ICCTA 2026)
  • 2025Two-phase deep learning method for image fusion-based extended depth of focus The Journal of Supercomputing
  • 2025Deep Learning-Based Image Fusion with Novel Multi-Focus Dataset for Extended Depth of Focus in Microscopy International Conference on Smart Multimedia
  • 2024DL-EDOF: Novel Multi-Focus Image Data Set and Deep Learning-Based Approach for More Accurate and Specimen-Free Extended Depth of Focus Journal of Imaging Informatics in Medicine
  • 2023Investigation of Effects of Deep Features on Focus Values Extraction in Extended Depth of Focus Eskişehir Osmangazi Üniversitesi Mühendislik ve Mimarlık Fakültesi Dergisi
02

Shape from Focus & 3D Microscopy

Recovering 3D structure from focus

The same focal stack also encodes depth. From handcrafted focus operators to unsupervised and transformer-based deep models, we reconstruct accurate 3D shapes of specimens and blood cells, and develop deep-learning autofocusing for microscopes.

Projects

  • KTÜ BAPA Dataset and Machine Learning Algorithms for 3D Imaging of Blood Cells in Microscopic Systems (2023–2024)
  • TÜBİTAKNovel Deep Learning Approaches for Optimally Focused Images and 3D Shape Reconstruction in Microscopic Systems (2023–2024)
  • TÜBİTAKDeep Learning-Based Autofocusing in Light Microscopy (2023–2024)

Publications

03

Cellular Image Analysis

Segmenting and classifying cells

Accurate segmentation and classification of blood cells, fluorescent nuclei and cell-culture lines (HaCaT) with auto-encoder and U-Net families, supported by new expert-annotated microscopy datasets.

Projects

  • TÜBİTAKOptimal Imaging and Novel Deep Learning Approaches for 2D and 3D Cell Segmentation in Microscopic Systems (2025–2026)

Publications

  • 2026Creation of a Novel Dataset and VGG-UNet Based Deep Learning Approach for Automated Blood Cell Segmentation 11th International Health Sciences and Management Congress (IHMC 2026)
  • 2024Kan Hücrelerinin Optimum Odaklı Görüntülenmesi için Derin Öğrenme Tabanlı Yaklaşımın Geliştirilmesi Kahramanmaraş Sütçü İmam Üniversitesi Mühendislik Bilimleri Dergisi
  • 2024Performance Improvement in Blood Cell Segmentation with Deep Learning-based Image Fusion Approach 2024 8th International Artificial Intelligence and Data Processing Symposium (IDAP)
  • 2023Floresan Mikroskop Sistemlerinde Yüksek Doğruluklu Nükleer Segmentasyonu için Otomatik Kodlayıcı Tabanlı Modellerin Geliştirilmesi Karadeniz Fen Bilimleri Dergisi
  • 2023An Extensive Study: Creation of A New Inverted Microscope Image Data Set and Improving Auto-Encoder Models for Higher Accuracy Segmentation of HaCaT Cell Culture Line 2023 5th International Congress on Human-Computer Interaction, Optimization and Robotic Applications (HORA)
04

Drug Interactions & Repurposing

Predicting how drugs interact

Predicting drug–drug and drug–target interactions with convolutional and multi-head self-attention models, towards safer combinations and data-driven drug repurposing.

Publications

  • 2025Enhanced Multi-Head Self-Attention Transformer-Based Method for Prediction of Drug-Drug Interactions 2025 13th International Symposium on Digital Forensics and Security (ISDFS)
  • 2023İlaç-İlaç Etkileşimi Tahmini için Konvolüsyonel Sinir Ağı Tabanlı Yeni Bir Yaklaşım Süleyman Demirel Üniversitesi Fen Bilimleri Enstitüsü Dergisi
05

Mechanism of Action from Signals

Reading drug effects in contractility

Organ-bath experiments record how smooth muscle contracts under a pharmacological agent. We learn from these contractility signals to predict the agent's mechanism of action automatically — bringing deep learning to the pharmacology lab bench.

Projects

  • TÜSEBNovel Deep Learning Approaches for Automatic Prediction of the Mechanisms of Action of Agents Affecting Smooth Muscle Contractility (2024–2025)

Code & data

Publications

  • 2026Mechanism of Action of Pharmacological Agents: A New Deep Learning-based Prediction Model Using Smooth Muscle Contractility Signals The 12th International Conference on Computer Technology Applications (ICCTA 2026)
06

Clinical Decision Support

From lab models to patient care

Translating our methods to clinical questions, such as automatic staging of acute kidney injury (AKI) with deep learning.

Projects

  • TÜSEBAutomatic Staging of Acute Kidney Injury (AKI) with Deep Learning (2025–2026)