STARSTEM Publications

STARSTEM Journal Articles

2020

  • Dimitri A.Kessler, James W.MacKay, Victoria Crowe, Frances Henson, Martin J. Graves, Fiona J. Gilbert, Joshua D. Kaggie (2020), The Optimisation of Deep Neural Networks for Segmenting Multiple Knee Joint Tissues from MRIs.” Computerized Medical Imaging and Graphics. DOI: https://doi.org/10.1016/j.compmedimag.2020.101793
    Download the pdf here.

Automated semantic segmentation of multiple knee joint tissues is desirable to allow faster and more reliable analysis of large datasets and to enable further downstream processing e.g. automated diagnosis.

In this work, we evaluate the use of conditional Generative Adversarial Networks (cGANs) as a robust and potentially improved method for semantic segmentation compared to other extensively used convolutional neural network, such as the U-Net. As cGANs have not yet been widely explored for semantic medical image segmentation, we analysed the effect of training with different objective functions and discriminator receptive field sizes on the segmentation performance of the cGAN. Additionally, we evaluated the possibility of using transfer learning to improve the segmentation accuracy. The networks were trained on i) the SKI10 dataset which comes from the MICCAI grand challenge “Segmentation of Knee Images 2010”, ii) the OAI ZIB dataset containing femoral and tibial bone and cartilage segmentations of the Osteoarthritis Initiative cohort and iii) a small locally acquired dataset (Advanced MRI of Osteoarthritis (AMROA) study) consisting of 3D fat-saturated spoiled gradient recalled-echo knee MRIs with manual segmentations of the femoral, tibial and patellar bone and cartilage, as well as the cruciate ligaments and selected peri-articular muscles. The Sørensen–Dice Similarity Coefficient (DSC), volumetric overlap error (VOE) and average surface distance (ASD) were calculated for segmentation performance evaluation.

DSC ≥ 0.95 were achieved for all segmented bone structures, DSC ≥ 0.83 for cartilage and muscle tissues and DSC of ≈0.66 were achieved for cruciate ligament segmentations with both cGAN and U-Net on the in-house AMROA dataset. Reducing the receptive field size of the cGAN discriminator network improved the networks segmentation performance and resulted in segmentation accuracies equivalent to those of the U-Net. Pretraining not only increased segmentation accuracy of a few knee joint tissues of the fine-tuned dataset, but also increased the network’s capacity to preserve segmentation capabilities for the pretrained dataset.

cGAN machine learning can generate automated semantic maps of multiple tissues within the knee joint which could increase the accuracy and efficiency for evaluating joint health.

What are you aiming to find out?

We used machine learning to automatically segment multiple knee tissues, including bones, cartilage, muscles, and cruciate ligaments. This will help us enable better quantitative analysis of these tissues in future studies, and use the best machine learning network to do this.

Why does this research need to be done?

Tissue segmentation is a time-consuming process, but very useful for obtaining quantitative metrics – such as shape or signal intensity. Machine learning can accelerate this.

While machine learning is increasingly performed for the segmentation of tissues, this work shows that it can be used for many (10) different tissues, with highly varying characteristics. We expect this type of work to be constantly iterated on, although we were the first to show all of these tissues together and tissues like the ACL/PCL.

Describe the methods chosen.

We used several deep learning methods to automatically segment knee tissues – called the U-Net and the GAN, and compared these with different parameters. We wanted to know the best process to perform this measurement, as it can speed up analysis in future sets and patients.

We used publicly available knee datasets (SKI10, ZIB) and local data (AMROA).

The AMROA participants were questioned on their patient experience as they left the MRI system.

What are the next steps?

We would like to apply this to looking at how signal intensities change with the introduction of stem cells or nanoparticles.

  • Kessler, D.A., MacKay, J.W., McDonald, S., McDonnell, S., Grainger, A.J., Roberts, A.R., Janiczek, R.L., Graves, M.J., Kaggie, J.D. and Gilbert, F.J. (2020), Effectively Measuring Exerciserelated Variations in T1ρ and T2 Relaxation Times of Healthy Articular Cartilage.” Journal of Magnetic Resonance Imaging. DOI: 10.1002/jmri.27278
    Download the pdf here.

Background: Determining the compositional response of articular cartilage to dynamic joint loading using magnetic resonance imaging may be a more sensitive assessment of cartilage status than conventional static imaging. However, distinguishing the effects of joint loading versus inherent measurement variability remains difficult as the repeatability of these quantitative methods is often not assessed or reported. Purpose: To assess exercise-induced changes in femoral, tibial and patellar articular cartilage composition and compare these against measurement repeatability. Study Type: Prospective observational study. Population: Phantom and 19 healthy participants. Field Strength/Sequence: 3T; 3D fat-saturated spoiled gradient recalled-echo; T1ρ- and T2-prepared pseudo-steady-state 3D fast spin echo. Assessment: The intra-sessional repeatability of T1ρ and T2 relaxation mapping, with and without knee repositioning between two successive measurements, was determined in 10 knees. T1ρ and T2 relaxation mapping of nine knees was performed before and at multiple time points after a 5-minute repeated, joint-loading stepping activity. Three-dimensional surface models were created from patellar, femoral and tibial articular cartilage. Statistical Tests: Repeatability was assessed using root-mean-squared-CV (RMS-CV). Using Bland-Altman analysis, thresholds defined as the smallest detectable difference (SDD) were determined from the repeatability data with knee repositioning. Results: Without knee repositioning, both surface-averaged T1ρ and T2 were very repeatable on all cartilage surfaces with RMS-CV<1.1%. Repositioning of the knee had the greatest effect on T1ρ of patellar cartilage with the surface-averaged RMS-CV=4.8%. While T1ρ showed the greatest response to exercise at the patellofemoral cartilage region, the largest changes in T2 were determined in the lateral femorotibial region. Following thresholding, significant (> SDD) average exercise-induced in T1ρ and T2 of femoral (-8.0% and -5.3%), lateral tibial (-6.9% and -5.9%), medial tibial (+5.8% and +2.9%) and patellar (-7.9% and +2.8%) cartilage were observed. Data Conclusion: Joint loading with a stepping activity resulted in T1ρ and T2 changes above background measurement error.

Key Words: Articular Cartilage; MRI; Quantitative Imaging; Repeatability; Exercise; Relaxation Time

  • Nandan Das, Sergey Alexandrov, Yi Zhou, Katie Gilligan, Róisín M Dwyer, Martin J. Leahy  Nanoscale structure detection and monitoring of tumour growth with optical coherence tomography.” Nanoscale Advances 2020 DOI: doi.org/10.1039/D0NA00371A
    Download the pdf here.
Approximately 90% of cancers have their origins in epithelial tissues and this leads to epithelial thickening, but the ultrastructural changes and underlying architecture is less well known. Depth resolved label free visualization of nanoscale tissue morphology is required to reveal the extent and distribution of ultrastructural changes in underlying tissue, but is difficult to achieve with existing imaging modalities. We developed a nanosensitive optical coherence tomography (nsOCT) approach to provide such imaging based on dominant axial structure with a few nanometre detection accuracy. nsOCT maps the distribution of axial structural sizes an order of magnitude smaller than the axial resolution of the system. We validated nsOCT methodology by detecting synthetic axial structure via numerical simulations. Subsequently, we validated the nsOCT technique experimentally by detecting known structures from a commercially fabricated sample. nsOCT reveals scaling with different depth of dominant submicron structural changes associated with carcinoma which may inform the origins of the disease, its progression and improve diagnosis.

What are you aiming to find out?

Proposed research demonstrated detection of depth resolved submicron structure with few nanometre accuracies and quantified its alteration as tumour growth in mammary fat pad (MFP).

Why does this research need to be done?

Early detection of cancer can save millions of lives. It is known that tissue goes through submicron structural changes as cancer development initiates. However, developed technique rely on labelling tissue sample and information available from superficial region only.

Describe the methods chosen.

Here we have applied nano sensitive optical coherence tomography (nsOCT) which can detect depth resolved submicron structure and its alteration. Proposed method validated numerically and tested on mammary fat pad (MFP) in mice.

How will this benefit citizens/patients?

Clinical system can be develop based on proposed method to detect cancer in early stage. In initial study, we found a consistent change of submicron structure over tissue depth as tumour developed. Early detection of cancer can provide opportunity for better treatment and can improve quality of life.

What are the next steps?

We are presently developing nsOCT which expected to detect smaller submicron structure and can provide more insight about nano-sensitive changes as precancer progress.

  • Cerine Lal, Sergey Alexandrov, Sweta Rani, Yi Zhou, Thomas Ritter, and Martin Leahy Nanosensitive optical coherence tomography to assess wound healing within the cornea.” Biomedical Optics Express 2020 DOI: 10.1364/BOE.389342
    Download the pdf here.
Optical coherence tomography (OCT) is a non-invasive depth resolved optical imaging modality, that enables high resolution, cross-sectional imaging in biological tissues and materials at clinically relevant depths. Though OCT offers high resolution imaging, the best ultra-high-resolution OCT systems are limited to imaging structural changes with a resolution of one micron on a single B-scan within very limited depth. Nanosensitive OCT (nsOCT) is a recently developed technique that is capable of providing enhanced sensitivity of OCT to structural changes. Improving the sensitivity of OCT to detect structural changes at the nanoscale level, to a depth typical for conventional OCT, could potentially improve the diagnostic capability of OCT in medical applications. In this paper, we demonstrate the capability of nsOCT to detect structural changes deep in the rat cornea following superficial corneal injury.
What was the aim of this study?

The aim of the study is to demonstrate the capability of nsOCT to detect structural changes deep in the cornea following superficial corneal injury and subsequent healing.

Why is this important?

Studying nanoscale structural and dynamic changes in vivo is fundamental to understanding changes occurring at cellular level before the changes manifest at the tissue level. Detecting these submicron structural changes can help scientists and clinicians to diagnose the onset of a disease, its progression and in determining treatment effectiveness of drugs.

The cornea is the transparent, avascular layer of the eye that controls the entry of light into the eye and helps to refract the light onto the retina. Corneal transparency is vital to preserve its structure and function. Corneal injuries generally arise from thermal and chemical burns. Of these, 11.5–22% of all ocular injuries occur from chemical burns, from both acids and alkali. Among chemical induced corneal burns, alkali burn causes more damage to the corneal stroma and anterior chamber compared to acid injury. Alkali ions being lipophilic, penetrate into the corneal stroma disrupting the cells and denaturing the collagen matrix, which promotes further penetration into the anterior chamber.

Hence, it is imperative to understand the nanoscale structural changes occurring during ocular injury and subsequent wound healing process in vivo for assessment of wound repair and monitoring treatment efficacy.

What methods did you use?

In this paper, we have elucidated the capability of the nanosensitive OCT (nsOCT) technique to detect structural changes within the cornea to assess the impact of alkali injury and also to study the wound healing process. nsOCT offers much higher sensitivity to structural changes within the cornea compared to conventional OCT processing. The study reveals that nsOCT is able to detect structural changes with nanoscale sensitivity between healthy cornea, injured cornea and also during the reparative phase of the injury at all depths within the cornea with high statistical significance (p < 10−10).

How could this benefit citizens/patients?

The method presented offers potential for in vivo imaging applications especially in clinical imaging where sensitivity to changes in structure is of significance either to detect the onset of a disease or to evaluate the efficacy of treatment which cannot be obtained from conventional OCT processing.

It can be used for real time monitoring of corneal health following ocular trauma or ocular diseases.

What are the next steps?

Future steps will be the application of nsOCT technique to spatial and temporal structural changes within cornea following injury, treatment and healing and changes occurring due to elevated intra ocular pressure. The technique can also be used to detect pathophysiological changes in retina following diabetic retinopathy, glaucoma and other retinal diseases.

Further applications of the technique can be used to study morphological changes in biomedical samples, for example, to image progression of cancerous cells and tumours as they are known to undergo nanoscale structural changes within their vicinity long before the manifestation of the disease.

  • Yi Zhou, Sergey Alexandrov, Andrew Nolan, Nandan Das, Rajib Dey, Martin Leahy Noninvasive detection of nanoscale structural changes in cornea associated with cross‐linking treatment.” Journal of Biophotonics 2020 DOI: 10.1002/jbio.201960234
    Download the pdf here.
Corneal cross‐linking (CXL) using ultraviolet‐A (UVA) irradiation with a riboflavin photosensitizer has grown from an interesting concept to a practical clinical treatment for corneal ectatic diseases globally, such as keratoconus. To characterize the corneal structural changes, existing methods such as X‐ray microscopy, transmission electron microscopy, histology and optical coherence tomography (OCT) have been used. However, these methods have various drawbacks such as invasive detection, the impossibility for in vivo measurement, or limited resolution and sensitivity to structural alterations. Here, we report the application of oversampling nanosensitive OCT for probing the corneal structural alterations. The results indicate that the spatial period increases slightly after 30 minutes riboflavin instillation but decreases significantly after 30 minutes UVA irradiation following the Dresden protocol. The proposed noninvasive method can be implemented using existing OCT systems, without any additional components, for detecting nanoscale changes with the potential to assist diagnostic assessment during CXL treatment, and possibly to be a real‐time monitoring tool in clinics.

What are you aiming to find out?

The proposed method aims to detect nanoscale changes in cornea with the potential to assist diagnostic assessment during cross-linking treatment

Why does this research need to be done?

Corneal ectasia can severely impair vision, especially in the progressive form caused by the inherent structural weakness of the cornea. Keratoconus, the most common form of corneal ectasia affecting nearly 1 in 375 individuals globally, is an ocular disorder. Corneal cross-linking (CXL), was proved to be an effective way in halting the progression of keratoconus, meaning patients can avoid a corneal transplant.

What methods did you use and why?

In this study, we have presented the application of over-sampling nano-sensitive optical coherence tomography (nsOCT), which is proposed to retain the high spatial frequency information in the interference spectra, to probe the structural alterations inside ex vivo bovine cornea during CXL treatment with nanoscale sensitivity. The results suggest that the over-sampling nsOCT can be used to detect nano-sized structural changes valuable for corneal treatment methods.

How could this benefit citizens/ patients?

Due to its fast, non-invasive detecting method and nanoscale sensitivity, this unique technology is potential to be an indicator in diagnostic assessment associated with CXL treatment, and possibly to be a real-time monitoring tool in clinics as a fast way to receive feedback from patient’s tissue.

What are the next steps?

Future work will aim to implement this method for in vivo corneal detections associated with CXL treatment, including monitoring the nanoscale structural variations at different treating steps and also the postoperative assessment.

  • S. Carluccio, et. al  Progenitor Cells Activated by Platelet Lysate in Human Articular Cartilage as a Tool for Future Cartilage Engineering and Reparative Strategies.” Cells 2020 9(4) DOI: https://doi.org/10.3390/cells9041052
    Download the pdf here.
Regenerative strategies for human articular cartilage are still challenging despite the presence of resident progenitor cell population. Today, many efforts in the field of regenerative medicine focus on the use of platelet derivatives due to their ability to reactivate endogenous mechanisms supporting tissue repair. While their use in orthopedics continues, mechanisms of action and efficacy need further characterization. We describe that the platelet lysate (PL) is able to activate chondro-progenitor cells in a terminally differentiated cartilage tissue. Primary cultures of human articular chondrocytes (ACs) and cartilage explants were set up from donor hip joint biopsies and were treated in vitro with PL. PL recruited a chondro-progenitors (CPCs)-enriched population from ex vivo cartilage culture, that showed high proliferation rate, clonogenicity and nestin expression. CPCs were positive for in vitro tri-lineage differentiation and formed hyaline cartilage-like tissue in vivo without hypertrophic fate. Moreover, the secretory profile of CPCs was analyzed, together with their migratory capabilities. Some CPC-features were also induced in PL-treated ACs compared to fetal bovine serum (FBS)-control ACs. PL treatment of human articular cartilage activates a stem cell niche responsive to injury. These facts can improve the PL therapeutic efficacy in cartilage applications.
  • W. Hotham, and F.M.D. Henson “The use of large animals to facilitate the process of MSC going from laboratory to patient-‘bench to bedside’.” Cell Biol Toxicol 2020 DOI: 10.1007/s10565-020-09521-9
    Download the pdf here.

Large animal models have been widely used to facilitate the translation of mesenchymal stem cells (MSC) from the laboratory to patient. MSC, with their multi-potent capacity, have been proposed to have therapeutic benefits in a number of pathological conditions. Laboratory studies allow the investigation of cellular and molecular interactions, while small animal models allow initial ‘proof of concept’ experiments. Large animals (dogs, pigs, sheep, goats and horses) are more similar physiologically and structurally to man. These models have allowed clinically relevant assessments of safety, efficacy and dosing of different MSC sources prior to clinical trials. In this review, we recapitulate the use of large animal models to facilitate the use of MSC to treat myocardial infarction-an example of one large animal model being considered the ‘gold standard’ for research and osteoarthritis-an example of the complexities of using different large animal models in a multifactorial disease. These examples show how large animals can provide a research platform that can be used to evaluate the value of cell-based therapies and facilitate the process of ‘bench to bedside’.

2019

  • Sergey Alexandrov, Paul M. McNamara, Nandan Das, Yi Zhou, Gillian Lynch, Josh Hogan, and Martin Leahy. “Spatial frequency domain correlation mapping optical coherence tomography for nanoscale structural characterization” Applied Physics Letters 115(2) 2019 DOI: https://doi.org/10.1063/1.5110459
    Download the pdf here.

Most of the fundamental pathological processes in living tissues exhibit changes at the nanoscale. Noninvasive, label-free detection of structural changes in biological samples pose a significant challenge to both researchers and healthcare professionals. It is highly desirable to be able to resolve these structural changes, during physiological processes, both spatially and temporally. Modern nanoscopy largely requires labeling, is limited to superficial 2D imaging, and is generally not suitable for in vivo applications. Furthermore, it is becoming increasingly evident that 2D biology often does not translate into the real 3D situation. Here, we present a method, spatial frequency domain correlation mapping optical coherence tomography (sf-cmOCT), for detection of depth resolved nanoscale structural changes noninvasively. Our approach is based on detection and correlation of the depth resolved spectra of axial spatial frequencies of the object which are extremely sensitive to structural alterations. The presented work describes the principles of this approach and demonstrates its feasibility by monitoring internal structural changes within objects, including human skin in vivo. Structural changes can be visualized at each point in the sample in space from a single image or over time using two or more images. These experimental results demonstrate possibilities for the study of nanoscale structural changes, without the need for biomarkers or labels. Thus, sf-cmOCT offers exciting and far-reaching opportunities for early disease diagnosis and treatment response monitoring, as well as a myriad of applications for researchers.

What are you aiming to find out?

Development of the new technologies for visualization of the sub-micron structure with nanoscale sensitivity to structural changes.

Why does this research need to be done?

For both the fundamental study of biological processes and early diagnosis of pathological processes, information about nanoscale tissue structure is crucial. Furthermore, it is becoming increasingly evident that 2D biology often does not translate into the real 3D situation. The research is motivated by current needs of optical science and technology for biomedical and other applications and is devoted to an important fundamental problem: investigation of new possibilities to detect the structural changes within 3D objects without labels by providing nanoscale sensitivity to structural alterations. One of the most appreciated and fast developed techniques for 3D biomedical imaging is optical coherence tomography (OCT), but resolution and sensitivity to structural alterations is typically limited to microscale. Proposed approach permits to dramatically improve sensitivity to structural changes, up to nanoscale, using just single frame.

Describe the methods chosen.

Label free non-contact optical imaging technologies have been developed. The ability to detect nano-scale structural changes has been demonstrated using different phantoms and human skin in vivo. Healthy volunteer was involved in this study.

What is the intended impact and how can others use the research?

According to the STARSTEM project these techniques will be used to detect changes in cell morphologies and extracellular vesicles at the nanometer scale; to investigate the potential of imaging of extracellular vesicle-mediated disease responses; for nanosensitive detection of tissue responses to disease.

What are the next steps?

  • Experiments to perform nanosensitive detection of tissue responses to disease.
  • Further development of these optical technologies.

Related research.

  1. Alexandrov, P. M. McNamara, N. Das, Y. Zhou, G. Lynch, J. Hogan, and M. Leahy “Spatial frequency domain correlation mapping optical coherence tomography for nanoscale structural characterization”. Appl. Phys. Lett. 2019, v.115, N12, 121105. https://doi.org/10.1063/1.5110459.
  2. Alexandrov, N. Das, J. McGrath, P. Owens, C. J. R. Sheppard, F. Boccafoschi, C. Giannini, T. Sibillano, H. Subhash, and M. Leahy. “Label free ultra-sensitive imaging with sub-diffraction spatial resolution”. 21st International Conference on Transparent Optical Networks ICTON, July 9-13, 2019 Angers, France, Invited, IEEE-Xplore Proceedings 2019 Fr.A6.3, pp.1-4. https://ieeexplore.ieee.org/xpl/conhome/1000766/all-proceedings

STARSTEM Conference Publications

2019

  • Sergey Alexandrov, Nandan Das, James McGrath, Peter Owens, Colin J. R. Sheppard, Francesca Boccafoschi, Cinzia Giannini, Teresa Sibillano, Hrebesh Subhash, and Martin Leahy. “Label free ultra-sensitive imaging with sub-diffraction spatial resolution” Paper presented at the 21st International Conference on Transparent Optical Networks, ICTON’2019, Angers, France, 09-13 July.  DOI: https://dx.doi.org/10.1109/ICTON.2019.8840220
    Download the pdf here.

In this paper, we show a new way to break the resolution limit and dramatically improve sensitivity to structural changes. To realize it we developed a novel label free contrast mechanism, based on the spectral encoding of spatial frequency (SESF) approach. The super-resolution SESF (srSESF) microscopy is based on reconstruction of the axial spatial frequency (period) profiles for each image point and comparison of these profiles to form super-resolution image. As a result, the information content of images is dramatically improved in comparison with conventional microscopy. Numerical simulation and experiments demonstrate significant improvement in sensitivity and resolution.

  • Sergey A. Alexandrov, James McGrath, Colin Sheppard, Francesca Boccafoschi, Cinzia Giannini, Teresa Sibillano, Hrebesh Subhash, Josh Hogan, and Martin Leahy. “Ultra-sensitive label free imaging below the resolution limit” Paper presented at the SPIE BiOS 2019 meeting, San Francisco, California, USA, 4 March 2019.  DOI: https://doi.org/10.1117/12.2502479
    Download the pdf here.

Almost all known nanoscopy methods rely upon the contrast created by fluorescent labels attached to the object of interest. This causes limitations on their applicability to in vivo imaging. A new label-free spectral encoding of spatial frequency (SESF) approach to nanoscale probing of three-dimensional structures has been developed. It has been demonstrated that spatial frequencies, encoded with optical wavelengths, can be passed though the optical system independent of the resolution of the imaging system. As a result information about small size structures can be detected even using a low resolution imaging system. Different versions of the SESF imaging have been published [1-7], including a novel contrast mechanism for high resolution imaging [1], real time nano-sensitive imaging [2], reconstruction the axial (along depth) spatial frequency profiles for each point with nano-sensitivity to structural changes [3], and the adaptation of the SESF approach to depth resolving imaging [4,5]. Recently the SESF approach has been applied to break the diffraction limit and dramatically improve resolution [6,7]. Here we present further development of the SESF approach including correlation mapping SESF imaging. Both results of numerical simulation and preliminary experimental results, including biological objects, will be presented. [1] Alexandrov, et.al., Opt. Lett. 36 3323 (2011). [2] Alexandrov, et.al., Opt. Express 20 (8) 9203 (2012). [3] Alexandrov, et.al., Appl. Phys. Let., 101 033702 (2012). [4] Uttam, et.al., Opt. Express, 21, 7488 (2013). [5] Alexandrov, et.al., Nanoscale, 6, 3545 (2014). [6] Alexandrov, et.al., Sci. Rep., 5, doi: 10.1038/srep13274 (2015). [7] Alexandrov, et.al., J. Biophotonics, https://doi.org/10.1002/jbio.201700385 (2018).