FindA University Ltd Featured PhD Programmes
University of Bristol Featured PhD Programmes
FindA University Ltd Featured PhD Programmes

DETECTION AND ANALYSIS OF METASTATIC AND PRIMARY ORAL CANCERS AND STROMAL FEATURES WITH ADVANCED DEEP LEARNING.


   School of Clinical Dentistry

This project is no longer listed on FindAPhD.com and may not be available.

Click here to search FindAPhD.com for PhD studentship opportunities
  Dr Syed Ali Khurram  No more applications being accepted  Funded PhD Project (UK Students Only)

Sheffield United Kingdom Cancer Biology Data Analysis Pathology Software Engineering

About the Project

Supervisor(s): Dr Ali Khurram (School of Clinical Dentistry, Sheffield), Dr Pavitra Krishnaswamy (I2R, A* Star, Singapore).

Oral squamous cell carcinoma (OSCC) or oral/mouth cancer is amongst the top ten most common cancers in the world with increasing numbers and a poor prognosis. The prognosis is worsened when OSCC metastasis to the regional lymph nodes (LN) (<50% five-year survival). Metastatic OSCC (mOSCC) diagnosis can be tedious and challenging particularly for small/micrometastases. 25-30% of mOSCC can infiltrate out of a LN resulting in extranodal extension (ENE) reducing the survival to 20%. Diagnosis of ENE is subjective and only possible on resection specimens as existing imaging/radiological modalities lack the required accuracy. Better understanding of the metastatic and ENE process and identification of novel markers can facilitate improved detection and treatment.

Primary OSCC (pOSCC) cross-talks with the surrounding stromal cells, which can promote tumour growth, invasion and metastasis. However, whether mOSCC-associated stroma in LNs plays a role in metastasis growth or ENE development has never been investigated. Our preliminary data of matched primary and metastatic tumours shows that ENE +ve nodes have more abundant stroma, increased number of myofibroblasts and blood vessels compared to ENE-ve nodes suggesting an important role for tumour microenvironment (TME) in ENE.

Deep learning algorithms have been shown to be important in analysing ‘big data’ from whole slide images (WSI) to facilitate automated detection of different types of cells as well as their digital footprint. The CAMELYON2016 challenge showed that deep learning models performed better than experienced pathologists at detecting metastatic breast cancer deposits in LN. However, to our knowledge deep learning has yet to be used for mOSCC and ENE detection and analysis.

Automated prediction and detection of mOSCC and ENE on tissue sections could significantly reduce diagnostic turnaround times and improve accuracy to aid patient treatment and improve outcomes.

Aims

To investigate the use of deep learning-based approaches in automated detection of mOSCC and ENE and compare tumour and stromal features between primary and metastatic tumours.

Methods

A cohort of archived OSCC cases will be identified using the local database. Metastatic OSCC tissue sections as well as the matched primary tumours (from tongue and floor of mouth, the two most common pOSCC sites) will be scanned to obtain WSI in addition to non-metastatic pOSCC. A subset of these cases will serve as the training set and exhaustively annotated to aid programming of deep learning algorithms. The developed models will be applied to the test cohort to determine accuracy. Correlation with other clinico-pathological variables such as primary tumour size, perineural/vascular/lymphatic invasion, bone invasion, age, gender, recurrence and survival will also be carried out. Furthermore, analysis of stromal and epithelial mesenchymal transition (EMT) markers will also be undertaken to establish whether these features can be detected on routine H&E images as well as correlation with radiological images, prognosis and whether digital features from pOSCC can be used to predict risk of metastasis and ENE.

The findings from this study will be completely novel with a potential to make a significant impact on diagnostic practice and aid patient treatments in future.

The project has a duration of 4 years with 2 years spent at A*STAR and 2 years at University of Sheffield. The latter will include data analysis and write up to ensure submission within the allocated 48 months. The 2 years in each site are likely to be alternating on a yearly basis but may be changed depending upon how the data acquisition and analysis progresses. Research in Sheffield will involve scanning of slides, retrieval of demographic data as well as one-on-one training in digital pathology and histological analysis of normal, cancerous and metastatic tissue slides. The student will also be involved in retrieval and analysis of clinical images, clinical metadata and follow-up data as well as statistical analysis, presentation of results at relevant computer science/computational pathology/deep learning conferences as well pathology/radiology/oncology and biomedical meetings. Research in Singapore will involve development of deep learning algorithms. Statistical correlation to clinicopathological and primary tumour variables as well as radiological imaging will also be performed at Singapore in liaison with Sheffield.

Entry Requirements:

• Only UK students are eligible for this funded project.

• 2:1 or above honours degree and possession of a PhD/Masters or equivalent in Computer Science or a closely related discipline.

• Experience of research (or interest in) in one or more of the following: deep learning; big data management; computational pathology; medical imaging; computer vision; and machine learning.

• Ability to initiate, plan, organise, implement and deliver programmes of work to tight deadlines.

• Ability to work with people from different backgrounds and in team across academic, healthcare and industry, with good interpersonal skills.

• Good written and oral communication skills, with the ability to translate complex information into accessible language.

• Sufficient breadth or depth of specialist knowledge in the discipline and of research methods and techniques to work independently and develop relevant research.

• Ability to work independently or as part of a multi-disciplinary team on research programmes.

• Ability or potential to contribute to the development of funding proposals and publications in order to generate external funding to support research projects.

Enquiries:

Interested candidates should in the first instance contact (Dr Ali Khurram- 01142159378 - [Email Address Removed])

How to apply:

Please complete a University Postgraduate Research Application form available here: www.shef.ac.uk/postgraduate/research/apply

Please clearly state the prospective main supervisor in the respective box and select 'School of Clinical Dentistry' as the department.


Funding Notes

Fully funded by the A*Star-University of Sheffield Research Attachment PhD Programme - further details below.

References

Proposed start date: 1st October 2021
For the selected student admitted to the 4-year programme, A*Star will provide the following financial support, whilst the student is in Singapore:
● Living allowance: A monthly stipend of two thousand, five hundred Singapore Dollars (~£1,050) whilst in Singapore.
● A one-off "settling-in allowance" of one thousand Singapore dollars (~£420).
● A one-time airfare allowance of one thousand five hundred Singapore dollars (~£630).
● Consumables and Bench Fees incurred by students when based at A*Star in Singapore.
● Cost of medical insurance while the student is based at A*Star.
Whilst in Sheffield, students receive fees and stipend (at the RCUK rate, £4,500 & £15,609 in 2020/21). In addition, students will be able to claim up to £500 from Sheffield towards the costs of an airfare back to the UK whilst they are in Singapore in order to make a home visit. This will normally be available for students who meet the normal expectations of spending approximately half of the programme in Singapore. This allowance is not available for those NOT spending 50% of the programme time in Singapore
Search Suggestions
Search suggestions

Based on your current searches we recommend the following search filters.