Inside the product: the fundamental principle
Rilemo’s portable and battery-powered devices are based on Electromagnetic Radar Spectroscopy Imaging (ERSI), a safer alternative to X-rays and CT, based on transmitting electromagnetic waves towards a patient and receiving their reflections to generate images.
ERSI uses non-ionising waves, which do not cause DNA damage as X-rays can. These waves propagate and reflect inside patients depending on the dielectric permittivity of their tissues, a property closely related to fluid content.
This relationship between the high sensitivity of ERSI to dielectric permittivity and the link between dielectric permittivity and fluid content is what makes it possible to build a device capable of detecting the multiple pathologies described in the 
Product Overview section.
How does it work?
ERSI is built on three steps: signal transmission, reflection reception, and image generation.
The device is secured on the anatomical area of interest with a soft pressure comparable to a sweatband. Because ERSI is not sensitive to most clothing or hair, the patient can remain dressed without complex preparation.
Once in position, the antennas inside the device emit electromagnetic waves at low power towards the patient. These waves operate in the 0.5–4 GHz range, technically classified as microwaves and similar in frequency to Wi-Fi and Bluetooth signals. At these frequencies, electromagnetic waves are non-ionising and do not pose a health risk even with long-term exposure.
After emission, signals reflect off body tissues depending on the dielectric permittivity of each region and travel back to the device, where they are captured and processed into an image. The entire acquisition completes in under 5 seconds, with a Specific Absorption Rate (SAR) below 200 mW/kg, which is 8x lower than a typical smartphone and 16x lower than an MRI.
This imaging method was developed at the Italian National Institute for Nuclear Physics (INFN) and the European Organisation for Nuclear Research (CERN), derived from Microwave Imaging (MWI) with three fundamental advantages:
- it is quantitative rather than purely differential;
- it is real-time rather than minutes-long;
- it is compatible with portable, low-power devices.
- it is quantitative rather than purely differential;
- it is real-time rather than minutes-long;
- it is compatible with portable, low-power devices.
Image generation: from signal to clinical image
Once the electromagnetic waves interact with the body, the dielectric permittivity of tissues introduces distortions in the reflections. These reflected waves are captured by the system and processed by Rilemo's proprietary AI pipeline, producing three complementary image modalities simultaneously for every acquisition:
- Tomography: the anatomical reconstruction of the imaged region, providing a field of view comparable to a CT scan, but without ionising radiation.
- Radar (contrast) image: a contrast-based representation of internal echoes, highlighting boundaries and structural transitions inside the imaged volume.
- Dielectric Permittivity map: a quantitative representation of the dielectric properties of the tissues, directly correlated with fluid content, blood, cerebrospinal fluid, and inflammatory or oedematous changes.
These three modalities are produced simultaneously for every acquisition, allowing the clinician to interpret anatomy, structural change, and quantitative tissue composition side by side.
The quantitative reconstruction is achieved by solving complex inverse scattering problems, made possible by proprietary AI algorithms. The electromagnetic data are processed and reconstructed into images in only a few seconds, a speed advantage that is among the core differentiations from the state of the art, and what enables the vision of bringing brain imaging to the bedside.
In addition, the algorithms automate motion correction, noise reduction, image superposition, and alignment with other imaging modalities such as X-ray or CT, unlocking fusion imaging capabilities with existing anatomical references. The fusion approach combines the anatomical context that clinicians already know with fluid-related functional details, supporting fast clinical decisions while reducing the learning curve.
What the device output looks like
The figures below show the three-modality output applied to both a healthy head model and a pathological case, simulated through the device's electromagnetic digital twin and reconstructed by the production AI pipeline.
The tomography panel reconstructs the anatomical structure of the head: the skull, the brain parenchyma, and the main internal compartments are clearly visible, with a field of view comparable to a low-resolution CT scan. The radar panel highlights internal echoes and structural contrast, useful for identifying boundaries and abrupt transitions in tissue properties. The dielectric permittivity panel quantifies the dielectric properties of the tissues at every point, showing the homogeneous and physiologically expected distribution of a healthy head.
In the pathological case, the radar panel clearly identifies the haemorrhagic event, while the dielectric permittivity map shows the localised change in fluid-related tissue properties. The diagnostic discrimination between healthy and pathological cases, including central haemorrhagic events and wedge haemorrhagic strokes, is presented in 
Clinical Validation.
The dataset used to validate the imaging pipeline is built from 30 anatomically realistic head models drawn from the IT’IS Foundation PHM repository. Tissue dielectric properties are taken from the IT'IS Foundation database, the de facto standard for electromagnetic medical-device simulation, with literature-guided variability to reflect inter-patient differences. The full electromagnetic acquisition chain is simulated at the antenna, signal-generation, and acquisition level, and the output is reconstructed by the same AI pipeline that runs on the production device.
Why Rilemo?
Microwave Imaging was first proposed as a biomedical imaging method in the 1980s [1], but the components needed to make it practical and portable became affordable only in the 2010s, on the back of decades of telecom-driven RF investment. From that point onward, MWI demonstrated functionality in a series of clinical investigations including breast cancer detection [2], lung cancer diagnostics [3], brain stroke detection [4], and extremities imaging [5].
The technology is now ready, and the structural demand is documented. The clinical and economic evidence in 
Market Overview confirms the gap, and the 
Competitive Landscape section confirms that the category is beginning to form. What has been missing is a device built around what clinicians actually need: fast, portable, easy to use, and capable of providing functional insight rather than a purely differential image.
Rilemo's team addressed the three constraints that historically prevented microwave imaging from leaving the laboratory:
1. Image generation time. Common systems work by solving complex inverse electromagnetic problems and can require up to 20 minutes for a single image [6]. The device works with a proprietary AI algorithm, kept as a trade secret, trained to recognise patterns and producing the final image in under 5 seconds.
2. Physical dimensions. Most current microwave systems are built around an expensive third-party unit integrated into their assembly, making the final product at most transportable [6]. Rilemo moved to an architecture that miniaturises down to a single integrated electronic system, as portable as a headband.
3. Functional insight. Current systems focus on differential images. Although well suited for cancer screening, they cannot identify the kind of fluid present in the body [6]. Rilemo developed a proprietary approach, kept as a trade secret, that allows the device to tag and recognise different body fluids, essential for fast and accurate triage. The technique is built on known tissue parameters and on synthetic datasets generated in Rilemo's laboratories.
Defensibility
Rilemo's defensibility rests on a stack of complementary advantages, none of which can be easily replicated by a new entrant.
Layer | What protects Rilemo |
Algorithmic trade secrets | The AI reconstruction pipeline delivering images in under 5 seconds, and the multi-fluid recognition approach, are kept as trade secrets built on years of in-house R&D. |
Proprietary training datasets | The AI pipeline is trained on proprietary synthetic datasets generated internally. Data that competitors cannot access or replicate. |
Calibration architecture | Each Calibration Phantom is uniquely serial-number-linked to the device, with an expiry algorithm tied to cumulative environmental exposure. This architecture binds imaging performance to a controlled consumable that only Rilemo supplies. |
RF and antenna know-how | Designing and manufacturing a portable 16-port phased antenna array operating in the 0.5-4 GHz range requires deep specialist knowledge, developed through the CTO’s work at INFN and CERN. |
Regulatory head start | The regulatory documentation completed to date, including a comprehensive system requirements baseline and a closed hazard analysis, represents a head start that a new entrant would need 12 to 24 months to reproduce. |
CDMO and supply-chain | An established manufacturing partnership covers design transfer, production, and regulatory support. The audited bill of materials and the qualified supplier network already in place create an operational advantage that would take years to replicate. |
Clinical relationships | The structured interviews and IRCCS-class advisory relationships are not capital-replaceable. They translate directly into early reference accounts, clinical co-authorship, and faster regulatory and commercial cycles. |
Patent prosecution | Three patents cover the core imaging method and portable implementation, including one granted in Italy, Europe, and the United States. Two additional patents filed in 2025 have received positive Freedom to Operate and Absence of Prior Art on independent searches. Further prosecution and new filings are planned. Full details in the  |
Technical performance
The following parameters reflect the device's current design target. Safety and electrical classifications are aligned with MDR Class IIa requirements, as described in the 
Regulatory Pathway section.
Parameter | Value |
Time from device off to ready for imaging | < 60 seconds |
Time per single image | < 5 seconds |
Time to result (TTR) | < 2 minutes |
Linear image resolution | < 5 mm |
Maximum electromagnetic power | ~30 mW (~2% of a typical smartphone's maximum power) |
Specific Absorption Rate (SAR) | < 200 mW/kg (smartphones reach up to 1,600 mW/kg, MRIs up to 3,500 mW/kg) |
Frequency of operation | 0.5–4 GHz |
Battery powered | Yes, with seamless transition between battery and mains |
Battery autonomy | ≥ 20 scans (Head Unit, single-image mode) or 12 hours (serial imaging with Bedside Unit connected) |
Maximum patient skin temperature | < 41°C |
Water and dust protection | IP22 |
Drop resistance | 1 m |
Electrical safety class | Class II, Type BF applied parts (IEC 60601-1) |
Software safety class | IEC 62304 Class B |
System weight | < 5 kg, distributed across the three system components |
References
This Data Room is provided to you by:
Rilemo S.r.l.
Via Don Sebastiano Colleoni 20, Robbiate (LC), 23899, Italy
EU VAT: IT 04187140134 - ATECO: 26.60.02
PEC: rilemo@legalmail.it
Made with Bullet