Our AI technologies for computational pathology at the service of medicine

We built the most advanced technology to help

clinicians end cancer as a deadly disease.

Tribun Health is pushing the limits of technology to deliver solutions that have the potential to radically transform the
way cancer is diagnosed and treatment plans are decided upon. Our technology stack is as advanced as it gets:
Convolutional Neural Network powered Computer Vision.

We use those technologies in a very different way to circumvent seemingly blocking factors: digital images quality
variability, insufficient data and privacy and regulatory frameworks limiting data transfer.



Stain, scanner and tissue processing vaariability are inherent to slide preparation and digitalization. We built advanced AI tools that recognize and compensate for the variability and ensure consistent and accurate analyses.



When data is rare or difficult to acquire, our AI tools efficiently model data based on existing data, allowing our models to deliver accurate analyses despite limited amount of available data.



Privacy and compliance often limit the ability to transfer and centralize data. We leverage federated learning to run our algorithms where the data is, as opposed to in a central repository, thus removing obstacles to training our algorithms and delivering accurate analyses.

It all starts with data:

Digitalized slides, detailed annotations and medical records.

Whole slide images,
where every pixel tells a story.

It all starts with high-resolution digitalized slide images through sophisticated and dedicated scanners. Our platform is compatible with all major formats including Philips (.isyntax), Aperio (.svs), Hamamatsu (.ndpi), Leica (.scn), Sakura (.svslide), MIRAX (.mrxs), Ventana (.bif) and more.

Detailed annotations
by pathologists.

Pathologists analyze images to recognize and classify the tissue components, such as normal and abnormal tissue, including tumors, inflammation, necrosis, fibrosis, and more. With tumors, they annotate regions of interest including the tumor itself, invasive borderline and invasive front. Consistent and precise annotation by pathologists helps train our algorithms to recognize these regions automatically and help guide future analyses.

Medical history
and clinical data.

Patient health history and clinical data from other sources than biopsies are a powerful data-set to enrich the knowledge and allow for a more complete and multi-modality diagnostic.

Then, we leverage the power of AI automation

To extract and deliver relevant clinical insights.

Computer Vision:
Using Deep Learning.

We use Computer Vision to program algorithms to interpret visual information contained within image data (pathology images from digitalized slides) in order to make better sense of the digital data.

These algorithms translate this data into meaningful insights, using contextual information provided by pathologists in order to make better diagnostics decisions.

Deep Learning (DL) models are based on neural networks capable to learn from raw data and deliver new insights. When dealing with slide images we use a particular type of DL called convolutional neural network (CNN) capable to automatically extract relevant image features to make accurate decisions.

Big Data analysis:
Using Machine Learning.

Big data analytics helps make sense of the data by uncovering trends and patterns. Machine learning is used to accelerate this process with the help of decision-making algorithms. It can categorize the incoming data, recognize patterns and translate the data into insights helpful for clinically actionable insights in diagnosis, prognosis and treatment selection.

And finally, we deliver the clinical insights

to help clinicians with diagnosis and
treatment selection.

Clinical insights for
diagnosis, prognosis and
treatment selection.

Computational pathology aims to improve diagnostic accuracy and optimize patient care. It helps advance individualized precision medicine by delivering reliable and actionable clinical insights to treating physicians. Our technological approach delivers the highest level of accuracy and confidence in those insights.

We are currently developing algorithms for breast Immunohistochemistry (IHC), mitosis count, gastrointestinal (GI) biopsies triage, Non-Small Cell Lung Cancer (NSCLC) outcome prediction, and more are in our development pipeline.