seeing beyond


SEEING BEYOND THE VISIBLE FUTURE HEALTHCARE 2 S e n s . A I | 2 0 1 8 ABOUT FUTURE PROCESSING Our main goal and the central part of our operations is the use of software to offer solutions to business problems. The company


  2. 2 S e n s . A I | 2 0 1 8 ABOUT FUTURE PROCESSING Our main goal and the central part of our operations is the use of software to offer solutions to business problems. ▪ The company was founded in 2000 by Jarosław Czaja (CEO). ▪ Revenues above the level of PLN 104 million (2017). ▪ 900 people. ▪ 18 years of experience. ▪ 150 clients in portfolio. ▪ R&D project management. ▪ Development of our own products.

  3. 3 S e n s . A I | 2 0 1 8 OUR COMPETENCES 10+ 69 SUPPORT YEARS OF EXPERIENCE EXPERTS SPECIALISTS There are machine learning and Our experts will support the Future Processing has more than 10 We have 69 experts specialized medical imaging specialists customer’s projects and R&D years of experience in the in computer vision departments with their working for us development of IT solutions knowledge and competences that support medical imaging

  4. 4 S e n s . A I | 2 0 1 8 WE WORK IN ACCORDANCE WITH ISO Our specialists have knowledge of standards applicable in the medical industry. ▪ We meet the requirements of the ISO 13485 standard, as confirmed by the Certificate for the Management System according to ISO 13485:2016. ▪ Knowledge of standards ISO 13485 and 14971 and the ability to create software while complying with the 62304 standard allows us to cooperate with institutions and authorities in the field of medical diagnostics.

  5. 5 S e n s . A I | 2 0 1 8 Sens.AI ( E nhancing the diagnostic efficiency of dynamic C ontrast-enhanced imaging in personalised O ncology by extracting New and I mproved B iomarkers) Seeing beyond the visible

  6. 6 S e n s . A I | 2 0 1 8 WHAT IS Sens.AI? Sens.AI is a system for comprehensive and automatic DCE (dynamic contrast-enhanced imaging) analysis. ▪ It is a tool supporting the diagnosis of brain lesions through the analysis of magnetic resonance images after contrast enhancement. ▪ Sens.AI analyzes sequences of medical images in search of relevant diagnostic information. ▪ The purpose of this is: • to support the diagnosis of brain lesions, • to improve the efficiency of diagnosis of patients with tumors, • to save time spent on manual segmentation and image analysis.

  7. 7 S e n s . A I | 2 0 1 8 CLINICAL PARTNER Sens.AI is being created in close cooperation with: Oncology Center – Institute named after Maria Skłodowska -Curie, Branch in Gliwice (Centrum Onkologii – Instytut im. Marii Skłodowskiej -Curie, Oddział w Gliwicach) Gliwice Center of Oncology (Centrum Onkologii) is a multidisciplinary oncological center offering cancer patients all the highly specialized methods of combination therapy of all types of cancer that are recognized in the world. The Sens.AI project is carried out with the Department of Radiology and Imaging Diagnostics of the Oncology Center.

  8. 8 S e n s . A I | 2 0 1 8 MULTIDISCIPLINARY TEAM Two teams of experts are working on the Sens.AI system: RESEARCH TEAM DEVELOPMENT TEAM (RESEARCH) This team deals with the design and Consisting of experienced software analysis (theoretical and experimental) engineers – this team works on the of algorithms. It consists of experienced implementation of the final product. scientists (doctors of physical and technical sciences).

  9. 9 S e n s . A I | 2 0 1 8 DESCRIPTION OF THE SOLUTION ▪ It is a tool supporting the diagnosis of neoplastic lesions. ▪ Performing of analysis of medical images from magnetic resonance (MR) after contrast enhancement. ▪ Following contrast-enhanced MRI, the images obtained are analysed by Sens.AI. The system performs automated analysis of imaging data by performing image segmentation and analysis of contrast flow in brain tissue. After the cycle is completed, Sens.AI generates a clear and easy to interpret report in the form of DICOM files. ▪ The entire analysis process is carried out without user’s intervention.

  10. 10 S e n s . A I | 2 0 1 8 INNOVATIVE ALGORITHMS ▪ Automatic and very fast DCE analysis of medical images is possible thanks to the innovative algorithms created by Future Processing using machine learning techniques, especially deep learning. ▪ Algorithms, using machine learning techniques, allow effective analysis of a series of medical images , obtained during magnetic resonance imaging. ▪ The report (in DICOM format) includes segmented neoplastic lesions in dynamic brain imaging, contrast flow graphs, parametric maps and easy to interpret and analyze biomarkers.

  11. 11 S e n s . A I | 2 0 1 8 POSSIBILITY OF RISK ASSESSMENT ▪ Biomarkers, graphs and parametric maps of the brain can be used to analyze the characteristics and stage of development of the tumor, allowing risk assessment in patients with cancer. ▪ Thanks to this, the medical staff receives support in the process of diagnosing the patient. ▪ The medical staff interprets the results and makes the final diagnosis.


  13. 13 S e n s . A I | 2 0 1 8 SEGMENTATION - EXAMPLE Neoplastic lesion imaging in MR head image J. Nalepa et al.: ECONIB: A deep learning-powered DCE tool for assessing brain tumor perfusion without region-drawing, RSNA 2018 (in review). Pixels correctly classified as tumor are highlighted in yellow, false negatives in green, and false positives in red.

  14. 14 S e n s . A I | 2 0 1 8 PARAMETRIC MAPS - EXAMPLE Ktrans – the first and most important DCE parameter; the volume transfer constant for gadolinium between blood plasma and the tissue extravascular extracellular space (EES). Like a chemical reaction rate, its units are given in values of (1/time), such as min −1 . In brief, Ktrans reflects the sum of all processes (predominantly blood flow and capillary leakage) that determine the rate of gadolinium influx from plasma into the EES.* * References Ferrier MC, Sarin H, Fung SH, et al. Validation of dynamic contrast-enhanced magnetic resonance imaging-derived vascular permeability measurements using quantitative autoradiography in the RG2 rat brain tumor model. Neoplasia 2007; 9:546-555. (Good correlation between Ktrans and autoradiographically measured influx rate in a rat glioma model with high permeability). Tofts PS. T1-weighted DCE imaging concepts: modelling, acquisition and analysis. MAGNETOM Flash 2010; 3:30-35.

  15. 15 S e n s . A I | 2 0 1 8 CONTRAST CONCENTRATION GRAPHS IN TUMOR TISSUE AND IN AIF - EXAMPLE Flow graphs J. Nalepa et al.: ECONIB: A deep learning-powered DCE tool for assessing brain tumor perfusion without region-drawing, RSNA 2018 (in review). Sample graphs of contrast saturation - orange dots indicate the average concentration of contrast in tumor tissue (the graph displays the model adjustment), and pink dots – the average concentration of contrast in AIF (artery input function; the graph presents the model adjustment).

  16. 16 S e n s . A I | 2 0 1 8 REPORT Sample report (in DICOM format) generated by the Sens.AI system.

  17. 17 S e n s . A I | 2 0 1 8 ADVANTAGES ▪ Automatic execution of both segmentation and analysis of medical images for each patient. ▪ Time saving – the system eliminates the necessity of manual segmentation and image analysis. ▪ DCE analysis and results in real time. ▪ Support for the diagnosis and prediction process. The results of the algorithm are not affected by the error of the human eye. ▪ Processing and selection of data for analysis without user’s intervention. ▪ Repeatability of results obtained during the segmentation and DEC analysis conducted. ▪ Analysis and comparison of results possible with use of publicly available DICOM file viewers.

  18. 18 S e n s . A I | 2 0 1 8 WHAT DO WE OFFER? 1. Sens.AI platform for easy integration with the hospital image archiving and communication (PACS) system, as well as algorithms for segmentation and DCE analysis of MR images after contrast enhancement 2. Innovative algorithms for: segmentation of neoplastic lesions in the brain and ▪ arteries from medical images coming from magnetic resonance (MR) after contrast enhancement – for customers who are looking for a modern tool for segmentation, segmentation of various organs (imaged in various ▪ modalities, such as MR, CT or PET), DCE analysis – for customers who already have a ▪ segmentation tool and want to extend it with automatic and fast DCE analysis. 3. Know-how regarding the application of machine learning methods in medical applications. 4. Support of specialists in the field of machine learning combining academic competencies with technological skills.

  19. THANK YOU What can we do together? Contact us and find out how our specialists can support your project. Future Processing ul. Bojkowska 37A 44-100 Gliwice, Poland +48 32 438 43 06

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.


More recommend