Influencing factors are important considerations in the application of non-invasive thermal diagnostics for the early detection of breast tumour. In this paper, experimental studies of artificial tumours embedded inside silicone breasts coupled with numerical simulations using 3D finite-element method in ANSYS were used to investigate the effects of tissue conductivity, tumour size and tumour depth on the heat patterns at the breast skin surface. After validating the numerical breast model, the analysis was extended to examine the heat patterns of a growing tumour. The findings revealed that thermal patterns of the breast surface over time could be useful for the detection of tumours. The existence of tumours would be more noticeable from thermal images of breasts with more fatty tissues, and breasts of lower density. The method would be more suitable for the detection of large tumours near the skin surface. The simulated results suggested that it is possible to detect an initial 4 mm HER2-positive tumour at a depth of 44 mm after about 196 days when it had grown to 7 mm using a temperature sensor with resolution of 0.01 ℃.