Where Data Meets Computer Science


Boguslaw’s Lab interdisciplinary research focuses on advancing the state of the art in image/video analysis, processing and understanding, with potential for applications in biology, medicine, arts and humanities, and engineering.


The local Gaussian distribution fitting (LGDF) energy model is a state-of-the-art method, capable of segmenting inhomogeneous objects with poorly defined boundaries, but it is computationally expensive. In our approach, we port the LGDF energy functional to the GPU, and introduce a novel set of interactive brush functions to segment challenging datasets where an active contour would not ordinarily capture the target object. Furthermore, we expose a smaller and more intuitive parameter space to the user, and enhance usability by including a built-in ray tracer to visualise the evolving 3D segmentation results in real time. Quantitive and qualititive validation is presented, demonstrating both a faithful port of the original algorithm and the practical efficacy of our interactive elements, for a wide variety real-world datasets.


Deep Learning

  • Image features-based deep learning system for counterfeit goods detection.
  • Modelling human factors to predict form conversion for consumer goods products.



Evolution-in-materio concerns the computer controlled manipulation of material systems using external stimuli to train or evolve the material to perform a useful function. In this paper we demonstrate the evolution of a disordered composite material, using voltages as the external stimuli, into a form where a simple computational problem can be solved. The material consists of single-walled carbon nanotubes suspended in liquid crystal; the nanotubes act as a conductive network, with the liquid crystal providing a host medium to allow the conductive network to reorganise when voltages are applied. We show that the application of electric fields under computer control results in a significant change in the material morphology, favouring the solution to a classification task.


Macular Oedema

The objectives of this project are to develop and assess novel image analysis techniques of 3D OCT scans for volumetric assessment of different morphological patterns in diabetic macular oedema (DMO) such as:

  • Overall volume of the generalised outer retinal thickening in early DMO.
  • Volume of accumulated serous fluid under the neurosensory retina (SRF).
  • Residual volume of retinal tissue passing between the inner and outer plexiform retinal layers, which is an optimal measurement of potential residual macular function.


In this project we introduce a new method for the detection of parametrisable shapes in N dimensions; this method has been developed for ellipses and uses simple morphological operations, building on aspects of granulometry and signal processing techniques. With this method it is possible to easily extract the position, size and rotation of elliptical objects in any image data. This method is low parameter, accurate and robust to noise and object clustering in greyscale images; key contributions are the abiltiy to find an unknown number of ellipses with no a priori information and no arbitrary thresholding and robustness to both noise and clustering.



This project focuses on a robust method for extracting a natural interpolating 3D piecewise cubic spline from a 2D input image of a helix object. We are able to use the generated spline to extract 3D metrics such as tortuosity and curvature, and validate the results against real-world samples at both macro and microscopic levels. The algorithm analytically chooses locations to sample the image to extract properties of the curve, such as its amplitude and perpendicular width, to ensure robustness. The generated 3D spline has few input parameters, and only requires a single view of the 2D dataset, making it suitable for a range of applications in engineering, physics, and biology.



Runways are vital descriptive features of airports and knowledge of their location is important to many aviation and military applications. With the recent wide availability of remote sensing data, there is demand for an automatic process of extracting runway geometry from satellite imagery. In particular, Very High Resolution (VHR) data makes it feasible to extract a runway’s area precisely. In this project we establish a novel method for accurate and precise extraction of geometric polygons for an arbitrary number of runways in VHR remote sensing imagery.


Big Data

Projects focuses on big data challenges including capturing data, data storage, data analysis, search, sharing, transfer, visualization, querying, updating and information privacy.


Leaf Vein Networks

Quantitative characterisation of the leaf vein network topology

Vein network density reflects how much energy and resources the leaf has invested in the network, while distance between veins shows how well the veins are supplying resources to the leaf. The number of loops is a measure of the leaf’s resilience and plays a role in determining its lifespan. If the veins reconnect often and part of the leaf becomes damaged, resources can be circulated through different pathways.

We have developed image processing approach for leaf vein network extraction, analysis and graph-based  representation.


Cell Tip

Analysis and understanding of fungal cell tip growth

Fungi cause devastating plant and human diseases. There is considerable evidence that much of the cellular machinery driving growth of invasive fungal hyphae is common across all fungi, including plant and mammalian pathogens, and involves localized tip growth.
We have developed and evaluated high-throughput automated microscope-based multi-dimensional image analysis systems to segment and characterize fungal growth, and characterize the patterns of protein localization within the tip that control development.


Cytoskeletal Networks

Data driven approaches to understanding cytoskeletal network complexity

A 3D/4D cytoskeletal network may contain thousands of nodes and connections, in specific but variable geometric organisations which vary with time. Such data exceeds the capacity of human analysis, creating a barrier, which can only be overcome with robust, automated image analysis and informatics tools to extract, characterise and model networks. Having such topological characteristics, some key open questions that can then be explored are:

  • How are topology and function related to each other?
  • What are the underlying dynamics that shape the network patterns?
  • How important is the formation of alternative paths through the network via crosslinks?
  • How does the network evolve to confer high robustness?
  • How local responsive changes affect the global network?


Cell Protrusions

Identification, description and quantification of the cell behavior observed by time-lapse confocal microscopy

Research, development and implementation of novel 4D image analysis and processing methods for identification, description and quantification of the cell behavior (protrusion, adhesion and invasion), as observed by time-lapse confocal microscopy. (more…)


An image processing approach for quantitative understanding of early-print books

In their research, historians often work with dated manuscripts and early-print books, as these can often provide valuable, authentic information that may not be possible to obtain from other sources. However, despite their popularity with historians, a number of problems can arise when using manuscripts and early-print books in research. One of the most significant problems is that these sources can be challenging to read. This is due to the combination of difficult-to-read handwriting, and also noise on the page (such as textures, damage, or ink seepage). This project therefore proposes the developement of a software system which will facilitate the efficient digitalisation of primary sources, which in turn, will ease the burden on the historian. The project will take a skeletonization based image processing approach, which aims to extract and quantitatively describe the shapes of letters on pages of early-print books, despite any noise on the sample. This will allow the text in the source to then be presented to the historian in an easy-to-read, digital format. In order to realise this aim, the project will involve looking into noise reduction techniques in image processing, image segmentation techniques and algorithms, and character recognition techniques.



Bacterial cells identification in Differential Interference Contrast (DIC) microscopy images

Microscopy image segmentation lays the foundation for shape analysis, motion tracking, and classification of biological objects. Despite its importance, automated segmentation remains challenging for several widely used non-fluorescence, interference-based microscopy imaging modalities. For example in differential interference contrast microscopy which plays an important role in modern bacterial cell biology. Therefore, new revolutions in the field require the development of tools, technologies and work-flows to extract and exploit information from interference-based imaging data so as to achieve new fundamental biological insights and understanding.

We have developed and evaluated a high-throughput image analysis and processing approach to detect and characterize bacterial cells and chemotaxis proteins. Its performance was evaluated using differential interference contrast and fluorescence microscopy images of Rhodobacter sphaeroides.


Zebrafish Lens

Quantitative understanding of the geometrical order of the lens

The eye lens is a tissue in which its cell structure and cell organisation is intimately linked to function.  We are combining cell biology, imaging, image processing and modeling to understand the geometrical order of the lens.


Cell Tracking

Linking BRCA1-regulated processes with cell motility pathways

With the increasing availability of live cell imaging technology, tracking cells and other moving objects in live cell videos has become a major challenge for bioimage informatics. In particular, studying cell motility has become an important factor in understanding numerous biological processes such as tissue repair, metastatic potential, chemotaxis, or the analysis of drug performance. Cell tracking has therefore become a major application for biological image processing.


Active Mesh

Speeding up active mesh segmentation by local termination of nodes

Active meshes and other deformable models are very popular in image segmentation due to their ability to capture weak or missing boundary information; however, where strong edges exist, computations are still done after mesh nodes have settled on the boundary. This can lead to extra computational time whilst the system continues to deform completed regions of the mesh. We propose a local termination procedure, reducing these unnecessary computations and speeding up segmentation time with minimal loss of quality.



An image processing approach for colocalization analysis of objects in dual-color confocal images

The colocalization technique is important to many cell biological and physiological studies during the demonstration of a relationship between pairs of bio-molecules.


Fungal Networks

Microbial biology: investigating critical developmental transitions via imaging, image analysis and mathematical modelling

Saprotrophic fungi are critical in ecosystem biology as they are the only organisms capable of complete degradation of wood in temperate forests. These fungi form extensive interconnected mycelial networks that scavenge efficiently for scarce resources in a heterogeneous environment. The architecture of the network continuously adapts to local nutritional cues, damage or predation, through growth, branching, fusion or regression. Such biological networks, honed by evolution, may exemplify potential solutions to real-world compromises between cost, coverage, resilience and persistence. We need to be able to automatically analyse the dynamic network architecture to evaluate their performance efficiently. We propose an image processing approach for fungal network segmentation and graph-based analysis. (more…)

Ascidian Cells

Segmentation of ascidian notochord cells in DIC timelapse images

We have developed a method to automatically segment notochord cell boundaries from differential interference contrast (DIC) timelapse images of the elongating ascidian tail. The method is based on a specialized parametric active contour, the network snake, which can be initialized as a network of arbitrary but fixed topology and provides an effective framework for simultaneously segmenting multiple touching cells. Several modifications to the original network snake were necessary for high-quality segmentation, including linear Gaussian derivative filtering to reconstruct edge maps from DIC images and a new energy function to improve the segmentation of critical cell-cell vertices.