Measurements in medical imaging

We’ve created and continue to develop algorithms for extracting valuable data from medical images various modalities.

Take advantage of our know-how and capabilities to create customized solutions that will give you the quantitative data you need.


Medical image segmentation with machine learning

Our solutions based on machine learning automatise the segmentation process. The models we worked on contour organs and detect as well as localize structures within organs on various types of examinations.

Segmentation based on our deep learning method:

  • is precise and fast, provides repeatable results as well as enables extraction of measurable features of anatomical structures and lesions
  • has achieved an accuracy of more than 90% (Sørensen-Dice coefficient of roughly 0.9) [1] 
  • is characterized by effectiveness proven in many prestigious global AI challenges (such as BraTS and FeTS)
  • can be embedded into existing solutions


Extraction and measurement of quantitive data from medical images

The bespoke algorithms we’ve been developing for our clients are designed to meet particular needs and requirements. They are able to measure and analyze as well as provide valuable quantitative data. This technology can be used to detect any lesion, for any internal anatomical structure of a single patient.

icon medical examination

Detection of a single lesion on a scan, as well as multiple parts of a larger lesion

icon medical examination

Evaluation of measurable lesion according to a predefined requirements

icon medical examination

Lesions detection in time – detection of new lesions on a follow up scan 

icon medical examination

Evaluation of patient response to treatment also outside standard criteria


Proven experience 
Automated RANO and RECIST+

We created and applied fully automated algorithms for precise and comprehensive patient response measurements using RANO and RECIST criteria (automated RANO and automated RECIST) to real-world data. Furthermore, our technology is capable of much more than what the guidelines suggest.

Case study of automated RANO calculation

  • Volumetric measurements of predefined regions or subregions
  • Repeatability and objectivity of results
  • Integration with the segmentation algorithm, as all takes place in one step
  • Higher bidimensional measurements than those reported by most human readers [2]

Tumor parts segmentation and RANO calculation

Modality: MRI

Body part: brain

Input data

Sequences: T1, T1CE, T2, T2-FLAIR

Output data

Segmented necrosis: 14.907 cm3

Segmented edema: 99.434 cm3

Segmented enhancing tumor: 27.789 cm3

Segmented TOTAL: 142.13 cm3

Calculated RANO value: 1241.68 mm2

Brain MRI image with segmented lesion and its subregions

Automated segmentation examples


[1] Filip Malawski, Jarosław Gośliński, Mikołaj Stryja, Katarzyna Jesionek, Marcin Kostur, Karol Miszalski-Jamka, Jakub Nalepa, Deep Learning Meets Computational Fluid Dynamics to Assess CAD in CCTA, Lecture Notes in Computer Science book series (LNCS,volume 13540), 2022

[2] Jakub Nalepa, Krzysztof Kotowski, Bartosz Machura, Szymon Adamski, Oskar Bozek, Bartosz Eksner, Bartosz Kokoszka, Tomasz Pekala, Mateusz Radom, Marek Strzelczak, Lukasz Zarudzki, Agata Krason, Filippo Arcadu, Jean Tessier, Deep learning automates bidimensional and volumetric tumor burden measurement from MRI in pre- and post-operative glioblastoma patients, Computers in Biology and Medicine, Volume 154, 2023,

Let’s work on your challenges together!

Contact us:

Let’s work on your challenges together!

Contact us: