The main objective of this thesis is the development of personalised diastolic cardiac mechanics models for the study of dilated cardiomyopathy. A model personalisation pipeline is developed, which deals with two core challenges in patient-specic modelling, namely data integration and parameter identiability. Important modelling aspects of the pipeline are selected based on the potential and limitations of the data at hand, following systematic investigation using in silico and in vivo tests. To assist in the model development process, numerical schemes for dealing with incompressibility/ near incompresibility in the heart are compared in terms of accuracy and eciency. In silico testing is extended to a parameter identiability study, where commonly used passive constitutive laws are compared in terms of identiability and model delity, using synthetic 3D tagged MRI. The model personalisation pipeline is developed based on these modelling considerations, with a focus on optimising the use of the available data to improve model delity. The proposed pipeline is tested on a group of volunteers and patients, enabling an assessment of modelling attributes which can improve model accuracy and parameter identiability in vivo. Finally, the developed patient-specic models are employed for a comparative analysis of diastolic heart function, in patients with dilated cardiomyopathy and healthy volunteers.