The use of Design of Experiments for steady-state and transient inverse melanoma detection problems

Abstract

Melanoma is one of the most fatal skin cancers; for this reason, there is a need for the development of new safe, non-invasive and efficient diagnostic techniques. Dynamic thermography is showing to be a promising technique for the early detection of skin cancers. Therefore, this paper investigates two different inverse bioheat problems using steady-state and transient skin temperature measurements. Both problems are investigated numerically to estimate how accurate blood perfusion rate, metabolic heat generation, diameter and thickness of the tumour can be estimated simultaneously under exact and noisy measurement data, based on a complex numerical model describing multilayer tissue. The inverse problems have been tested using different melanoma size, Clark II and Clark IV. The Design of Experiments (DOE) technique has been used to solve and analyse the inverse problems. A substantial number of numerical model evaluations, totalling 2,306,486 simulations, had to be undertaken as part of the full factorial DOE. The results show that it is always possible to obtain tumour parameters using exact static or dynamic measurement data. However, for noisy temperature data, the use of a dynamic approach showed an advantage over the steady-state one, which failed because of the very small temperature differences between the healthy skin and the tumour. The dynamic thermography can retrieve blood perfusion rate, thickness and diameter of the tumour as well as the metabolic heat generation despite the low sensitivity for low and high levels of measurement error; however, to detect melanoma lesions at an early stage, the measurement and model errors should be kept as low as possible.