Internet Explorer is not supported
Sorry, please use another browser such as Google Chrome or MozillaFirefox

Introduction to AI in custom-fit ear product design

/ Updated

Dr. Paolo Masulli

Balancing automation, flexibility, and configurability.

BTE and RIC moulds, hearing aid shells, and in ear monitors, are among the most common custom ear products. Each design must adapt to unique anatomy while meeting acoustic, aesthetic, and production requirements.

Across the industry, styles and specifications vary widely: canal lengths, retention styles, and material constraints all differ from one manufacturer to another. This diversity is a strength, but it makes automation a complex challenge. AI can help optimize the design process, but only if it respects these variations and integrates seamlessly into production workflows.

Why anatomy comes first

Every successful automated design begins with accurate anatomical understanding. Before any modelling can occur, the system must reliably identify key regions of the ear: the canal and its bends, the concha, tragus and antitragus, and more. These landmarks define the reference frame for every subsequent operation, such as tip cutting, spline generation, and vent placement. If landmarking is inconsistent, the entire pipeline becomes fragile. When it is robust, automation becomes predictable and trustworthy.

This is why input quality matters so much. A scan that lacks canal length or omits parts of the concha will compromise even the most advanced algorithm. 

 

High quality impressions or scans free from artefacts are the foundation of reliable automation.

High quality scans often translate into fewer remakes and faster manufacturing.

Modular AI flows for quality and control

Rather than treating design as a single “AI task,” the most effective approach is to break it into a series of operations. Typical steps include:

  • impression clean up
  • anatomical landmarking
  • tip cutting
  • shape generation
  • the addition of functional features such as bores and vents

This modular structure offers two key benefits: flexibility and transparency. Each operation can be optimised independently, validated against production examples and quality standards, and configured to reflect a manufacturer’s unique style.
 

Some operations lend themselves naturally to AI approaches. The tip cut, for example, involves subtle decisions about where and how to trim the canal and choose the angle of the cut. A specialised AI model trained on a large, diverse dataset can capture these nuances and deliver consistent results across thousands of cases. Other steps, such as generating shape splines are better handled by mathematical models with explicit parameters. These allow production managers to fine tune outcomes, maintain predictable behaviour, and ensure compliance with regulatory requirements.

How AI tackles the hard parts

Applying AI to custom fit ear product design is not about a single algorithm; it’s about using the right technique for each challenge. The process begins with landmark detection, where deep learning models trained on thousands of annotated scans identify critical anatomical regions. These models must generalise across a wide range of ear shapes, impression qualities, and scanning conditions. Accuracy here is non negotiable because every downstream decision depends on it.

Once landmarks are established, AI can assist in operations that require nuanced judgement. Machine learning models stemming from the field of Geometric Deep Learning prove to be a precious tool to obtain accurate results on three-dimensional geometric data, since these kinds of models can learn from expert decisions. If you want to determine the ideal tip cut for your earmould, finding the right location where to cut and how to angle the termination – balancing comfort, retention, and printability – this is the way to go. In fact, it is a task that resists simple rules but thrives on pattern recognition from large datasets.

For other steps, such as spline and surface generation, AI often plays a supporting role rather than taking full control. Here, the use of precise mathematical models with carefully-tuned parameters remains dominant because it offers explicit control over thickness, spline position, and blending behaviour. However, AI can still contribute by predicting optimal parameter ranges based on anatomy and sample sets.

Another area where AI adds value is quality assurance. Models can detect anomalies, such as incomplete canals or missing regions, before the design process begins, saving time and reducing remakes.

This hybrid approach, applying AI to perception and judgement-heavy tasks, and geometric models for the more deterministic steps, delivers the best of both worlds.


The 3Shape approach builds on a long history of innovation

With more than two decades of experience in the custom hearing aid and dental industries, 3Shape’s products build on a wealth of knowledge accumulated over the years, covering both 3D geometry algorithms and advanced AI and machine learning. Multiple patents held by 3Shape in these fields are a testimony to the continued innovation that made us market-leaders in solutions for custom 3D products and automation.

Our solution for automated modelling in the hearing industry, 3Shape Advance, demonstrates how the careful combination of modern AI with solid 3D geometry algorithms can produce high-quality custom earmoulds at scale.

Every time 3Shape Advance generates an output, its neural networks and algorithms – with tens of thousands of parameters – translate your unique styles and specifications into a 3D model customized to the individual ear. Thanks to its modular and transparent architecture, Advance is continuously improved as we fine tune and optimize each operation in its pipeline.

Automation that works for you

AI in custom fit ear product design is not about replacing expertise; it’s about supporting it and making high-quality products possible at scale. Success depends on three principles: 
  1. reliable anatomical detection
  2. modular architecture
  3. a balanced mix of AI and geometric methods

When these elements align, automation delivers faster turnaround, greater consistency, and fewer remakes, without erasing the unique design language that defines your products.

3Shape Advance embodies this philosophy: a platform built to integrate AI where it adds the most value, maintain transparency at every stage, and give manufacturers the control they need. Thanks to this approach, you can scale production confidently, delivering custom products that fit perfectly, perform reliably, and carry your signature style into the future. 

 

Dr. Paolo Masulli

Dr. Masulli is a Senior Product Manager in 3Shape Audio and has been with the company since 2023. Before joining 3Shape, he worked in Research and Development within academia and the health technology industry for more than nine years.

Dr. Masulli has a strong background in technical sciences and product development. He is passionate about the applications of advanced technologies to promote health and well-being, and proud to contribute to the transformation of the hearing industry at 3Shape Audio.

Curious about how AI-powered automation can benefit your business?