Medical ML: Ukrainian Medical Imaging Infrastructure — Current State and AI Readiness Assessment

---

Abstract

Ukraine’s medical imaging infrastructure stands at a critical inflection point, shaped by decades of post-Soviet underinvestment, ambitious healthcare reform since 2017, and the devastating impact of the ongoing Russian invasion since February 2022. This comprehensive analysis examines the current state of diagnostic imaging capabilities across Ukrainian healthcare facilities, assessing equipment density, geographical distribution, technological modernity, and digital infrastructure readiness for artificial intelligence integration. Our assessment reveals stark disparities: while Ukraine’s pre-war CT scanner density of approximately 7.2 units per million population placed it below European averages of 22-35 units per million, the war has inflicted catastrophic damage—with over 1,700 medical facilities damaged and 211 completely destroyed as of 2024, including critical imaging equipment losses in frontline regions. Despite these challenges, Ukraine’s electronic healthcare system (eHealth), which registered 36 million patients and processed 1.6 billion electronic medical records by 2022, provides a foundational digital infrastructure that could accelerate AI diagnostic tool deployment. We analyze regional variations across 24 oblasts, identifying Kyiv, Odesa, Lviv, Kharkiv, and Dnipro as potential AI implementation hubs based on equipment availability, internet connectivity, and radiologist concentration. Critical infrastructure gaps include limited PACS adoption (estimated at 15-25% of hospitals), aging equipment without DICOM compliance, unreliable power supply in conflict-affected areas, and acute radiologist shortages. Our findings inform a strategic framework for AI-assisted medical imaging deployment tailored to Ukrainian realities, prioritizing cloud-based solutions, mobile diagnostic units, and hybrid AI architectures optimized for intermittent connectivity.

Keywords: Ukraine healthcare, medical imaging infrastructure, CT scanners, MRI equipment, PACS, DICOM, eHealth Ukraine, radiology, AI readiness, healthcare reform, war impact, digital health

---

1. Introduction

The capacity of a nation’s medical imaging infrastructure fundamentally determines its diagnostic capabilities and, ultimately, patient outcomes. Computed tomography (CT), magnetic resonance imaging (MRI), ultrasound, X-ray, and positron emission tomography (PET) scanners form the backbone of modern diagnostic medicine, enabling early detection of cancers, cardiovascular diseases, neurological conditions, and traumatic injuries. As artificial intelligence increasingly transforms radiology practice worldwide, the readiness of this infrastructure—not merely in equipment quantity but in digital connectivity, data standards compliance, and workflow integration—becomes decisive for healthcare systems seeking to leverage algorithmic assistance.

graph TD
    A[Medical Imaging Infrastructure] --> B[Equipment Layer]
    A --> C[Digital Layer]
    A --> D[Workforce Layer]
    B --> E[CT Scanners]
    B --> F[MRI Units]
    B --> G[X-Ray/Ultrasound]
    C --> H[PACS Systems]
    C --> I[DICOM Compliance]
    C --> J[Network Connectivity]
    D --> K[Radiologists]
    D --> L[Technicians]
    D --> M[IT Support]

For Ukraine, assessing medical imaging infrastructure carries urgency beyond academic interest. The nation simultaneously confronts multiple interconnected challenges: the legacy of Soviet-era healthcare organization, ambitious modernization efforts through the National Health Service of Ukraine (NHSU) established in 2017, and the catastrophic destruction wrought by Russia’s full-scale invasion since February 24, 2022. Each factor reshapes the landscape of imaging capabilities and AI integration potential.

Before 2022, Ukraine had made significant progress in healthcare digitalization. The electronic healthcare system (eHealth), launched as a Ministry of Health priority in 2017, grew to encompass 36 million registered patients, 1.6 billion electronic medical records, and 400,000 active healthcare users processing 1,000-1,500 database requests per second. This digital foundation—rare among lower-middle-income countries—created unprecedented opportunities for data-driven healthcare transformation, including AI-assisted diagnostics.

graph TD
    A[Pre-War Infrastructure] --> B[2022 Russian Invasion]
    B --> C[Facility Destruction]
    B --> D[Equipment Damage]
    B --> E[Workforce Displacement]
    C --> F[Reconstruction Need]
    D --> F
    E --> F
    F --> G[AI Integration Opportunity]

Yet the war has inflicted wounds that statistics struggle to convey. The World Health Organization documented over 1,500 attacks on healthcare facilities by February 2024. Researchers verified 334 confirmed attacks on 267 healthcare facilities in just the first year of full-scale conflict. The Ukrainian Cabinet of Ministers reports 211 medical facilities completely destroyed and over 1,700 damaged—including diagnostic imaging departments housing irreplaceable CT and MRI equipment. Frontline regions—Kharkiv, Donetsk, Luhansk, Zaporizhzhia, and Kherson oblasts—bear disproportionate infrastructure losses while simultaneously facing elevated diagnostic demands from trauma casualties.

Understanding this infrastructure landscape—its pre-war baseline, wartime degradation, regional variations, and reconstruction trajectory—is essential for designing realistic AI implementation strategies. This article provides that foundational analysis, examining equipment density and distribution, digital infrastructure and standards compliance, workforce considerations, and actionable pathways for AI integration within Ukrainian medical imaging.

2. Literature Review

2.1 Global Medical Imaging Equipment Density Standards

International comparisons provide crucial context for assessing Ukrainian imaging infrastructure. According to OECD Health Statistics 2024, CT scanner density varies dramatically across developed economies. Japan leads globally with 115.7 CT scanners per million population, followed by Australia at 69.6 per million. Most Western European nations cluster between 20-40 units per million: Germany at 35.3, Italy at 34.8, and the United Kingdom at 9.5 per million reflecting its more centralized National Health Service approach.

Country	CT Scanners per Million	MRI Units per Million	Year
Japan	115.7	57.4	2021
Australia	69.6	16.0	2021
Germany	35.3	34.7	2021
Italy	34.8	31.6	2021
Poland	17.4	11.2	2021
Ukraine (estimated pre-war)	7.2	3.8	2021
United Kingdom	9.5	8.0	2021

MRI equipment density follows similar patterns, with Japan again leading at 57.4 units per million, Germany at 34.7, and the United States at approximately 40 per million. The European average hovers around 15-25 MRI units per million population. These benchmarks highlight that achieving EU-candidate country standards would require Ukraine to approximately triple its CT capacity and quadruple MRI availability—a challenge complicated by wartime destruction.

2.2 The Ukrainian eHealth System Architecture

Ukraine’s digital health transformation, while interrupted by conflict, established critical infrastructure for future AI integration. The electronic healthcare system architecture, as documented by Malakhov et al. (2023) in their comprehensive PMC analysis, operates on a two-level structure: a centralized national database managed by the State Enterprise “eHealth” and approximately 40 regional Medical Information Systems (MIS) developed by private IT companies addressing local needs.

flowchart TD
    subgraph National["National Level"]
        A[State Enterprise eHealth]
        B[Central Database]
        C[API Gateway]
    end
    
    subgraph Regional["Regional MIS (40 systems)"]
        D[Kyiv MIS]
        E[Odesa MIS]
        F[Lviv MIS]
        G[Other Regional]
    end
    
    subgraph Healthcare["Healthcare Facilities"]
        H[Hospitals]
        I[Clinics]
        J[Diagnostic Centers]
    end
    
    A --> B
    B --> C
    C --> D & E & F & G
    D & E & F & G --> H & I & J

This architecture offers both advantages and challenges for AI imaging deployment. The centralized component ensures standardized data formats and interoperability across regional systems—essential for training machine learning models on diverse patient populations. By 2022, the system handled enormous transaction volumes, demonstrating scalability that many healthcare IT infrastructures lack. However, the war decimated the MIS ecosystem: of 40 operational systems pre-war, only 4-6 are projected to survive, creating workflow disruption as medical staff trained on defunct systems must migrate to survivors.

2.3 International Experience with Post-Conflict Healthcare Reconstruction

Historical precedents inform expectations for Ukrainian healthcare reconstruction. The WHO’s Health as the Bridge for Peace initiative documented healthcare rebuilding in Bosnia-Herzegovina, Kosovo, and other post-conflict settings, identifying equipment procurement, workforce retention, and supply chain resilience as interconnected challenges. Notably, these cases demonstrate that conflict often accelerates digital transformation—displaced healthcare workers adopt telemedicine, centralized systems replace fragmented local records, and international donors prioritize modern equipment installation.

The RAD-AID International organization’s work in Ukraine since 2015 provides directly relevant experience. Their Radiology-Readiness Assessment and PACS-Readiness Assessment conducted in 2022 identified low-resource Ukrainian health institutions struggling with infrastructural damage. Their 2024 agreement to support PACS and IT implementation at Kyiv City Clinical Hospital #18 represents a model for combining international expertise with local capacity building—including plans for AI integration by 2025.

2.4 AI Readiness Frameworks for Healthcare Systems

Assessing AI readiness requires examining multiple infrastructure dimensions beyond equipment counts. The WHO’s 2021 guidance on AI for health identifies six foundational requirements: digital infrastructure and connectivity, data governance frameworks, workforce digital literacy, regulatory environments, ethical oversight mechanisms, and interoperability standards. Scholarly work by Rajpurkar et al. on radiology AI deployment emphasizes additional technical requirements: DICOM-compliant imaging equipment, PACS connectivity for workflow integration, and sufficient bandwidth for cloud-based AI processing.

For Ukraine, each dimension presents specific challenges. DICOM (Digital Imaging and Communications in Medicine) compliance varies significantly across equipment vintage—older Soviet-era machines and early post-Soviet acquisitions often lack digital output capabilities entirely. PACS adoption, estimated at 15-25% of Ukrainian hospitals pre-war, lags far behind Western European levels of 70-90%. Meanwhile, internet connectivity in rural areas and conflict-affected regions may not support real-time AI analysis requiring significant data transmission.

3. Methodology

3.1 Data Sources and Assessment Framework

This infrastructure assessment synthesizes data from multiple sources given the absence of comprehensive, current Ukrainian medical equipment registries. Primary sources include:

graph LR
    A[WHO Statistics] --> E[Synthesis]
    B[NHSU Reports] --> E
    C[Academic Studies] --> E
    D[RAD-AID Assessments] --> E
    E --> F[Infrastructure Analysis]
    F --> G[AI Readiness Score]

• World Health Organization Global Health Observatory: Medical equipment density statistics through 2021, providing pre-war baseline
• Ukrainian Ministry of Health and NHSU publications: eHealth system statistics, healthcare reform documentation, and facility reports
• Peer-reviewed academic literature: Including PMC-indexed studies on Ukrainian digital health (Malakhov et al. 2023, Mudge et al. 2025)
• International organization reports: USAID LHSS telemedicine assessment (September-November 2022), WHO attack documentation, RAD-AID readiness assessments
• Ukrainian Cabinet of Ministers announcements: Facility damage statistics and reconstruction initiatives
• News reports and humanitarian organization documentation: Verified incident reports from Physicians for Human Rights, IOM Ukraine, and Ukrainian Healthcare Center

3.2 Regional Analysis Approach

Ukraine’s 24 oblasts (regions) plus Kyiv city exhibit substantial heterogeneity in healthcare infrastructure, economic development, and conflict exposure. Our analysis categorizes regions into five tiers based on AI readiness potential:

graph TD
    subgraph Tier1["Tier 1: Major Hubs"]
        A[Kyiv City]
        B[Kyiv Oblast]
    end
    
    subgraph Tier2["Tier 2: Regional Centers"]
        C[Lviv]
        D[Odesa]
        E[Dnipro]
        F[Kharkiv]
    end
    
    subgraph Tier3["Tier 3: Secondary Centers"]
        G[Poltava]
        H[Vinnytsia]
        I[Zaporizhzhia]
    end
    
    subgraph Tier4["Tier 4: Rural/Underserved"]
        J[Western Oblasts]
        K[Central Oblasts]
    end
    
    subgraph Tier5["Tier 5: Conflict-Affected"]
        L[Kharkiv Frontline]
        M[Donetsk]
        N[Zaporizhzhia South]
    end
    
    Tier1 -->|High Readiness| AI[AI Implementation]
    Tier2 -->|Medium-High| AI
    Tier3 -->|Medium| AI
    Tier4 -->|Low| AI
    Tier5 -->|Critical Need| AI

Tier	Regions	Characteristics	AI Readiness
Tier 1: Major Hubs	Kyiv City, Kyiv Oblast	Highest equipment density, digital infrastructure, specialist concentration	High
Tier 2: Regional Centers	Lviv, Odesa, Dnipro, Kharkiv	Strong university hospitals, growing private sector, variable war impact	Medium-High
Tier 3: Secondary Centers	Poltava, Vinnytsia, Zaporizhzhia, Chernivtsi	Regional hospitals with imaging, limited PACS, telemedicine emerging	Medium
Tier 4: Rural/Underserved	Western and Central oblasts	Basic imaging only, poor connectivity, workforce shortages	Low
Tier 5: Conflict-Affected	Kharkiv frontline, Donetsk, Luhansk, Kherson, Zaporizhzhia south	Severe infrastructure damage, displaced populations, highest trauma burden	Critical Need

4. Current Infrastructure Assessment

4.1 Equipment Density and Distribution

Pre-war estimates placed Ukraine’s CT scanner density at approximately 7.2 units per million population—roughly one-third of the European Union average. MRI density stood even lower at 3.8 units per million. However, these national figures mask significant regional disparities: Kyiv city enjoyed equipment density approaching Polish levels (15-17 CT units per million city population), while rural western oblasts operated below Sub-Saharan African averages in some districts.

War-related destruction has further skewed this distribution. Kharkiv Oblast, which housed Ukraine’s second-largest imaging infrastructure pre-war, lost an estimated 40-60% of operational capacity through direct attacks and equipment evacuation. Mariupol’s destruction eliminated all advanced imaging capability in the city. Meanwhile, western Ukraine—particularly Lviv—has experienced equipment concentration as facilities relocated from conflict zones.

4.2 Digital Infrastructure and PACS Adoption

Picture Archiving and Communication Systems (PACS) represent the critical bridge between imaging equipment and AI integration. PACS enable digital storage, retrieval, and transmission of medical images—prerequisites for AI algorithm deployment. Pre-war estimates suggested 15-25% of Ukrainian hospitals had functional PACS, concentrated in Kyiv, Lviv, Odesa, and private facilities.

This low adoption rate creates both challenges and opportunities. The challenge: most existing images remain inaccessible for AI analysis, stored on local workstations or film. The opportunity: greenfield PACS deployments can incorporate AI-ready architectures from inception rather than retrofitting legacy systems.

4.3 Workforce Considerations

Ukraine’s radiologist density of approximately 3.2 per 100,000 population falls below WHO recommendations but exceeds many middle-income countries. The war has disrupted this workforce through internal displacement, emigration, and—most critically—concentration of specialists in safer western regions while eastern and southern areas face acute shortages. AI-assisted diagnosis could help address this maldistribution by enabling remote specialist oversight of AI-flagged cases from underserved regions.

5. Strategic Framework for AI Integration

Given the infrastructure assessment, we propose a three-phase AI integration strategy tailored to Ukrainian realities:

gantt
    title Ukrainian Medical AI Integration Roadmap
    dateFormat  YYYY-MM
    section Phase 1
    PACS Expansion           :2026-01, 12M
    Connectivity Upgrade     :2026-01, 18M
    Workforce Training       :2026-03, 15M
    section Phase 2
    Pilot AI Deployment      :2027-01, 12M
    Validation Studies       :2027-06, 12M
    Regulatory Framework     :2027-01, 18M
    section Phase 3
    National Rollout         :2028-01, 24M
    Quality Monitoring       :2028-06, ongoing
    Continuous Improvement   :2028-12, ongoing

5.1 Phase 1: Foundation Building (2026-2027)

• Expand PACS deployment to 50% of imaging-capable facilities
• Establish DICOM compliance requirements for new equipment procurement
• Deploy edge computing nodes enabling offline AI inference
• Train radiologists and technicians on AI-assisted workflows

5.2 Phase 2: Pilot Implementation (2027-2028)

• Deploy validated AI algorithms in Tier 1 and Tier 2 facilities
• Conduct prospective validation studies comparing AI-assisted vs. standard diagnosis
• Develop regulatory framework aligned with EU Medical Device Regulation
• Establish telemedicine links connecting Tier 5 facilities with specialist oversight

5.3 Phase 3: National Rollout (2028-2030)

• Extend AI deployment to Tier 3-5 facilities with appropriate infrastructure
• Integrate AI results into eHealth system for longitudinal tracking
• Establish continuous quality monitoring and algorithm update protocols
• Develop Ukrainian AI training datasets addressing local disease patterns

6. Conclusions

Ukraine’s medical imaging infrastructure faces unprecedented challenges from war-related destruction compounding decades of underinvestment. Yet the nation’s digital health foundation—particularly the eHealth system’s scale and interoperability—provides a platform for AI integration that many higher-income countries lack. Success requires realistic assessment of regional capabilities, prioritized investment in PACS and connectivity infrastructure, and AI deployment strategies adapted to intermittent connectivity and workforce constraints.

The path forward is neither simple nor guaranteed, but the potential for AI-assisted medical imaging to extend specialist capacity to underserved regions—including conflict-affected areas with the highest diagnostic needs—justifies sustained investment and international partnership. This infrastructure assessment provides the foundation for targeted interventions that can transform Ukrainian medical imaging capabilities within the reconstruction timeline.

---

References

1. World Health Organization. Global Health Observatory Data Repository. 2024.
2. Malakhov et al. Ukrainian eHealth System Architecture. PMC. 2023.
3. OECD Health Statistics. Medical Equipment Density. 2024.
4. RAD-AID International. Ukraine Radiology-Readiness Assessment. 2022.
5. Ukrainian Ministry of Health. eHealth System Statistics. 2022.
6. USAID LHSS. Ukraine Telemedicine Assessment. 2022.
7. WHO. AI for Health Guidance. 2021.
8. Rajpurkar et al. Radiology AI Deployment Considerations. Nature Medicine. 2022.