|According to World Health Organization, 13.1 million people will die in the world just because of cancer by 2030. Early tumor detection is very crucial to saving the world from this alarming mortality rate. However, it is an insurmountable challenge for the existing medical imaging techniques with limited imaging resolution to detect microscopic tumors. Hence, the need of the hour is to explore novel cross-disciplinary strategies to solve this problem. The rise of nanotechnologies provides a strong belief to solve complex medical problems such as early tumor detection. Nanoparticles with sizes ranging between 1-100 nanometers can be used as contrast agents. Their small sizes enable them to leak out of blood vessels and accumulate within tumors. Moreover, their chemical, optical, magnetic and electronic properties also change at nanoscale, which make them an ideal probing agent to spatially highlight the tumor site. Though, using nanoparticles to target malignant tumors is a promising concept, only 0.7% of the injected nanoparticles reach the tumor according to the statistical results of last 10 years.
In PhD work, we proposed novel in vivo computational frameworks for fast, accurate and robust nanobiosensing. Specifically, the peritumoral region corresponds to the “objective function”; the tumor is the “global optimum”; the region of interest is the “domain” of the objective function; and the nanoswimmers are the “computational agents” (i.e., guesses or optimization variables). First, in externally manipulable in vivo computation, nanoswimmers are used as contrast agents to probe the region of interest. The observable characteristics of these nanoswimmers, under the influence of tumor-induced biological gradients, are utilized by the external tracking system to steer nanoswimmers towards the possible tumor direction. To take it one step ahead, we provide solutions to the real-life constraints of in vivo natural computation such as uniformity of the external steering force and finite life span of the nanoswimmers. To overcome these challenges, we propose a multi-estimate-fusion strategy to obtain a common steering direction for the swarm of nanoswimmers and an iterative memory-driven gradient descent optimization strategy for faster tumor sensitization.
Next, we proposed a parallel framework called autonomous in vivo computation, where the tumor sensitization is highly scalable and tracking-free. We demonstrate that the tumor-triggered biophysical gradients can be leveraged by nanoparticles to collectively move toward the potential tumor hypoxic regions without the aid of any external intervention. Although individual nanoparticles have no target-directed locomotion ability due to limited communication and computation capability, we showed that once passive collaboration is achieved, they can successfully avoid obstacles and detect the tumor.
Finally, to address the respective limitations of externally manipulable and autonomous settings such as constant monitoring and slow detection, we proposed a semi-autonomous in vivo computational framework. We showed that the spot sampling strategy for an autonomous swarm of nanoswimmers can achieve faster tumor sensitization in complex environments. This approach makes the swarm highly scalable along with giving it the freedom from constant monitoring.
The performance of the aforementioned tumor sensitization frameworks is evaluated through comprehensive in silico experiments that mimic the realistic targeting processes in externally manipulable, self-regulatable and semi-autonomous settings. The efficacies of the proposed frameworks are demonstrated through numerical simulations that incorporate various physical constraints with respect to controlling and steering of computational agents, their motion in discretized vascular networks and their motion under the influence of disturbance and noise.