Breast Cancer HER2 Subtype Identification

Predicting HER2 Status Using Histopathological Images.

Summary

Breast cancer is a major cause of mortality among women, with HER2-positive cases being particularly aggressive. Identifying HER2 status early and accurately is essential to improving treatment outcomes. However, the conventional FISH test, widely used for this purpose, is labor-intensive, time-consuming, and requires highly trained professionals.

About the Customer

The Dow University of Health Sciences (DUHS), one of Pakistan's oldest public universities, has been a leader in health sciences since 1945. Renowned for its focus on biomedical, health, and medical research, DUHS encompasses esteemed institutions like Dow International Medical College and Dow Medical College. The university offers a comprehensive range of undergraduate, postgraduate, and doctoral programs and has a strong postgraduate department overseeing medical sciences research.

Team composition
4 members
Expertise used

Computer Vision, Machine Learning.
Duration

4 months
Services provided

UI/UX development, App Development, Reporting, and dashboard development
Region

UK
Industry

IT Services and Technology

Understanding the Challenge

DUHS faced challenges with the traditional FISH testing method, as it is both time-consuming and dependent on specialized expertise, which can limit the speed and accessibility of HER2 testing. Given the high mortality rate associated with HER2-positive breast cancer, the university needed an efficient, accurate, and scalable solution to streamline HER2 testing.

Solution

Folio3 AI in partnership with Fidel AI, developed an AI-powered system automating HER2 testing for faster, accurate diagnosis.

Contrast Enhancement

Contrast enhancement is essential for highlighting details in histopathological images, allowing for clearer differentiation between structures in low dynamic range areas. This technique enhances image quality, enabling practitioners to perform noise reduction and improve overall clarity.

Cell Segmentation

Cell segmentation divides images into distinct segments, grouping pixels based on similar attributes such as intensity and texture. This step ensures each cell is accurately isolated, which is crucial for subsequent analysis.

Image Binarization
Image binarization converts grayscale images into binary format, a critical step for effective segmentation. Using an adaptive thresholding method (ISO Data Algorithm), the system identifies pixel tones, creating a clear center for segmentation.

Spot Counting
In the context of HER2 testing, each spot represents a "cytokine signature" of a single cell. The solution identifies true spots by detecting a dense center that fades towards the edges, measuring intensity and size to estimate cytokine levels accurately.

Book a Demo

Result

Thanks to this AI-powered solution, DUHS has achieved significant improvements in both the efficiency and precision of HER2 testing. Practitioners can now perform the test faster and with greater accuracy, storing digitized images for easy access and future analysis.

Technologies Used