AI-Based Diagnostic Tools

Artificial intelligence in pathology is not a concept of the future but a clinical reality. CE-IVD-marked and FDA-cleared solutions are already used to screen for prostate, breast, lung and colorectal cancer. KombiMED integrates AI modules into a laboratory's existing digital infrastructure.

Artificial Intelligence in Pathology: Current Status

Modern AI solutions for pathology are based on deep learning and analyze Whole Slide Images (WSI), full-resolution digital images of histological slides. The algorithms are trained on hundreds of thousands of annotated slides and can:

AI does not replace the pathologist; it acts as a "second opinion," reducing the risk of missed findings and speeding up the processing of routine cases.

Detect malignant cells and micrometastases with a sensitivity comparable to that of an expert pathologist
Quantify biomarkers (Ki-67, PD-L1, HER2, TILs)
Grade tumors using the Gleason score, the Nottingham grading system and other systems
Prioritize urgent cases in the laboratory workflow

Certified AI Solutions (CE-IVD / FDA)

The following solutions have undergone clinical validation and hold regulatory approval for diagnostic use:

Paige Prostate / Paige Breast

Manufacturer: Paige AI (USA)

The world's first FDA-approved AI solution for pathology. It analyzes WSI of prostate and breast biopsies and highlights areas with a high probability of malignancy.

Ibex Galen™ Prostate / Breast

Manufacturer: Ibex Medical Analytics (Israel)

A CE-IVD platform for histopathological diagnosis. Galen™ analyzes up to 20 pathological patterns on a single slide, including perineural invasion and cribriform growth.

Philips IntelliSite Pathology Solution + AI Apps

Manufacturer: Philips Healthcare (Netherlands)

An FDA-approved digital pathology platform with validated AI applications for prostate and lymph node analysis.

Aiforia Clinical Suite

Manufacturer: Aiforia Technologies (Finland)

A CE-IVD solution for WSI analysis. It covers tumors of the prostate, breast, lung, colon and stomach.

DeepDx® Prostate

Manufacturer: DeepBio (South Korea)

An MFDS- and CE-IVD-certified AI system for prostate cancer diagnosis with Gleason grading.

Research AI Platforms (RUO)

For scientific and research applications, powerful platforms are available that do not yet have clinical certification but are actively used in leading medical centers:

PathAI: a deep learning platform for biomarker discovery and companion diagnostics development
HALO AI (Indica Labs): classification and quantitative analysis of WSI with customizable models
Lunit SCOPE Pathology: AI scoring of TILs and PD-L1 for immuno-oncology
Proscia Concentriq AI Modules: plug-ins for the analysis of melanoma, breast cancer and prostate cancer
QuPath: an open-source platform for quantitative analysis of biomedical images

Next-Generation Foundation AI Models

A groundbreaking direction is Foundation Models for Pathology: large-scale models trained on millions of WSI without being tied to a specific task. After fine-tuning, they address a wide range of diagnostic tasks:

These models are still at the research stage, but they are setting the direction of clinical AI development for the next 5–10 years.

Virchow (Paige AI): a model with billions of parameters for general-purpose pathology
UNI (Harvard/BWH): a vision transformer for multi-task analysis
Prov-GigaPath (Microsoft Research): a gigapixel model for WSI
CONCH (Harvard/BWH): a multimodal model that combines text and images

Benefits of AI Diagnostics for the Laboratory

AI algorithms detect micrometastases and small foci that may be missed during routine review.

Quantitative biomarker analysis eliminates inter-observer variability, which is especially critical for Ki-67 and PD-L1.

Automatic sorting of cases by complexity allows the pathologist to focus on the most significant cases.

AI pre-screening reduces TAT by 30–50% for routine biopsies.

Certified solutions have undergone multicenter clinical studies with confirmed specificity and sensitivity.

Higher sensitivity
Standardized assessment
Optimized workflow
Shorter turnaround time (TAT)
Clinical safety

Multimodal AI and Early Cancer Detection

A special area is multimodal AI, which combines data from various sources: WSI images, genomic sequencing results (NGS), transcriptomic and proteomic data, radiological images and clinical information. This comprehensive approach improves diagnostic accuracy and makes it possible to create digital biomarkers that predict response to therapy.

Solutions for early cancer detection:

Hologic Genius Digital Diagnostics: AI for cytological cervical cancer screening
Landing AI Cytology: automated analysis of cytology slides
Tempus Digital Pathology: a multimodal platform for integrating morphological and molecular data
Lumea AI Integration Platform: linking digital pathology data with NGS results

Integrating AI into Existing Infrastructure

AI modules connect to the digital pathology platform and work as an add-on to the existing workflow:

1. The slide is scanned on a WSI scanner

2. The image is transferred to the IMS (for example, PathoZoom® Digital Lab)

3. The AI module analyzes the WSI and generates a heat map / report

4. The pathologist reviews the AI result and makes the final diagnosis

5. The data is integrated into the LIS and linked to the patient's case

It is important to emphasize that AI works in "second opinion" mode: the final decision always rests with a qualified pathologist. Regulatory authorities (FDA, EMA) approve AI solutions specifically as decision-support tools, not as a replacement for the physician.

KombiMED Solutions

KombiMED supports the full cycle of AI diagnostics implementation:

Request a consultation: we will help you select a certified AI solution that matches your diagnostic tasks and regulatory requirements.

*KombiMED: centralized procurement of medical equipment from Europe.*

Audit of the laboratory's current digital infrastructure
Selection of AI solutions to match the institution's clinical profile
Integration with existing scanners, IMS and LIS
Training pathologists to work with AI tools
Technical support and model updates