Abnormality Detection Inside Blood Vessels With Mobile Nanomachines

Abstract

Motivated by the numerous healthcare applications of molecular communication within Internet of Bio-Nano Things (IoBNT), this work addresses the problem of abnormality detection in a blood vessel using multiple biological embedded computing devices called cooperative biological nanomachines (CNs), and a common receiver called the fusion center (FC). Due to blood flow inside a vessel, each CN and the FC are assumed to be mobile. In this work, each of the CNs perform abnormality detection with certain probabilities of detection and false alarm by counting the number of molecules received from a source, e.g., infected tissue. These CNs subsequently report their local decisions to a FC over a diffusion-advection blood flow channel using different types of molecules in the presence of inter-symbol interference, multi-source interference, and counting errors. Due to limited computational capability at the FC, OR and AND logic based fusion rules are employed to make the final decision after obtaining each local decision based on the optimal likelihood ratio test. For the aforementioned system, probabilities of detection and false alarm at the FC are derived for OR and AND fusion rules. Finally, simulation results are presented to validate the derived analytical results, which provide important insights.

Sequential Decision Fusion for Abnormality Detection via Diffusive Molecular Communications

“…While some of the existing works have been focusing on a specific type of abnormality and sensing model (e.g. [4], [5]), others, such as [2] and [6], have employed abstract sensing models in order to achieve more general results. In this work, we chose to focus on the latter approach for the sake of generality.…”

“…(9) in the SPRT described in eqs. ( 10) and ( 11), resulting in a test which we will refer to as the SAPRT, where, in contrast to the usual LLR based SPRT model, the running ALLR functionΛ Y (l) of the observations cannot be expressed as a running sum of the log likelihood ratios of the individual samples due to the averaging operation over the pmf of the nuisance parameter X performed in (6), despite the fact that Y n is an independent (albeit not i.i.d.) sequence.…”

“…The work in [5] focuses on a similar type of abnormality as in [4], using a more realistic channel model based on the solution of the diffusion-reaction equations in an unbounded medium, and provides a sub-optimal decision fusion (DF) strategy. In [6], hard decisions are employed at the sensors, and a suboptimal DF approach based on logic OR and AND operations. Both [5] and [6] assume perfectly orthogonal sensor-to-FC communication channels.…”

“…In [6], hard decisions are employed at the sensors, and a suboptimal DF approach based on logic OR and AND operations. Both [5] and [6] assume perfectly orthogonal sensor-to-FC communication channels. This requires a receiver that can provide as many molecule receptor types as there are sensors, which may become impractical for applications involving a large number of sensors.…”

Cooperative Molecular Communication in Drift-Induced Diffusive Cylindrical Channel

“…For example, cooperative abnormality detection where several sensors are used to identify malignancy in a three-dimensional unbounded environment was presented in [15]. In [17], abnormality detection in a one-dimensional channel was proposed. Early detection of cancerous cells using mobile nanosensors (MNS) was considered in [18].…”