Artificial intelligence in digital pathology
Company
St. Elizabeth Cancer Institute
Project Duration
9 months
Implemented system
ANNA – a system for analyzing immunohistochemical samples
Area
Healthcare
Our customer
The St. Elizabeth Oncology Institute is a modern medical facility that serves the needs of oncology patients from all over Slovakia. There are 7 bedded wards in the Oncological Institute of St. Elizabeth: surgical, oncogynecological, maxillo-facial surgery, intensive care, internal medicine, radiotherapy and nuclear medicine. The Institute’s laboratories are one of the few departments in the Slovak Republic accredited for the entire European Union.
The St. Elizabeth Cancer Institute invests in the acquisition of new diagnostic and treatment equipment, as well as in the further improvement of patient care. We at BRAIN:IT are also helping this development to move forward thanks to a project that we have created together.
Our challenge and goals
According to research conducted by experts at SACI on the issue of sample evaluation, existing software solutions are available, but they are extremely costly and still fail to deliver satisfactory results. Therefore, SACI entered a contractual collaboration with BRAIN:IT to create a software solution that would address this problem using artificial intelligence.
The core problem was the time-consuming and often imprecise evaluation of quantitative parameters in immunohistochemical staining, including manual counting of mitotic figures in tumors, detection of micrometastases in lymph nodes, and screening of cytological samples.
The expected benefit of the research lies in improving the quality of pathology diagnostics, helping patients receive the most accurate results in the shortest possible time. The development of the software/application will assist pathologists in precisely determining and evaluating tumor status and growth progression — specifically the proliferative activity. The application will determine the exact percentage of positive tumor cells out of the total number of tumor cells, ignoring elements such as stroma, lymphocytes, or other artefacts.
Our solution
The solution to the problem described is a fully functional software system that assists pathologists in determining the stage of cancer. At BRAIN:IT, we developed ANNA, a software platform that provides a module for quantitative and semi-quantitative evaluation of immunohistochemical staining of nuclear markers. It can determine the percentage of positive tumor cells from staining without counting other cell elements present.
Our ANNA application, developed in cooperation with the St. Elizabeth Cancer Institute in Bratislava, is a decision-support tool that helps pathologists precisely determine and assess the state and progression of tumor growth — specifically, the proliferative activity. The evaluation of proliferative activity in patient samples is performed through immunohistochemical staining of nuclear markers (assessment of the proliferation index using Ki-67).
The pre-processed samples are scanned by a digital microscope into electronic form, and the resulting digital image is then analyzed by artificial intelligence, which detects the presence and quantity of tumor cells in the background. The application determines the exact percentage of positive tumor cells out of the total number of tumor cells, while ignoring elements that do not belong to tumor cells, such as stroma, lymphocytes, or other artifacts.
Multiple modules have been developed for the ANNA system, including:
ANNA Micro-metastasis – Module for detecting micrometastases in lymph nodes
ANNA Mitotic – Module for monitoring mitotic activity in tumors
ANNA Cytological – Module for evaluating cytology results
All modules function as decision-support tools with the aim of minimizing the time-consuming manual work of doctors/pathologists, increasing the efficiency of examinations, and saving time. The processing and evaluation workflow is the same as for the assessment of proliferative activity: images/scans are created and subsequently analyzed by artificial intelligence.
Benefits for the client
The application makes it possible to precisely determine the percentage of positive tumor cells and significantly increases the accuracy of quantitative parameters in immunohistochemical staining. As a result, pathologists save up to 60% of their time and can examine 30% more patients.
More patients treated
The application saves time, allowing pathologists to diagnose more patients.
Objectivity
The application ensures objectivity, as its output is an exact result rather than a subjective estimate.
Error reduction
The application eliminates errors that could affect the final diagnosis.
Project complexity
Size
1/6
Time
3/6
Finances
1/6
Complexity
5/6
Are you interested in a similar solution?
We will be happy to help you create a system tailored to your needs – from design to implementation.