Microsatellite instability (MSI) is a critical biomarker in colorectal cancer (CRC), influencing diagnosis, prognosis, and treatment decisions. Traditional MSI detection methods, such as immunohistochemistry (IHC) and polymerase chain reaction (PCR), are time-consuming and often require specialized expertise. This can lead to delays in diagnosis and treatment, ultimately impacting patient outcomes.
PAICON’s AI Solution for Faster CRC Diagnosis
To address these challenges, we have developed an innovative AI-based tool capable of rapidly and accurately predicting MSI status directly from hematoxylin-eosin (H&E) stained slides of CRC tissue. By leveraging advanced deep learning techniques, our tool can analyze the morphological features of tumor cells and surrounding stroma to provide a reliable MSI prediction.
We trained our model using a multi-centric cohort of over 1,800 slides. For evaluation, we used two independent cohorts: one with over 600 cases and another with 250 cases. With our approach we could achieve an Area Under the Curve (AUC) of 91% and 93%, respectively, demonstrating the robustness of our tool and its ability to generalize effectively across different centers and countries.
Want to test your image? Register now on SatSight DX to try our AI model!
Example SatSight DX Report
Below is an example report generated by our SatSight DX, showcasing the type of detailed insights and analysis you can expect from your own data.
Outlook
As we continue refining and validating the tool, we’re dedicated to making it widely available in research and clinical settings. We believe this technology could transform how MSI is detected, leading to more personalized and faster treatments for patients. The versatility of our approach allows to apply it to other histopathology subtyping tasks, addressing a variety of clinical challenges through image-based learning.
