Texture features are designed to quantitatively evaluate patterns of spatial distribution of image pixels for purposes of image analysis and interpretation. Unexplained variations in the texture patterns often lead to misinterpretation and undesirable consequences in medical image analysis. This project aims to explore the ability of Machine Learning (ML) methods to design a radiology test of a bone pathology such as Osteoarthritis (OA) at an early stage when the number of patients' cases is small. Experiments will be run on high-resolution X-ray images of knees in patients who were identified with Kellgren-Lawrence scores progressing from 1, the lowest recognisable score. The existing ML methods have provided a limited diagnostic accuracy, whilst the project aims to explore the Group Method of Data Handling (GMDH) strategy of Deep Learning shown to be capable of significantly extending the diagnostic test accuracy. The project aims to compare Machine Learning solutions on real-scale data to demonstrate that a new GMDH framework of learning the texture features will significantly improve the diagnostic test accuracy. This allows radiologists to design a decision model for early diagnosis of bone pathology and provide more accurate radiology tests verified on a sufficient number of patients' cases.
Research questions: (1) to explore the ability of a designed GMDH strategy of Deep Learning to extend the early detection of bone pathology (2) to explore ways of designing activation functions within the GMDH-type Deep Learning framework.
The deadlines are as follows:
For March starters:
International applicants - 30th November 2021
UK nationals - 18th January 2022
For October starters:
International applicants - 30th June 2022
UK nationals - 5th August 2022