Want to know more?

Projects

If you think you need our support for innovative activities of this caliber, contact us to discuss your needs with us

The work methodology and algorithms used can certainly be utilized and replicated in different contexts: from industry to civil engineering, from production to security.

In particular, all areas - not just medical ones - where it is necessary to retrieve information from image acquisition, lend themselves to the implementation of algorithms and projects that use the same technologies and skills, even if for different purposes.

Algorithms for the reconstruction of the breast geometry from medical imaging

The need was to improve the output of the detections of the new device by exploiting advanced algorithms to generate a result as representative of reality as possible.

The device that is the subject of our work bases its operation on mathematical models capable of identifying tissue properties from a set of optical and infrared detections. Deix played a crucial role in developing an algorithm capable of improving the reconstruction of breast geometry from the acquired images: the biggest challenge was to **identify the region of interest** for the analysis and **reconstruct its 3D volume**, eliminating the inherent noise of the acquisitions and considering the criticalities given by heterogeneous detections.

Week after week, the **joint R&D work** with the client has allowed for an increase in the accuracy of breast reconstruction, enabling the machine to provide increasingly effective support for the doctor's analysis.

The project has therefore achieved the agreed-upon objective, **improving the output** generated by acquisitions made by the device. One of the most interesting aspects was certainly working in incremental phases, refining the tool step by step, thanks to continuous feedback and experimentation. This is a typical approach for algorithm development projects, which proceeds through incremental testing in order to achieve the best possible result.

In the field of medical prevention, there are increasingly more technologies being developed to make analyses more effective and less invasive. One of the biggest challenges is that of **ensuring the accuracy** of the results so that they can support medical personnel in the diagnostic phase.

A biomedical equipment company has requested the support of Deix to improve the technology behind a device that they are developing, which is capable of providing support to medical personnel in diagnosing breast cancer

The algorithm is able to accurately identify the **region of interest** for analysis, through a **segmentation** process that jointly uses 2D images and 3D acquisitions.

The use of both sources of data allows to overcome individual limitations of each (noise and imprecision of 3D data and poorly illuminated areas in 2D images) by **reconstructing the 3D surface** of the region of interest.

Techniques of image rasterization were used to **map the information** derived from the 2D image to the 3D model.

Deix has developed a robust algorithm that leverages various techniques of image analysis and computational geometry to combine acquired information and generate output.

If you think you need a solution of this type for your company, please contact us to discuss your scenario together. We are ready to support you in your most ambitious projects.

The application represents a truely tailored solution, based on the specificity of the customer and their needs, starting from different and highly varied datasets.

Customization

The methodology used for the project allows for replicating and scaling the solution, adapting it to specific contexts. Whether it is about extending it to additional processes or expanding the scope to other production facilities, it is always possible to work by modulating the application to the context.

Replicability

The project allows to work with incremental releases, improving the application from every point of view over time. This enables faster data integration, continuous platform improvement thanks to constant feedback, the ability for the software to learn from the interaction between the application and human input, and the possibility to integrate artificial intelligence and machine learning tools.

Optimization

Data-driven management software for production plants

The data generated and collected daily by the machinery were underutilized for **decision-making and optimization processes**. Until the beginning of the project, they were mainly used to assess the situation in the event of a production failure or defect and to analyze the problem after the fact.

The context of the project posed an additional initial challenge: the need to integrate multiple heterogeneous data sources to arrive at a centralized and organized system of information on which to make strategic decisions with a **data-driven approach**.

Thanks to the integration and normalization of data, merged into a single ad hoc application, staff has been able to fully exploit the potential of the available measurements. The time needed to perform an analysis has gone from several hours to just a few minutes.

Simplification

In addition to providing a better overview of what is happening in the production plant, the solution has given operators the ability to highlight potential problems in advance, thanks to automatic alerting tools.

Real-time monitoring

Mapping the most common intervention cases, optimized over time, has made it possible to anticipate, predict and investigate issues in advance and optimize processes.

Predictive approach

The goal of building a customized tool to support production control activities has been fully achieved, revolutionizing the way equipment is supervised.

In addition to this, the following objectives have been met:

ADVANTAGES

The change in perspective brought about by the new tool has led to a data-driven approach. Moving from a reactive logic, to a context that allows making more informed and quick decisions, optimizing processes, reducing costs, and inefficiencies.

We were approached by a company in the manufacturing sector that needed to leverage data from sensorized machinery more effectively.

The project undertaken was aimed at providing control and maintenance managers with a tool to **evaluate data from different sources** in real time and perform p**reventive maintenance** on equipment, reducing production inefficiencies and extraordinary maintenance interventions.

Assessment: We identified all the relevant data for the project, **aggregating different sources** into a single repository. The work of normalizing and integrating the data allowed us to **obtain clean data** to compare with the engineers and experts of the client company.

Development: To immediately test the clarity of the information, statistical hypotheses, and heuristics, we **developed a web app**, which allowed for guided development and was functional to the objectives.

Feedback: Evolution of the system and application based on **continuous testing and exchange of information** among the stakeholders, to increase the effectiveness of the information produced and evolve the solution.

The implementation of the solution required a step-by-step working methodology, fully focused on incremental improvement.

The project stages were as follows:

The great advantage of the solution is the mode of development in progress, which through the feedback tool leads to improved modifications.

The output produced, in addition to a series of customized dashboards for data visualization, includes an **automatic warning system**. These warnings are generated both by some heuristic rules, defined in collaboration with the client, and by statistical analysis of historical data. The predictive capability of the tool is constantly updated, thanks to refinement work on the warnings themselves through analysis of the reports made by technicians on the generated alerts. Finally, the application offers the possibility of performing **retrospective analyses** on the data.

If you are interested in learning more about how the **development of advanced algorithms** can support predictive maintenance and other activities, please do not hesitate to contact us for consultation. We are ready to help you develop solutions tailored to your needs

The project represents only the first step towards an increasingly effective digital monitoring and classification activity. The ultimate goal is to implement effective **predictive maintenance** tools, using this technology to perform automated scans over time. The aim is to have faster and more efficient support for **structural analysis** and maintenance forecasting activities, making processes more effective and quicker in a context where currently many activities are time-consuming and require a significant amount of energy.

Digital twin reconstruction for structural analysis for infrastructure monitoring

The main challenge of the project was to **digitally reconstruct** a 3D model of the individual elements that make up the infrastructure, precisely identifying columns, beams, abutments, joints, and other typical components of road structures, following a well-defined classification.

The need was to **create a digital twin** with the highest level of conformity to reality, training algorithms to accurately "label" individual elements from data collected from aerial surveys conducted with drones through recordings and LIDAR scanning technologies.

The large amount of data collected by the drones translates into a cloud of hundreds of millions of points, whose **segmentation** process into its subcomponents represented the main challenge of the project.

The project was successfully completed and all the **accuracy** objectives for the digital reconstructions requested by the client were achieved. In particular, it was possible to create digital twins of the structures, identifying individual elements with a precision of over 90%. This paved the way for a new phase of the project aimed at leveraging what has been created to improve structural **monitoring and analysis activities** in a path of continuous innovation and growth.

Movyon is a leader in the development and integration of Intelligent Transport Systems, Tolling, and Infrastructure Management solutions, and a center of excellence for research and innovation within the Autostrade per l'Italia Group. The R&D department of Movyon approached Deix with the goal of experimenting with an innovative way to **periodically monitor and analyze** roads and highways.

With the advent of new technologies, data acquisition has become more accessible and faster, but data processing still requires precision and significant effort to ensure that collected data can be analyzed and made available to decision-makers.

The use of **Machine Learning**, through the use of neural networks for automatic learning

An **Unsupervised Approach**, using CAD models (BIM - Building Information Modeling) to create associations between the models and drone measurements

The project has seen the use of different approaches to reach the solution. In particular, two methods have been considered

Both approaches showed some difficulties that were overcome thanks to constant collaboration with the client.

In the case of machine learning and neural networks, the need to work with a reduced availability of labeled data led us to work together with the client to make the best use of the available information to generate an automatic tagging system capable of providing artificial intelligence with sufficient information to initiate automatic learning.

Regarding the unsupervised approach, the challenge was to create alignments and associations between the images acquired by drones and the CAD models used as a reference, which were not always exactly faithful to the real infrastructure. This was made possible through the use of ad hoc developed multistart optimization algorithms.

The R&D work performed in total synergy, which combined data labeling, cleaning and integration activities with algorithmic development, was crucial for the successful completion of the project.

If you are interested in improving safety in your workplaces and preventing accidents, we are at your disposal. Tell us about your needs and we will find the most suitable solution together.

Over time and through the use of neural networks, it is possible to train the technology to recognize all objects and movements deemed useful for better safety assessment.

This translates not only into the possibility of monitoring additional trucks and machinery, but even identifying safety devices to assess their proper use or verify in real time the actions performed comply with optimal procedures and directives to ensure the expected safety levels.

Starting with a limited area of the production plant, the solution will be replicated to the entire plant covered by the current activity and then to the company's additional production sites, in order to have an overall picture and control over all activities.

The tool enables safety officers and managers to have more reliable and comprehensive data to support strategic decisions. Real-time point data, constantly updated heat maps, and identification of common patterns are just some of the possibilities offered by the solution.

By capturing data over time, it will be possible to get to the point of identifying recurring situations at the plant level or specific areas, to facilitate effective safety measures, including, for example, the ability to report in real time dangerous situations and avert possible accidents.

Production workplace monitoring and incident prevention

Too often in the world of workplace safety, there is a **reactive approach to accidents**, implementing corrective actions to ensure that what happened does not happen again.

The challenge we set ourselves together with the client was to support the adoption of **preventive and proactive actions** to solve a widespread problem: accidents and injuries in manufacturing environments.

The scenario we were confronted with is characterized by environments with substantial **traffic of people and vehicles**, leading to a high risk of accidents.

Near Misses, situations where a potential accident has been escaped, were successfully mapped.

Mapping risk situations

The implemented solutions have increased awareness of operational spaces, contributing to a better understanding of everyday work, in terms of processes, actions performed and actors/elements involved.

Environment awareness

Identification of actual risk situations has improved all prevention activities, leading to revision of procedural directives and safety protocols based on acquired data.

Previsional perspective

The goals achieved have gone far beyond the needs for which the project was initiated.

This is a real shift in perspective from a reactive to a data-driven approach, based moreover not on generic industry models, but tailored to the client's specific situation.

The project was born in the area of workplace safety, with the aim of providing **operational and strategic support to the HSE department** (Health, Security and Environment) of a large manufacturing company.

Deix was involved as a partner to identify and implement a technological solution capable of improving **accident prevention in manufacturing environments**.

Our work began with the **installation of a series of cameras** capable of monitoring from different perspectives an intersection where pedestrian and forklift workers operate at the same time.

Objects and people identified by the cameras were **mapped onto a reference representation** of the area of interest so that their absolute positions and actual speeds could be known

We then trained the technology to **identify objects and movements** using **neural networks**, making detection and monitoring increasingly effective.

The output produced, a **custom-developed application**, enabled the collection of timely data and specific analyses, with the goal of identifying useful insights to support the daily activities of the security department.

To achieve the goal, we decided to monitor a specific work area in order to map high-risk situations and so-called *near misses.*

The project will involve continuous stages of evolution to improve the implemented solution and initiate large-scale expansion.

### BENEFITS

The adopted methodology and approach include as an essential element of success the constant discussion with the client, in order to acquire all the know-how necessary to customize the final solution based on specific needs and the implementation context.

Success Stories

We support our clients in developing innovative solutions and follow a proven approach to go from zero to production

Taylor-made solutions for your business

How We Work

Success Stories

Algorithms for the reconstruction of the breast geometry from medical imaging

Data-driven management software for production plants

Digital twin reconstruction for structural analysis for infrastructure monitoring

Production workplace monitoring and incident prevention