top of page
In order to achieve the successful complementation of LIBRA objectives, the work-plan is wisely organized and split into the following distinct work packages (WPs):
-
Objective 1: Develop miniaturized photonic integrated circuits (PICs) operating as Raman spectroscopy sensing module and Refractive index sensing moduleLIBRA will develop two types of PICs, a novel active PIC that will be the core technology of LIBRA system and versatile passive PICs that will couple to the active, will be disposable, and can be adjusted depending on the sensing needs, thus enabling a modular system. The active PIC will be used for realizing two sensing modalities, Raman spectroscopy and Refractometry. The first will be accomplished by means of an external cavity laser (ECL) and a high-resolution spectrometer (AWG structure), while the latter by means of a tunable VCSEL diode and PDs. All the components will be integrated on the same active PIC. The passive PICs will come in two types, in PICs integrating array of 8 aMZIs that when coupled to the active PIC will employ the tunable VCSEL and PDs and will realize aMZI refractometry for the detection of pathogens. The careful design of the structures and the utilization of a clever FFT-based algorithm will improve the limit of detection (LoD) and bring it down to ~10-8 RIU, while the aMZI array will allow the simultaneous detection of multiple pathogens. The second type of passive PICs consist of spiral structures that when coupled to the active PIC will employ the ECL and AWG spectrometer for Raman spectroscopy.
-
Objective 2: Develop a disposable microfluidics and pre-treatment module.The passive PICs will be enclosed in their corresponding microfluidic modules which will be configured according to the needs of each sensing module and will be coupled to the LIBRA fluidic stream. The Raman microfluidic module will drive the sample onto the spiral waveguides after filtering, towards the screening of nutrients, via Raman spectroscopy. The aMZI microfluidic module will include parts necessary for the preconcentration of the pathogens in the bioreactor, apart from filtering. Considering the large amount of liquid volume in a bioreactor (especially for the photonic bioreactor case scenario), as well as the need for early detection, it is necessary to include a preconcentration unit within the disposable microfluidic module, which will ensure the delivery of pathogens onto the aMZI sensors. To this aim, as the sample from the bioreactor enters the microfluidic module, it will be exposed to functionalized magnetic beads which will bind to the pathogens with great accuracy. Then the magnetic beads along with the pathogens, will be pre-concentrated with the use of a permanent magnet. The part of the microfluidic module that initiates the binding of the pathogens with the magnetic beads and their preconcentration, constitutes the pretreatment module. Furthermore, the LIBRA microfluidic modules will be disposable, thus maintaining sterility by using a closed system and disposable components, while enabling the modular approach with interchangeable components as those modules can be modified according to the end user’s needs.
-
Objective 3: To design and develop advanced data analysis, management and presentationLIBRA system will have embedded algorithms for the analysis of the data collected by each sensing module at each sensing event. To this aim, a data analysis pipeline will be developed, consisting of the data collection in both pre and post processing sensor data. The pre-processed data being in binary or other form of raw data. The collected data - referred as dataset – will be further organized into time-series. The dataset will be processed by: 1) performing an Exploratory Data Analysis (EDA) to identify the data distribution; 2) using the EDA report to remove outlier data points from the dataset; 3) identifying feature importance and correlation between features in the dataset; 4) perform steps to reduce the dataset dimensionality and avoid the feature multicollinearity issue; 5) perform feature engineering and data augmentation, removing data bias and noise, improving the overall prediction accuracy of AI models.
-
Objective 4: To provide efficient and reliable AI techniques for contamination predictionThe LIBRA system will become a smart multisensing platform. The multiple sensors (Raman sensing module, aMZI sensing module with multiple sensing elements) present in the system will offer large amounts of sensory data that will help the consortium to create an AI-model for smart contamination prediction models. In order to include a multitude of different pathogens within the AI-model, multiple machine learning and optimization algorithms will be coupled. Neuroevolution will be used as a primary technology of choice as it is a combination of deep neural networks and evolutionary algorithms, enabling: 1) large datasets to be processed and models created within minutes; 2) analyze and train models originated from different data formats, for example, quantitative and binary wavelength data together; 3) the created models can be upgraded and calibrated without the need to retrain from the beginning with the original data; 4) the Neuroevolution allows explainable AI features, allowing the model users to understand the reasoning behind each predictions and explaining the AI calculations in a natural language format - avoiding the need of statistics personnel to analyze and explain the data. Furthermore, the Neuroevolution is able to solve the challenge of creating insights when dealing with multiple data formats, in this context, nutrient quantification and pathogen data.
-
Objective 5: Integrate the individual modules into LIBRA advanced multi-sensing systemLIBRA will deliver a smart multi-sensing system that combines two photonic technologies in a modular, robust, cost-effective, fast and easy-to-use sensing system (TRL 5) that can be integrated to various types of bioreactors according to the desired application. The long-term use of the active photonic platform ensures the reduction of the manufacturing costs while the use of interchangeable disposable components, such as the Raman and aMZI passive chips, ensure the adjustment of the final system according to the needs as specified by the end users. Furthermore, to achieve the straightforward coupling of the disposable modules with the active PIC chip, novel passive and active alignment methods will be developed. Two types of microfluidic modules will be developed, according to the type of treatment that sample need for each sensing technique (Raman spectroscopy, refractometry). For the detection of pathogens, bio-functionalization technologies, will be incorporated, taking advantage of the binding affinity of antibodies to the pathogens. According to the application, LIBRA system will be prepared to be attached at the outer cell of the bioreactor and thus the focus of the instrument design will be not only the implementation of all interfaces to the cartridge types but also robustness, easy handling and minimal maintenance so as to be adapted to bioreactors that exist in indoor and outdoor environment. A user-friendly interface will be available (a touch screen on the device, an application on a smartphone or a PC) for the execution of the measurements and the basic handling of the data.
-
Objective 6: To demonstrate and validate a reliable and fast modular system for the detection of pathogens and nutrients in bioreactors for pharma industry and energy production industry.LIBRA will combine two key photonic sensing techniques that are able to provide complementary information whose combination leads to a comprehensive screening of the metabolic rate of a cell culture in a bioreactor. The Raman module will constantly provide information on nutrient use of the cell culture allowing for better control of addition of culture media or specific supplements when required. The aMZI module will provide additional information by providing early detection of micro-organisms inside the cell culture. LIBRA system will be validated in industrially relevant environment (SCE, AU, CNR) with real samples (including required matrix) for the detection of nutrients and various micro-organisms such as Bacillus cereusÌ´, bacillus fusiformis, pseudomonas protegens and mycoplasma, for the two proposed case scenarios. The results acquired with the LIBRA system will be compared to standard methods for biosafety testing, in compliance with global pharmacopoeia standards, according to the criteria formed by the US Food and Drug Administration (FDA) and the European Medicines Agency (EMA).
-
Objective 7. To evaluate the increased yield and reduced waste throughout the value chain for end-products and to achieve at least 50% reduced energy consumption throughout the whole bioreactor production process.LIBRA will employ Life Cycle Analysis (LCA) methodology to assess the environmental benefits of the PIC based multi-sensing technologies developed within the project. The impact assessment will first allow to calculate the waste reduction indicator and the energy reduction indicator, with a statistic uncertainty evaluation for each result. In addition, a carbon footprint of this technology will be performed, benchmarked against the reference bioreactor technologies. This carbon footprint will be calculated over 1 kg of produced bioproduct. LIBRA’s on-line sensing technology will enable early diagnosis of pathogens and nutrients, which can be detrimental to the culturing process. For the mammalian cell use case, early detection of contamination of out-of-spec CQAs can save overall power consumption by eliminating unnecessary cleanroom uptime. For the PMs use case, the accurate monitoring of parameters such as nutrients and other metabolites is a crucial requirement for cell culture processes in bioreactors to achieve high product yields with consistent product quality. These parameters can be kept close to the optimum, relying on the real-time sensors proposed here. Upon detection, the culture process can be stopped or adapted to prevent use of unnecessary resources and energy.
bottom of page