Muscle Impedance Signal Processing for biomarker development in Amyotrophic Lateral Sclerosis

Amyotrophic Lateral Sclerosis (ALS) is a progressive neurodegenerative disease that affects motor neurons, leading to muscle weakness and ultimately death. Early and accurate diagnosis of ALS is crucial specifically given the common occurrence of misdiagnosis due to overlapping symptoms with mimic diseases. Electrical Impedance Myography (EIM) is a non-invasive and promising technique for assessing the electrical properties of muscle. In this thesis, multidimensional EIM data were assessed as a potential biomarker for ALS, with a focus on two sections: EIM analysis of the tongue to assess ALS bulbar disease, and EIM analysis of multiple muscles in the limbs to address the contribution of the subcutaneous fat layer and how this effects the biomarker capabilities for the assessment of a range of neurodegenerative disorders.
The first section of the thesis involved EIM analysis of the tongue, where a finite element method (FEM) based model of the tongue was optimised to fit the measurement dataset. This work revealed the underlying differences in the dielectric properties of healthy muscle compared to ALS patient groups of different severities. The model also established that lateral measurements on the edge of the tongue were viable. Additionally, a novel analysis framework, termed tensor EIM, was developed for dimensionality reduction of the multi-dimensional EIM dataset. Tensor EIM was able to characterise the direction of the spectral shift associated with worsening disease, which can be attributed to an increase in the centre frequency. The methodology successfully improved classification and correlation with the bulbar disease burden. Additionally, tensor EIM was found to be more sensitive to longitudinal disease changes than other biomarkers. Furthermore, complex multi-directional spectral shifts were identified throughout progression, where preliminary findings suggest these differences may pertain to acute and chronic denervation.
The second section addressed the challenge of subcutaneous fat contribution to limb EIM data. Initial assessment of the dataset found that the subcutaneous fat thickness had a significant correlation with the EIM data as well as with the disease severity, suggesting a causal relationship between EIM and the disease state. Through a sophisticated optimisation algorithm, an FEM model of the human limb was refined and demonstrated clear agreement with measured data. With use of the optimised FEM model, a methodology was developed to separate the signal of the skin-subcutaneous fat layer from the muscle in limb EIM measurement data. Following signal separation the dominance of the subcutaneous fat was considerably re- reduced, and the biomarker performance of the data improved in some areas. Furthermore, the tensor EIM framework was used to identify the spectral pattern associated with neurodegenerative disease, with some preliminary findings indicating differences in the spectral pattern for myopathic compared to neurogenic disease.
Overall, this thesis provides valuable insights into the use of multidimensional EIM data as a biomarker for ALS. The findings from the tongue and limb EIM analysis, combined with the development of optimised FEM models and a signal separation procedure, contribute to the understanding of the diagnostic and prognostic potential of EIM in ALS. These results have implications for clinical practice and may aid in early and accurate diagnosis of ALS, as well as improving clinical trial design, enabling the development of more effective therapies.