Custom Image Analysis

Mantalys replaces cumbersome manual annotations with streamlined, automated solutions. Whether you are looking to save time, improve reproducibility, or unlock quantitative insights for the first time, our team crafts bespoke image analysis pipelines precisely tailored to your requirements.

We offer custom image analysis solutions across imaging scales, from nanoparticles and cells to organoids, tissue, and whole-animal imaging. This includes electron microscopy, super-resolution microscopy, confocal microscopy, tissue imaging, intravital microscopy, and bio-imaging. Not sure if your imaging modality fits our scope? Email us at contact@mantalys.fr .

Our technology supports "few-shot learning," which means that only a handful of annotated images are required to train our AI solutions. The exact requirements may vary by project, but between 1-10 images is a typical starting point. Use high-quality, expert-annotated images to provide the best results with the fewest images.

No. Your images are only used to train your own algorithm, unless you explicitly give permission for them to be used for research purposes.

Please reach out to us at contact@mantalys.fr to discuss specific needs surrounding confidentiality and data security.

We have three approaches in our tool box to develop your customized solution: conventional deep learning (not transparent, but well-validated models can be used as a starting point); evolutionary computing (which generates fully transparent image processing pipelines), and hybrid AI (includes elements of both). If transparency is important for your project, we can use evolutionary computing approaches to develop a fully transparent, human-interpretable image processing pipeline.

Upon request, we can provide you with a written summary of the key methods used in your analysis, as well as provide a step-by-step overview of the image processing pipeline.

Yes! Reach out to us for more details at contact@mantalys.fr .

Yes! In addition to offering automation solutions to replace manual annotations, we can also develop individual image processing solutions to help with tasks such as denoising images or correcting uneven illumination.

Yes, we can train light, fast algorithms that are perfect for using on-the-fly during imaging or for integrating into your existing in-house software. Reach out to us for more details at contact@mantalys.fr .

For inquiries, please reach out to us using the "Contact Us" form or at contact@mantalys.fr . We will begin with an exploratory discussion about your analysis needs, and conduct preliminary feasibility testing with your images at no charge. If you then wish to proceed with our custom services, we can provide several options for different service packages within your budget.

Yes we do! Reach out to us for more details at contact@mantalys.fr .

Unlike LLMs, which require vast amounts of training data, our custom algorithms are trained specifically on your data.

We train our models on a subset of your image cohort, and then provide industry-standard quantitative readouts for how well our model recapitulates your manual annotations on previously unseen images. Depending upon your context, your project may have different performance requirements and these can be addressed during our initial exploratory discussions.

Our image analysis solutions are "deterministic", which means they are fully reproducible - the same image input will produce the same quantitative output no matter how many times you run the analysis.

You are welcome to request a solution with exclusive IP rights. Discounts can be offered in cases where exclusive IP rights are not required.

Yes! Depending upon your needs, we can customize our solution delivery in a format that is best suited to you. This can include a user-friendly interface. Reach out to us for more details at contact@mantalys.fr .

We work with a range of different budgets for our custom services and we can adapt the scale of your project to fit within your desired budget.

Depending upon your scientific question, the results of our custom analysis pipeline may take the form of an image (e.g. with noise removed); a set of polygons (e.g. segmentation masks), or numerical data (e.g. a single-cell feature table). During our consultation, we will discuss what format and file type you would like your results delivered in.

Yes, we can train algorithms for you that are deployable within MantaPlex. Reach out to us for more details at contact@mantalys.fr .