The Breast IHC Tumor Tissue Detection App is a pre-trained HALO AI masking classifier designed to detect, quantify, and segment tumor area across hematoxylin and DAB-stained whole-slide digital images. Outputs include the area and percentage area classified as tumor or other. The App can classify tissues within regions of interest, fields of view, or across entire images and can be combined with HALO® Analysis Modules and other HALO AI Apps and to derive further quantitative outputs and measurements. In the examples shown below, the App is used with HALO Analysis Modules to facilitate tumor-specific analysis of biomarker expression. As an RUO product, the Breast IHC Tumor Tissue Detection Classifier can be further trained, tuned, and improved by end users using their own images and data.
Intended Use
For Research Use Only and not intended for clinical diagnostic use.
What's included?
- Breast IHC Tumor Tissue Detection App
- Benign Epithelia Classifier – Used to identify and exclude benign epithelia from analysis. Please note, the Breast IHC Tumor Tissue Detection App can be used with or without the Benign Epithelia Classifier.
Training Information
- Breast IHC Tumor Tissue Detection App training was performed using 4500+ annotations from hundreds of DAB and hematoxylin-stained whole slide images of primary invasive breast cancer.
- Tissues used for training and validation were probed with four different tumor biomarkers localizing to nuclear and membrane compartments and the final training annotation set included biomarker-positive and negative samples.
- Tissues were obtained from multiple institutes using different scanner types to improve overall generalizability of tumor detection. Note, not all of the file formats shown under ‘File Format Compatibility’ tab were included in the training set.
Prerequisites
All pre-trained HALO AI Apps require an existing license of HALO and HALO AI upgraded to version 4.0.5.
File formats supported by the HALO image analysis platform:
- Non-proprietary (JPG, TIF, OME.TIFF)
- Nikon (ND2)
- 3D Histech (MRXS)
- Akoya (QPTIFF, component TIFF)
- Olympus / Evident (VSI)
- Hamamatsu (NDPI, NDPIS)
- Aperio (SVS, AFI)
- Zeiss (CZI)
- Leica (SCN, LIF)
- Ventana (BIF)
- Philips (iSyntax, i2Syntax)
- KFBIO (KFB, KFBF)
- DICOM (DCM*)
*whole-slide images
Accelerate your AI Development
AI is a powerful tool in your image analysis toolbox, but AI development is time consuming and data intensive. Pre-trained using hundreds of images and training annotations, HALO AI Apps give users a jumpstart to accelerate AI development.
Refine with your Training Data
For most applications, HALO AI Apps will work right out-of-the-box, but we realize that it is impossible to test every application. Importantly, HALO AI Apps are ‘open’ and can be further trained by users to optimize and refine performance for specific applications and stains.
Complement your Expertise
HALO AI Apps are designed to handle time-consuming and tedious tasks and providing consistent, standardized measurements. You are free to apply your scientific expertise where it’s needed most – in the interpretation of data to make informed decisions.
Optimized for Generalizability and Flexibility
Having been trained and tested on tissues with variable staining from different scanners and with multi-institutional data, our HALO AI Apps are designed to achieve the highest level of generalizability right out-of-the-box, but with the flexibility to be trained further with your data. Here we see the Breast IHC Tumor Tissue Detection App accurately detects tumor in breast biopsies and resections with cytoplasmic (left), membraneous (middle), and nuclear (right) staining.
Seamlessly Integrate with Other HALO AI Apps and HALO Image Analysis Modules
The Breast IHC Tumor Tissue Detection App can be used in conjunction with other HALO AI Apps or HALO modules to derive additional quantitative outputs. In the examples shown here, the App is used with HALO Area Quantification BF (top row) and the Multiplex IHC Module (bottom row) to quantify biomarker positivity specifically within the tumor tissue class. In the top example, there is significant non-specific staining within the stroma which is eliminated from analysis by the App. These examples further highlight the flexibility and robustness of the Breast IHC Tumor Tissue Detection App to identify tumor with heterogeneous patterns and intensities of staining.
Please note: HALO AI Apps and associated applications are intended for research use only. Please visit our Clinical Products page to discover clinical AI products deployable through our HALO AP® platform.
Free Proof-of-Concept Analysis
See HALO AI Apps in action on up to three of your images with a free proof-of-concept analysis.
Related HALO AI Apps
The NSCLC IHC Tumor Detection App is a pre-trained HALO AI classifier designed to detect, segment, and quantify tumor area and non-tumor area across hematoxylin and DAB-stained whole-slide digital images of NSCLC.
Learn More
The Breast IHC Cancer Cell Phenotyper App is a pre-trained HALO AI object phenotyper designed to detect, segment, and quantify cancer cells and other cells across hematoxylin and DAB-stained whole-slide digital images of breast cancer.
Learn More
The Pan-Cancer H&E Lymphocyte Cell Phenotyper App is a pre-trained HALO AI object phenotyper designed to detect and quantify lymphocytes across whole slide H&E-stained images of multiple tumor types.
Learn MoreUse the arrows above to view additional related AI apps
Want to Learn More?
Fill out the form below to request information about any of our software products.
You can also drop us an email at info@indicalab.com