DeepDx Prostate (RUO)
DeepDx Prostate (RUO) is an AI algorithm for analysis of H&E-stained prostate core needle biopsies. Leveraging gland-level analysis, it detects and segments tumor regions to automatically generate Gleason Scores, ISUP Grade Groups, and quantitative tumor metrics. The software provides precise segmentation overlays for Gleason patterns and separately identifies adverse histological features such as cribriform pattern 4, IDC-P, and perineural invasion, enabling researchers to standardize data and streamline workflows.
DeepDX Prostate is For Research Use Only. Not for use in diagnostic procedures.
Zoom image Automated Gleason Grading & Detection
DeepDx Prostate (RUO) automates the analysis of prostate biopsies using deep learning to standardize research workflows:
Gland-Level Analysis: The algorithm utilizes advanced neural networks to perform segmentation at the gland level, automatically differentiating between benign tissue structures and adenocarcinoma.
Instant Scoring: It instantly calculates the Gleason Score (e.g., 3+4) and ISUP Grade Group for the slide, providing a consistent, automated baseline.
Visual Verification: Results are visualized via gland-level segmentation masks that overlay the tissue.
- Gleason Pattern 3: Yellow overlay
- Gleason Pattern 4: Orange overlay
- Gleason Pattern 5: Red overlay
Quantitative Pattern Analysis
The algorithm calculates quantitative metrics for the specific Gleason Patterns present in the biopsy core. Rather than providing only a summary Grade Group, DeepDx Prostate (RUO) computes the precise area occupied by each pattern relative to the total tumor area.
The following pattern-specific metrics are reported as percentages:
- % Pattern 3: (Pattern 3 Area / Total Tumor Area)
- % Pattern 4: (Pattern 4 Area / Total Tumor Area)
- % Pattern 5: (Pattern 5 Area / Total Tumor Area)
These continuous variables provide granular data on tumor composition, supporting research into risk stratification for intermediate-risk cases.
High-Risk Histological Features
DeepDx Prostate (RUO) includes specific detection capabilities for adverse histological features associated with high-risk disease. These features are identified independently of the standard Gleason research masks for research purposes.
The algorithm generates distinct heatmap overlays for the following morphological subtypes:
- Cribriform Pattern 4
- Intraductal Carcinoma of the Prostate (IDC-P)
- Perineural Invasion (PNI)
- High-Grade Prostatic Intraepithelial Neoplasia (HGPIN)
The presence of these features is flagged in the results, and the localized masks allow for immediate visual confirmation of the specific regions containing these markers within the biopsy core.
Tumor Burden – Length Metrics
The algorithm automatically measures the linear extent of tumor across the biopsy core, including fragmented or curved samples. Linear metrics are derived from the segmentation mask and reported in millimeters.
The following length-based metrics are calculated as outputs for each slide:
- Total Tissue Length (mm): The cumulative length of all benign and malignant tissue fragments.
- Total Tumor Length (mm): The cumulative linear length of all detected tumor foci.
- Percent Tumor Involvement by Length: (Total Tumor Length / Total Tissue Length) × 100.
These standardized outputs replace manual measurements, providing consistent data for assessing linear tumor burden.
Tumor Burden – Area Metrics
DeepDx Prostate (RUO) calculates volumetric tumor burden metrics based on the two-dimensional surface area of the detected tissue, offering granular data essential for precision research. These metrics are derived from the total pixel count of the segmentation masks.
The following area-based metrics are calculated as outputs for each slide:
- Total Tissue Area (mm2): The total surface area of the biopsy core.
- Total Tumor Area (mm2): The total surface area classified as adenocarcinoma.
- Percent Tumor Involvement by Area: (Total Tumor Area / Total Tissue Area) × 100.
DeepDx Prostate (RUO) in Aperio HALO AP Demonstration
Explore the future of prostate cancer research with DeepDx Prostate (RUO). DeepDx Prostate seamlessly integrates within the Aperio HALO AP platform to analyze H&E-stained prostate core needle biopsies.
After selecting a case, run the DeepDx Prostate (RUO) assay to view the H&E slides and analysis results. Results include comprehensive quantitative data displayed in the assay panel, featuring ISUP Grade Groups, Gleason Scores, and Gleason pattern proportions. Visual overlays include color-coded, gland-level segmentation masks for Gleason Patterns 3, 4, and 5, as well as tumor length and tissue length measurements. Finally, the algorithm independently detects and localizes a comprehensive suite of critical adverse features—Cribriform Pattern 4, Intraductal Carcinoma (IDC-P), Perineural Invasion (PNI), and High-Grade PIN—which can be toggled as heatmaps for data-driven risk assessment.