Biosensing by Learning: Cancer Detection as Iterative optimization

We propose a novel cancer detection procedure (CDP) based on an iterative optimization method. The global minimum of a tumor-induced biological cost function indicates the tumor location, the domain of the cost function is the tissue region at high risk of malignancy, and the time-variant guess input is a swarm of externally controllable and trackable nanorobots for tumor sensing. We consider the spatial distrib-ution of fibrin as the cost function; the fibrin is formed during the coagulation cascade activated by tumor-targeted signalling modules (nanoparticles) and recruits clot-targeted receiving modules (nanorobots) towards the site of disease. Subsequently, the CDP can be interpreted from the iterative optimization perspective: the guess input (i.e., a swarm of nanorobots) is continuously updated according to the gradient of the cost function in order to find the optimum (i.e., cancer) by moving through the domain (i.e., tissue under screening). Along this line of thought, we consider the gradient descent (GD) iterative method, and propose the GD-inspired CDP, which takes into account the realistic in vivo propagation scenario of nanorobots. Finally, we present numerical examples to demonstrate the features of the GD-inspired CDP.