Intelligent learning systems for hotspots detection in ISFET arrays

Electrochemical sensors have already started to revolutionise healthcare by applying state-of-the-art Lab-on- Chip technology to a variety of medical applications. The ion-sensitive field-effect transistor (ISFET) offers significant advantages due to their compatibility with large-scale Complementary Metal-Oxide-Semiconductor (CMOS) technology, allowing for a robust, low cost and scalable solution [1]. ISFETs are transistors that are capable of measuring the ion concentration in a solution [2]. This feature can be exploited to achieve DNA hybridisation of a specific DNA or RNA sequence. This has major applications in the detection of specific viruses and can have a huge impact in clinical applications [3] [4]. ISFET arrays are able to perform electrochemical imaging, providing a large amount of rich data at a fast frame rate which raises several opportunities and challenges for signal processing.

For example, ISFET arrays are used for rapid and accurate detection of the Coronavirus SARS- COV-2 at the point of care, resulting in a low-cost and portable diagnosis method [5]. This is essential during the current pandemic and would have a huge impact to allow the UK to return to standard activities while tracing people that contracted the virus. At the the Centre for Bio-inspired Technology, this is achieved by employing an array of thousand of sensors whose data is processed on-chip and off-chip.

However, the state of art of the ISFET technology is far from being optimal. There are challenges associated with this method that arise as a consequence of the non-idealities of the ISFET sensors: for instance, ISFETS are highly noisy and often introduce drift signals and trapped charge in the output signal that are not easily distinguishable from the desired data [6]. These are normally tackled by averaging and other techniques that reduce the spatial and temporal resolution of the samples.

The aim of the proposed research is to develop methods to improve diagnostic and patient outcomes of the ISFET arrays. This involves on-chip data processing to reduce the output signal non-idealities and off-chip creation of algorithms to reconstruct the significant features of the signal. These intelligent systems will explore different paradigms in signal processing and machine learning, alongside tools such as mathematical optimisation and graph signal processing.

The proposed research has the potential to be impactful beyond the specific ISFET application. I believe that under the supervision of Dr. Pantelis Georgiou, whose research focus matches the aforementioned challenges, there is a potential to conduct impactful research across many applications.

References
[1] Toumazou, Christofer, et al., Simultaneous DNA amplification and detection using a pH-sensing semicon- ductor system, Nature methods 10.7 (2013): 641
[2] N. Moser, T. S. Lande, C. Toumazou and P. Georgiou, ISFETs in CMOS and Emergent Trends in Instrumentation: A Review, in IEEE Sensors Journal, vol. 16, no. 17, pp. 6496-6514, Sept.1, 2016, doi: 10.1109/JSEN.2016.2585920.
[3] Rodriguez-Manzano, J., Moser, N., Malpartida-Cardenas, K. et al. Rapid Detection of Mobilized Colistin Resistance using a Nucleic Acid Based Lab-on-a-Chip Diagnostic System. Sci Rep 10, 8448 (2020). https://doi.org/10.1038/s41598-020-64612-1
[4] Malpartida-Cardenas K, Miscourides N, Rodriguez-Manzano J, Yu LS, Moser N, Baum J, Georgiou P., Quantitative and rapid Plasmodium falciparum malaria diagnosis and artemisinin-resistance de- tection using a CMOS Lab-on-Chip platform. Biosens Bioelectron. 2019 Dec 1;145:111678. doi: 10.1016/j.bios.2019.111678. Epub 2019 Sep 7. PMID: 31541787; PMCID: PMC7224984.
[5] Jesus Rodriguez-Manzano, Kenny Malpartida-Cardenas, Nicolas Moser, Ivana Pennisi, Matthew Cavuto, Luca Miglietta, Ahmad Moniri, Rebecca Penn, Giovanni Satta, Paul Randell, Frances Davies, Frances Bolt, Wendy Barclay, Alison Holmes, Pante