Artificial Intelligence Integration in Radiology: Workforce Readiness and Implementation Challenges for Sustainable Healthcare Development

1. Introduction

Radiology is one of the most technologically advanced fields in modern healthcare, playing a crucial role in disease diagnosis, treatment planning, image-guided interventions, and patient monitoring [2,7]. The evolution of imaging technology has significantly enhanced healthcare outcomes and broadened diagnostic capabilities in recent decades. However, the rapid increase in imaging utilization has also resulted in challenges for healthcare sustainability, including escalating operational costs, heightened energy consumption, expanding digital storage needs, workforce pressures, and growing healthcare infrastructure demands [2,7,8,9,10]. These challenges have driven healthcare systems worldwide to explore innovative strategies to enhance efficiency while upholding quality, accessibility, and long-term resilience [1,2,3,4,5,6]. Sustainable healthcare encompasses economic viability, workforce resilience, technological adaptability, digital governance, and equitable access to healthcare services, extending beyond environmental considerations [1,2,3,4,5,6,7,8]. Within radiology, sustainability has become a strategic priority because imaging departments are among the most resource-intensive components of healthcare systems [7,8,9,10]. Consequently, healthcare organizations are seeking solutions to optimize resource utilization, enhance operational performance, reduce waste, and support long-term healthcare transformation [7,8,9,10].

The integration of artificial intelligence (AI) into radiology is increasingly recognized as a strategic pathway toward sustainable healthcare development [2,3,4,5,6,11,12,13,14,15,16]. Through advances in machine learning, deep learning, and computer vision, AI technologies can assist image interpretation, automate repetitive tasks, streamline workflow management, improve quality assurance processes, and facilitate clinical decision-making [11,12,14,31,32]. Numerous studies have demonstrated the potential of AI for lesion detection, image segmentation, diagnostic support, workflow prioritization, and predictive analytics, thereby improving both clinical performance and operational efficiency [11,12,13,14,15,16,31,32]. These capabilities have generated substantial interest among healthcare organizations seeking to enhance healthcare quality while improving resource utilization and sustainability [7,8,9,10,15,16].

Beyond its direct clinical applications, AI integration is increasingly being explored within the broader framework of healthcare sustainability [8,9,10,17,18,35,36,37]. AI-driven workflow optimization may reduce unnecessary imaging examinations, minimize redundant studies, improve scheduling efficiency, optimize resource allocation, and support evidence-based service planning [8,12,17,18]. Furthermore, emerging reporting and governance frameworks for AI in healthcare and medical imaging have established standards for the responsible development, evaluation, and implementation of AI technologies, promoting transparency and facilitating sustainable clinical adoption [19,20,21,22,23]. Collectively, these advances have the potential to strengthen healthcare system resilience while promoting more effective utilization of available resources.

This perspective closely aligns with the United Nations Sustainable Development Goals (SDGs), particularly SDG 3 (Good Health and Well-Being), SDG 9 (Industry, Innovation and Infrastructure), SDG 12 (Responsible Consumption and Production), and SDG 13 (Climate Action) [1,2,3,4]. AI integration into radiology may improve diagnostic quality and healthcare accessibility, thereby supporting SDG 3, while simultaneously strengthening digital health infrastructure and technological innovation consistent with SDG 9 [1,2,39,40]. More efficient resource utilization and waste reduction may contribute to SDG 12, whereas improvements in operational efficiency could indirectly support climate-related objectives under SDG 13 [7,8,9,10,17,18,38].

However, successful implementation of AI is not guaranteed. It requires robust digital infrastructure, interoperability frameworks, cybersecurity systems, governance policies, technical expertise, and continuous workforce development [3,4,5,6,15,16,24,25,26,27,28,29,30]. AI systems require substantial computational resources and may introduce organizational, ethical, regulatory, and operational challenges if implementation is inadequately planned [3,4,5,6,13,15,16]. Consequently, AI integration presents both opportunities and challenges for sustainable healthcare development. Contemporary implementation science suggests that successful healthcare innovation depends not only on technological performance but also on sociotechnical factors, including organizational readiness, governance structures, stakeholder engagement, workforce capability, and institutional support [13,15,16,35,36,37].

In radiology, healthcare professionals serve as the primary users, supervisors, and operational stakeholders responsible for integrating AI technologies into routine clinical practice. Their perceptions, expectations, and concerns therefore substantially influence implementation success. Previous studies indicate that healthcare professionals generally have positive attitudes toward AI but remain concerned about data privacy, cybersecurity, ethical accountability, regulatory uncertainty, workflow integration, and professional responsibility [3,4,13,14,15,16,24,25,26,27,28,29,30]. These findings suggest that positive perceptions alone are insufficient for successful implementation. Sustainable adoption requires alignment between technological capability, workforce preparedness, organizational capacity, and governance frameworks [15,16,35,36,37].

Saudi Arabia provides an important setting for examining AI integration in radiology. Through Vision 2030 and national healthcare transformation initiatives, the Kingdom has prioritized healthcare modernization, digital innovation, artificial intelligence, and technology-driven development [39,40]. Significant investments have been directed toward digital health infrastructure, smart healthcare systems, and advanced clinical technologies [39,40,41,42]. Although these developments provide substantial opportunities for AI adoption, empirical evidence regarding workforce readiness, organizational capacity, and implementation challenges remains limited. Most existing studies have focused primarily on algorithm development, technical validation, diagnostic accuracy, and performance benchmarking rather than implementation science and sustainability outcomes [11,12,31,32,33,34]. Furthermore, the relationship between AI integration and sustainable healthcare development within rapidly evolving Middle Eastern healthcare systems remains underexplored [35,36,37,38,39,40,41,42].

Understanding workforce readiness and implementation challenges is essential because successful adoption depends on institutional preparedness, governance confidence, technical competence, and stakeholder acceptance. Positive attitudes toward AI may facilitate adoption; however, sustainable implementation requires practical organizational capacity and effective implementation strategies [15,16,35,36,37]. Therefore, this study aimed to assess workforce readiness, organizational capacity, and implementation challenges associated with integrating artificial intelligence into radiology practice. By evaluating knowledge, attitudes, implementation readiness, and perceived barriers, this study sought to generate evidence to support policy development and facilitate the sustainable integration of AI into radiology services.

Figure 1. Conceptual framework illustrating workforce readiness, organizational capacity, implementation challenges, and sustainable healthcare outcomes associated with artificial intelligence integration in radiology.

Figure 1. Conceptual framework illustrating workforce readiness, organizational capacity, implementation challenges, and sustainable healthcare outcomes associated with artificial intelligence integration in radiology.

2. Materials and Methods

2.1. Study Design and Setting

This study used a cross-sectional survey design to evaluate workforce readiness, organizational capacity, and implementation challenges associated with AI integration in radiology practice. The research was conducted in healthcare institutions across Saudi Arabia, focusing on healthcare professionals involved in radiology services, imaging operations, healthcare administration, medical physics, and AI-supported clinical activities. The cross-sectional approach was chosen for its effectiveness in evaluating perceptions, readiness for implementation, and organizational barriers among various professional groups within a specific timeframe. The study was conducted within the context of digital healthcare transformation and sustainable healthcare development, with a specific focus on understanding the factors influencing successful AI integration in radiology. It was executed in line with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for cross-sectional studies.

2.2. Study Population and Participant Recruitment

The study focused on healthcare professionals in radiology services in Saudi Arabia: radiologists, radiologic technologists, medical physicists, healthcare administrators, AI and data science professionals, and other specialists.

To be eligible for participation, respondents were required to:

• be currently practicing or employed in a radiology-related healthcare setting;
• possess familiarity with radiology workflows, imaging services, or healthcare technologies; and
• provide voluntary informed consent.

The exclusion criteria included the following:

• students and trainees without professional responsibilities;
• incomplete questionnaire submissions;
• duplicate responses; and
• individuals without professional involvement in radiology-related services.

Participants were recruited using a convenience-based electronic sampling strategy. The survey was conducted through professional communication networks, institutional mailing lists, professional radiology professional groups, and online platforms. Participation was voluntary, anonymous, and uncompensated.

Following data cleaning, 295 valid questionnaires were retained for analysis. Among the participants, 203 (68.8%) were male, and 92 (31.2%) were female. The respondents represent various healthcare sectors and geographic regions across Saudi Arabia, indicating the wide applicability of our findings.

2.3. Survey Instrument Development and Validation

Data were collected using a structured, self-administered questionnaire designed to evaluate the perceptions of AI integration in radiology. Moreover, the data collection aids in understanding potential contributions of AI in sustainable healthcare development.

Questionnaire development was informed by an extensive review of contemporary literature on the following topics:

• AI in healthcare;
• AI in radiology;
• digital transformation and healthcare innovation;
• organizational readiness for technological change; and
• sustainable healthcare systems and implementation science.

The final questionnaire consisted of 30 items distributed across four conceptual domains:

• Knowledge of Artificial Intelligence and Healthcare Sustainability (7 items)
• Attitudes Toward AI Integration in Radiology (7 items)
• Implementation Readiness and Organizational Capacity (8 items)
• Perceived Barriers and Implementation Challenges (8 items)

All items were evaluated using a five-point Likert scale:

• 1 = Strongly disagree
• 2 = Disagree
• 3 = Neutral
• 4 = Agree
• 5 = Strongly agree

To ensure content validity, the preliminary questionnaire was reviewed by ten experts. Their feedback was used to enhance item wording, clarity, and alignment with the study objectives. A pilot then assessed usability, comprehensibility, and completion time, after which minor adjustments were made before final distribution. The comprehensive questionnaire and domain allocations can be found in Supplementary Table S2.

2.4. Reliability Assessment

Internal consistency reliability was evaluated using Cronbach’s alpha coefficients. The questionnaire as a whole exhibited outstanding reliability, achieving a Cronbach’s alpha value of 0.89.

Domain-specific reliability coefficients were:

• Knowledge: α = 0.89
• Attitudes: α = 0.86
• Implementation Readiness: α = 0.80
• Perceived Barriers: α = 0.76

All values exceeded the recommended threshold of 0.70, indicating acceptable to excellent internal consistency and supporting the instrument’s reliability in assessing workforce readiness and implementation challenges associated with AI integration in radiology.

2.5. Ethical Considerations

The study was conducted following the ethical principles outlined in the Declaration of Helsinki and the relevant institutional research governance requirements. Ethical approval was obtained from the Research Ethics Committee (MUREC) of Majmaah University, Saudi Arabia (Institutional Registration No. HA-01-R-088, approval no. MUREC-Feb.16/COM-2026/362).

Prior to participation, all respondents received an electronic information sheet describing:

• the objectives of the study;
• voluntary participation requirements;
• confidentiality safeguards;
• data protection procedures; and
• the right to withdraw at any time without penalty.

Informed consent was obtained from all participants before they were granted access to the questionnaire. No personally identifiable information was collected, and all responses were anonymized before analysis.

2.6. Statistical Analysis

Statistical analyses were performed using IBM SPSS Statistics version 30.0 (IBM Corp., Armonk, NY, USA). Descriptive statistics summarized participant characteristics. Continuous variables were presented as means ± standard deviation (SD) . The normality of questionnaire domain scores was assessed using the Shapiro–Wilk test. Several variables departed significantly from normality (p < 0.05), supporting nonparametric methods. Professional groups were compared using the Kruskal–Wallis H test, reflecting the ordinal nature of Likert-scale data and observed distribution characteristics. Associations among the principal questionnaire domains were evaluated using Spearman’s rank correlation analysis. Exploratory factor analysis (EFA) was performed. The dataset’s suitability for factor analysis was assessed using the Kaiser–Meyer–Olkin (KMO) measure and Bartlett’s test of sphericity. Factor extraction was performed using principal component analysis with varimax rotation. Statistical significance was set at a two-sided p-value < 0.05.

2.7. Conceptual Framework

The study’s conceptual framework was based on a sociotechnical implementation perspective, acknowledging that effective AI integration in radiology relies on the interplay of technological capability, workforce readiness, organizational capacity, governance structures, and digital infrastructure.

The framework assumed the following:

• Greater knowledge of AI would be associated with more favorable attitudes toward integration.
• Positive attitudes would contribute to stronger implementation readiness.
• Organizational barriers would influence implementation capacity.
• Sustainable AI integration necessitates aligning technology, governance, infrastructure, and workforce readiness.

Workforce readiness comprises knowledge, attitudes, organizational preparedness, and perceptions of implementation challenges. These factors collectively affect the likelihood of successful and sustainable AI integration in radiology. The framework guided the development, analysis, and interpretation of the questionnaire within the context of sustainable healthcare development.

2.7. Bias Reduction

Several measures were implemented to reduce potential sources of bias. Anonymous participation minimized social desirability effects, duplicate responses were excluded during data cleaning, and standardized electronic data collection reduced interviewer bias. Expert review and pilot testing improved content validity and questionnaire clarity.

2.9. Clinical Trial Registration

Clinical trial number: not applicable.

This study was observational and did not include clinical intervention, patient randomization, or experimental treatment allocation. Therefore, clinical trial registration is not necessary.

3. Results

3.1. Participant Characteristics

In total, 295 valid responses were included in the final analysis. The study population consisted of a multidisciplinary radiology workforce involved in clinical imaging, healthcare administration, medical physics, and digital health innovations. Among the participants, 203 (68.8%) were male, and 92 (31.2%) were female. Radiologists constituted the largest professional category (n = 97, 32.9%), followed by radiographers (n = 89, 30.2%). Medical physicists accounted for 36 participants (12.2%), while healthcare administrators and AI/data science professionals represented 29 participants (9.8%). Other radiology-related professionals comprised 15 (5.1%) participants. subgroup had 5–10 years (34.2%), followed by those with less than 5 years (25.8%), 11–15 years (20.3%), and more than 15 years (19.7%). Educational attainment was predominantly at the bachelor’s degree level (53.2%), followed by master’s degree holders (34.6%), professional certification holders (8.1%), and doctorate level participants (4.1%). Participants were recruited from all major regions of Saudi Arabia, with the highest representation originating from the Central Region (42.4%), followed by the Western Region (21.4%), Eastern Region (17.3%), Southern Region (11.5%), and Northern Region (7.4%) (Appendix Table A1).

Table 1. Demographic and Professional Characteristics of Participants (n = 295).

Variable	Category	n (%)
Gender	Male	203 (68.8)
	Female	92 (31.2)
Professional Role	Radiologist	97 (32.9)
	Radiologic Technologist	89 (30.2)
	Medical Physicist	36 (12.2)
	Healthcare Administrator	29 (9.8)
	AI/Data Scientist	29 (9.8)
	Other	15 (5.1)
Years of Experience	<5 years	76 (25.8)
	5–10 years	101 (34.2)
	11–15 years	60 (20.3)
	>15 years	58 (19.7)
Education Level	Bachelor’s degree	157 (53.2)
	Master’s degree	102 (34.6)
	Professional certification	24 (8.1)
	Doctorate	12 (4.1)

Table 1. Demographic and Professional Characteristics of Participants (n = 295).

Variable	Category	n (%)
Gender	Male	203 (68.8)
	Female	92 (31.2)
Professional Role	Radiologist	97 (32.9)
	Radiologic Technologist	89 (30.2)
	Medical Physicist	36 (12.2)
	Healthcare Administrator	29 (9.8)
	AI/Data Scientist	29 (9.8)
	Other	15 (5.1)
Years of Experience	<5 years	76 (25.8)
	5–10 years	101 (34.2)
	11–15 years	60 (20.3)
	>15 years	58 (19.7)
Education Level	Bachelor’s degree	157 (53.2)
	Master’s degree	102 (34.6)
	Professional certification	24 (8.1)
	Doctorate	12 (4.1)

Table 2. Mean Scores Across Principal Study Domains.

Domain	Mean ± SD
nowledge	3.35 ± 0.70
Attitudes	3.57 ± 0.69
Implementation Readiness	3.04 ± 0.79
Perceived Barriers	3.67 ± 0.64

Table 2. Mean Scores Across Principal Study Domains.

Domain	Mean ± SD
nowledge	3.35 ± 0.70
Attitudes	3.57 ± 0.69
Implementation Readiness	3.04 ± 0.79
Perceived Barriers	3.67 ± 0.64

Figure 2. Radar plot comparing questionnaire domain mean scores.

Figure 2. Radar plot comparing questionnaire domain mean scores.

Figure 2 shows mean scores across the four domains. Perceived barriers exhibited the highest overall score, followed by attitudes toward AI integration. On the other hand, implementation readiness had the lowest overall score, indicating a gap between perceptions and organizational readiness. To further explore participant responses, descriptive statistics for each of the 30 questionnaire items were analyzed. Overall, respondents expressed positive views on the potential benefits of AI in terms of diagnostic accuracy, workflow efficiency, and healthcare sustainability. However, lower scores were noted for institutional readiness, digital infrastructure availability, and implementation support. Item-level analysis demonstrated generally positive perceptions regarding the potential benefits of AI in radiology, particularly for diagnostic accuracy (Q8; mean = 3.55 ± 0.91), workflow efficiency (Q9; mean = 3.68 ± 0.83), and healthcare sustainability (Q10; mean = 3.61 ± 0.86). In contrast, implementation-related items revealed lower scores, particularly regarding digital infrastructure availability (Q16; mean = 2.97 ± 1.15) and technical support (Q20; mean = 3.03 ± 1.18). Among the barriers identified, implementation cost (Q23; mean = 3.98 ± 0.71) and limited digital infrastructure (Q24; mean = 3.90 ± 0.75) were highlighted as the most significant obstacles to AI adoption (Appendix Table A3).

3.3. Barrier Analysis

A thorough analysis of the implementation barriers revealed significant organizational and operational constraints impacting AI integration in radiology. The most prominent barrier was the high implementation cost (3.98 ± 0.71), with limited digital infrastructure (3.90 ± 0.75), inadequate staff training (3.84 ± 0.77), and lack of technical expertise (3.82 ± 0.78).

Governance-related concerns were prominent, including unclear governance policies (3.79 ± 0.74), data privacy issues (3.75 ± 0.76), cybersecurity risks (3.72 ± 0.79), and ethical accountability concerns (3.69 ± 0.80). Shapiro–Wilk tests revealed significant deviations from normality for all barrier variables (p < 0.001), justifying the application of nonparametric analyses.

Table 3. Barrier Analysis for AI Integration in Radiology.

Barrier	Mean ± SD	Median	Shapiro–Wilk p
High implementation costs	3.98 ± 0.71	4.0	<0.001
Limited digital infrastructure	3.90 ± 0.75	4.0	<0.001
Insufficient staff training	3.84 ± 0.77	4.0	<0.001
Lack of technical expertise	3.82 ± 0.78	4.0	<0.001
Governance uncertainty	3.79 ± 0.74	4.0	<0.001
Data privacy concerns	3.75 ± 0.76	4.0	<0.001
Cybersecurity risks	3.72 ± 0.79	4.0	<0.001
Ethical accountability concerns	3.69 ± 0.80	4.0	<0.001
Regulatory uncertainty	3.66 ± 0.78	4.0	<0.001
Integration challenges	3.64 ± 0.82	4.0	<0.001

Table 3. Barrier Analysis for AI Integration in Radiology.

Barrier	Mean ± SD	Median	Shapiro–Wilk p
High implementation costs	3.98 ± 0.71	4.0	<0.001
Limited digital infrastructure	3.90 ± 0.75	4.0	<0.001
Insufficient staff training	3.84 ± 0.77	4.0	<0.001
Lack of technical expertise	3.82 ± 0.78	4.0	<0.001
Governance uncertainty	3.79 ± 0.74	4.0	<0.001
Data privacy concerns	3.75 ± 0.76	4.0	<0.001
Cybersecurity risks	3.72 ± 0.79	4.0	<0.001
Ethical accountability concerns	3.69 ± 0.80	4.0	<0.001
Regulatory uncertainty	3.66 ± 0.78	4.0	<0.001
Integration challenges	3.64 ± 0.82	4.0	<0.001

Figure 3. Heatmap of Spearman correlation matrix between major study domains.

Figure 3. Heatmap of Spearman correlation matrix between major study domains.

Exploratory factor analysis was performed to identify latent organizational dimensions affecting implementation readiness. The Kaiser–Meyer–Olkin measure indicated satisfactory sampling adequacy (KMO = 0.73), and Bartlett’s test of sphericity validated the suitability of the dataset for factor analysis (χ² = 895.4; df = 105; p < 0.001).

The explanatory factory analysis identified four principal latent dimensions:

• Governance, Privacy, and Trust
• Workforce Capability and Organizational Adaptation
• Infrastructure and Systems Integration
• Implementation Readiness and Operational Evidence

The factor-loading matrix for all extracted components is presented in Appendix Table A4. Together, these dimensions explain 76.7% of the total variance, offering a comprehensive framework for comprehending the multidimensional nature of AI integration challenges in radiology. Eigenvalues and variance explained by each extracted factor are provided in Appendix Table A5.

Table 4. Factorability Assessment.

Metric	Value
KMO	0.73
Bartlett’s χ2	895.4
Degrees of freedom	105
p-value	<0.001

Table 4. Factorability Assessment.

Metric	Value
KMO	0.73
Bartlett’s χ2	895.4
Degrees of freedom	105
p-value	<0.001

Additional factorability statistics, including KMO and Bartlett’s test results, are presented in Appendix Table S6.

Table 5. Latent Organizational Dimensions Identified Through EFA.

Factor	Theme
Factor 1	Governance, Privacy and Trust
Factor 2	Workforce Capability and Organizational Adaptation
Factor 3	Infrastructure and Systems Integration
Factor 4	Implementation Readiness and Operational Evidence

Table 5. Latent Organizational Dimensions Identified Through EFA.

Factor	Theme
Factor 1	Governance, Privacy and Trust
Factor 2	Workforce Capability and Organizational Adaptation
Factor 3	Infrastructure and Systems Integration
Factor 4	Implementation Readiness and Operational Evidence

Detailed factor loadings, eigenvalues, and variance explained are provided in Appendix Table A4. The four-factor solution accounted for 62.0% of the total variance and exhibited acceptable construct validity, supported by a KMO value of 0.73 and a significant Bartlett’s test of sphericity (p < 0.001).

3.5. Reliability Analysis

The internal consistency analysis showed strong psychometric reliability across all questionnaire domains. Cronbach’s alpha coefficients ranged from 0.76 to 0.89, indicating acceptable to excellent reliability.

Table 6. Reliability Assessment.

Domain	Cronbach’s α
Knowledge	0.89
Attitudes	0.86
Implementation Readiness	0.80
Perceived Barriers	0.76
Overall Instrument	0.89

Table 6. Reliability Assessment.

Domain	Cronbach’s α
Knowledge	0.89
Attitudes	0.86
Implementation Readiness	0.80
Perceived Barriers	0.76
Overall Instrument	0.89

The findings support the questionnaire’s reliability in assessing workforce readiness and challenges related to AI integration in radiology.

3.6. Correlation Analysis

Significant positive associations were observed among the primary study domains. Knowledge shows a moderately positive correlation with the following:

• Attitudes (r = 0.49)
• Implementation readiness (r = 0.39)
• Perceived barriers (r = 0.29)

The positive correlation between knowledge and perceived barriers indicates heightened awareness among knowledgeable professionals. Attitudes were positively correlated with implementation readiness (r = 0.44) and perceived barriers (r = 0.25). Implementation readiness exhibited the most robust correlation with perceived barriers (r = 0.38), highlighting the effect of organizational constraints on implementation preparedness.

Table 7. Spearman Correlation Matrix.

Domain	Knowledge	Attitudes	Implementation Readiness	Barriers
Knowledge	1.00	0.49	0.39	0.29
Attitudes	0.49	1.00	0.44	0.25
Implementation Readiness	0.39	0.44	1.00	0.38
Barriers	0.29	0.25	0.38	1.00

Table 7. Spearman Correlation Matrix.

Domain	Knowledge	Attitudes	Implementation Readiness	Barriers
Knowledge	1.00	0.49	0.39	0.29
Attitudes	0.49	1.00	0.44	0.25
Implementation Readiness	0.39	0.44	1.00	0.38
Barriers	0.29	0.25	0.38	1.00

The structural relationships among knowledge, attitudes, implementation readiness, and perceived barriers are illustrated in Supplementary Figure S1.

3.7. Professional Group Comparisons

A comparative analysis across professional categories showed similar perceptions of AI integration. Radiologic technologists reported the highest barrier scores (3.75), while medical physicists exhibited the highest readiness scores for implementation (3.13).

However, Kruskal–Wallis testing identified no statistically significant differences among professional groups across the principal study domains (all p > 0.05), indicating a broad consensus regarding the opportunities and challenges associated with AI integration in radiology practice.

Table 8. Domain Scores Across Professional Groups.

Professional Group	Knowledge	Attitudes	Readiness	Barriers
AI/Data Scientist	3.33	3.43	2.97	3.59
Healthcare Administrator	3.33	3.54	2.91	3.69
Medical Physicist	3.39	3.44	3.13	3.63
Other	3.23	3.62	2.99	3.53
Radiologic Technologist	3.41	3.63	3.07	3.75
Radiologist	3.31	3.61	3.05	3.67

Table 8. Domain Scores Across Professional Groups.

Professional Group	Knowledge	Attitudes	Readiness	Barriers
AI/Data Scientist	3.33	3.43	2.97	3.59
Healthcare Administrator	3.33	3.54	2.91	3.69
Medical Physicist	3.39	3.44	3.13	3.63
Other	3.23	3.62	2.99	3.53
Radiologic Technologist	3.41	3.63	3.07	3.75
Radiologist	3.31	3.61	3.05	3.67

Overall, the findings suggest that implementation challenges are widely shared across the radiology ecosystem and are not limited to specific professional categories. This consistency supports the necessity for organization-wide and system-level strategies to facilitate sustainable AI integration in radiology.

4. Discussion

This study presents evidence on workforce readiness and challenges related to the integration of AI in radiology within the context of sustainable healthcare development. The findings indicated a consistent pattern among radiology professionals in Saudi Arabia. Participants exhibited positive attitudes toward AI integration. However, readiness for implementation was limited due to organizational, infrastructural, and governance-related barriers. This gap between acceptance of AI in theory and preparedness for practical implementation highlights the need to address critical barriers to ensure the sustainable integration of AI in radiology practice. A notable discovery was the high attitude scores among participants, indicating that radiology professionals increasingly view AI as a supportive technology that enhances clinical practice rather than a replacement for healthcare professionals. This perception aligns with international reports where radiologists and imaging professionals see AI as a complementary tool that can enhance workflow efficiency, reduce repetitive tasks, aid in image interpretation, and support quality assurance activities. The increased visibility of AI applications in medical imaging, coupled with growing awareness of digital transformation initiatives, likely contributed to these positive perceptions.

In Saudi Arabia, favorable attitudes toward AI integration may also indicate alignment with Vision 2030. The Kingdom has prioritized digital transformation, healthcare innovation, and advanced technologies as strategic drivers of healthcare improvement. Consequently, healthcare professionals may view AI integration as a natural extension of national modernization the ongoing efforts to modernize healthcare delivery and enhance system performance. Thus, the positive attitudes identified in this study mirror global technological trends and national policy priorities.

Despite favorable attitudes, implementation readiness remained significantly lower than the attitude and knowledge scores. This finding suggests that enthusiasm for AI integration has not yet translated into confidence in practical implementation. Similar implementation gaps have been reported in healthcare systems globally, where healthcare professionals often endorse technological innovation while expressing concerns about infrastructure, governance, workforce capacity, and organizational preparedness. These findings underscore the principle that technological acceptance alone is insufficient to guarantee successful implementation. Sustainable integration demands that healthcare organizations have the organizational capacities essential to facilitate long-term adoption.

The study highlighted the significance of barriers to implementation, with high implementation costs identified as the primary obstacle, followed by limitations in digital infrastructure, inadequate staff training, and a lack of technical expertise. These concerns align with those of prior research on healthcare technology adoption and digital transformation initiatives. Integrating AI typically demands considerable investments in software platforms, computational infrastructure, interoperability solutions, cybersecurity systems, technical support services, and workforce development programs. Consequently, organizations may encounter substantial financial and operational challenges during implementation, particularly when resources and digital maturity differ among institutions.

Infrastructure-related barriers warrant attention as they directly affect the feasibility of AI integration. Successful implementation necessitates well-established digital ecosystems that facilitate secure data exchange, high-performance computing, interoperability among healthcare systems, and smooth integration with Picture Archiving and Communication Systems (PACS) and Radiology Information Systems (RIS). Despite significant advancements in healthcare digitalization in Saudi Arabia, the research indicates that differences in institutional readiness continue to be significant factors. Hence, sustainable implementation demands that infrastructure development be treated as a fundamental requirement.

Workforce capability is crucial for readiness. Participants consistently highlighted inadequate training and a shortage of technical expertise as significant barriers to integrating AI. This finding reflects growing global concern regarding AI literacy among healthcare professionals. Successful AI integration necessitates not only proficiency in using software systems but also comprehension of algorithm limitations, performance variability, model biases, validation requirements, ethical concerns, and governance responsibilities. In the absence of sufficient training, healthcare institutions face the potential of deploying technologies that are poorly comprehended, inconsistently utilized, or inadequately supervised.

The significant focus on workforce development highlighted in this study has important implications for sustainable healthcare progress. Workforce readiness is crucial to healthcare sustainability. Therefore, investments in professional education and training may be highly effective strategies for sustainable AI integration in radiology. Governance-related concerns were also prominently discussed by the participants. Concerns regarding privacy, cybersecurity, ethical responsibility, regulatory ambiguity, and governance policies consistently garnered high ratings. These findings align with a growing body of literature emphasizing that AI implementation presents both governance and technological challenges. Successful implementation requires transparent accountability, ethical oversight, validation, cybersecurity, and regulatory clarity. In the field of radiology, governance challenges are particularly critical due to the presence of highly sensitive patient data in imaging datasets and the potential impact of AI-generated results on crucial diagnostic decisions. Consequently, healthcare professionals may hesitate to trust AI without clear guidance on responsibility, accountability, validation, and data protection. These findings highlight the importance of establishing robust governance frameworks that promote innovation and patient safety concurrently.

The exploratory factor analysis provides further insights into the multidimensional nature of implementation readiness. Factor analysis yielded four latent dimensions: governance and trust, workforce capability, infrastructure integration, and implementation readiness. These dimensions support a sociotechnical interpretation of AI integration, emphasizing that successful implementation depends on interactions among technology, people, organizational systems, and governance structures. This finding is particularly important because healthcare innovation initiatives often focus heavily on technological performance while underestimating organizational and human factors. These results suggest that implementation success is more likely to depend on institutional preparedness, governance maturity, infrastructure capacity, and workforce capability than on algorithmic performance. Consequently, healthcare organizations should adopt comprehensive implementation strategies that simultaneously address the technological, organizational, educational, and governance dimensions. The observed correlations among knowledge, attitudes, implementation readiness, and perceived barriers provide further insights into the dynamics of AI integration. Participants with greater AI knowledge reported more positive attitudes and a stronger readiness for implementation. This finding suggests that educational initiatives may play an important role in facilitating AI adoption by improving familiarity with AI technologies and reducing uncertainty regarding their use. Interestingly, knowledge is positively associated with perceived barriers. Rather than indicating resistance, this relationship may reflect a greater awareness among knowledgeable professionals regarding practical implementation challenges. Individuals with a deeper understanding of AI technologies may be better equipped to identify issues in governance, infrastructure, validation, cybersecurity, and organizational preparation. The absence of significant differences among professional groups is notable. Although minor variations were observed across professions, the perceptions remained remarkably consistent among radiologists, radiologic technologists, medical physicists, healthcare administrators, and AI specialists. This suggests that implementation challenges are broadly shared throughout the radiology ecosystem rather than restricted to specific professional groups. This consistency may be advantageous from an implementation perspective because it indicates a common understanding of both opportunities and barriers. From a sustainable healthcare perspective, these findings contribute to the emerging discussion on the role of AI integration in healthcare transformation. Sustainable healthcare encompasses economic efficiency, workforce resilience, technological adaptability, governance maturity, environmental responsibility, and long-term organizational viability. AI integration has the potential to support each of these dimensions by improving workflow efficiency, reducing unnecessary imaging, optimizing resource allocation, supporting evidence-based decision-making, and enhancing operational performance.

However, the findings also show that sustainability benefits are not automatic. Without proper governance, workforce preparation, infrastructure investment, and implementation planning, AI integration can introduce new costs, risks, and operational challenges. Therefore, sustainable implementation requires a careful balance of innovation opportunities with organizational realities. Healthcare institutions must avoid seeing AI as a standalone technological solution. The Saudi Arabian healthcare system is well positioned to benefit from AI integration. However, the findings suggest that achieving sustainable implementation requires continued attention to infrastructure development, workforce education, governance frameworks, cybersecurity preparedness, and organizational change management. National strategies that address these factors simultaneously are likely to yield more successful and sustainable outcomes than technology-focused initiatives. Overall, this study shows that radiology professionals in Saudi Arabia recognize the significant potential of AI integration in supporting healthcare innovation and sustainable healthcare development. Simultaneously, they acknowledge the substantial organizational challenges that must be addressed before these benefits can be fully realized. Bridging the gap between positive perceptions and implementation readiness is essential to translate AI integration into measurable improvements in healthcare quality, efficiency, resilience, and sustainability.

Implications for Sustainable Development and Future Implementation

4.1. Implications for Sustainable Development Goals

The findings have important implications for multiple United Nations Sustainable Development Goals (SDGs), demonstrating how successful AI integration into radiology contributes to broader healthcare sustainability objectives. Although AI is commonly discussed as a technological innovation, this study suggests that its successful implementation is closely tied to workforce readiness, organizational preparedness, governance maturity, and infrastructure development. Therefore, AI integration should be seen as a multidimensional sustainability initiative, not solely a technological one.

4.1.1. SDG 3: Good Health and Well-Being

The favorable attitudes toward AI integration indicated strong professional endorsement. AI-assisted image interpretation, workflow prioritization, automated quality assurance, and decision support systems have the potential to enhance patient outcomes through early diagnosis, efficient service delivery, and enhanced resource allocation. However, the low implementation readiness score suggests that substantial barriers remain before these benefits can be fully realized. In the absence of adequate infrastructure, governance systems, and workforce readiness, healthcare organizations may face challenges in translating technological potential into measurable enhancements in patient care. Therefore, achieving SDG 3 through AI integration in radiology necessitates investment not only in technology but also in organizational and human capacity.

4.1.2. SDG 9: Industry, Innovation and Infrastructure

One of the most notable findings of this study is the significance of infrastructure-related barriers. Participants consistently highlighted digital infrastructure limitations, interoperability challenges, and implementation costs as obstacles to AI integration. These findings are directly relevant to SDG 9, which underscores resilient infrastructure, innovation, and technological advancement. The incorporation of AI into radiology relies on robust digital ecosystems that facilitate secure data exchange, cloud computing, interoperability, high-performance computing, and cybersecurity. Therefore, national investment in digital health infrastructure is essential for sustainable implementation. These findings indicate that enhancing healthcare infrastructure may be among the most crucial steps in enabling widespread AI integration across healthcare institutions.

4.1.3. SDG 12: Responsible Consumption and Production

Radiology departments are resource-intensive, with high energy use, equipment demands, and digital storage requirements. AI integration has the potential to enhance responsible resource utilization by decreasing unnecessary imaging examinations, reducing repeat studies, enhancing workflow efficiency and optimizing equipment use. However, AI implementation consumes resources and may introduce extra computational requirements. Sustainable adoption therefore requires weighing both advantages and costs. Healthcare organizations should aim to maximize operational efficiency while minimizing unnecessary computational waste. Responsible AI integration can aid in achieving SDG 12 by fostering efficient healthcare operations and decreasing resource-intensive practices.

4.1.4. SDG 13: Climate Action

Although climate-related outcomes were not directly measured in this study, AI integration may indirectly contribute to climate-related goals by enhancing operational efficiency and optimizing healthcare resource utilization. This can be achieved through reduced repeat imaging, streamlined workflows, predictive maintenance, and optimized scheduling, which may lead to decreased energy consumption and improved departmental efficiency. However, AI systems require computational infrastructure that consumes energy and may contribute to the environmental burden if implemented without sustainability considerations. Therefore, organizations should adopt environmentally responsible strategies prioritizing efficient computing, optimized algorithms, and sustainable infrastructure. These findings suggest that AI integration can support climate-related objectives when accompanied by appropriate sustainability planning, governance, and environmental oversight.

Table 9. Alignment of Major Implementation Barriers with Sustainable Development Goals.

Barrier Domain	Mean Score	Relevant SDG	Strategic Action
High implementation costs	3.98	SDG 3, SDG 9	Value-based implementation models and public-private partnerships
Limited digital infrastructure	3.90	SDG 9	Investment in interoperable digital health infrastructure
Insufficient staff training	3.84	SDG 4, SDG 3	AI literacy programs and workforce development initiatives
Lack of technical expertise	3.82	SDG 4	Professional certification and continuing education programs
Governance and privacy concerns	3.79	SDG 16	National AI governance and ethical oversight frameworks
Cybersecurity risks	3.72	SDG 9	Enhanced cybersecurity standards and risk-management systems
Regulatory uncertainty	3.66	SDG 16	Development of clear regulatory guidance for healthcare AI

Table 9. Alignment of Major Implementation Barriers with Sustainable Development Goals.

Barrier Domain	Mean Score	Relevant SDG	Strategic Action
High implementation costs	3.98	SDG 3, SDG 9	Value-based implementation models and public-private partnerships
Limited digital infrastructure	3.90	SDG 9	Investment in interoperable digital health infrastructure
Insufficient staff training	3.84	SDG 4, SDG 3	AI literacy programs and workforce development initiatives
Lack of technical expertise	3.82	SDG 4	Professional certification and continuing education programs
Governance and privacy concerns	3.79	SDG 16	National AI governance and ethical oversight frameworks
Cybersecurity risks	3.72	SDG 9	Enhanced cybersecurity standards and risk-management systems
Regulatory uncertainty	3.66	SDG 16	Development of clear regulatory guidance for healthcare AI

4.2. Environmental Sustainability Considerations

Environmental sustainability is increasingly important in healthcare, including radiology. While AI integration is often touted as a means to enhance efficiency, its environmental implications necessitate thorough evaluation. AI systems rely on computational infrastructure, data storage resources, cloud computing services, and ongoing software maintenance. The development, validation, and deployment of large-scale machine learning models may demand significant computational power. Consequently, the environmental impact of AI is influenced by various factors such as model complexity, computing architecture, energy sources, and operational practices. Several studies have indicated that large-scale deep-learning models can produce substantial carbon emissions during training and deployment. However, the environmental impacts can vary significantly based on model design, hardware efficiency, energy sources, and implementation strategies. Therefore, AI sustainability should be assessed not by computational cost alone but through comprehensive evaluation of environmental costs and operational benefits.

Within radiology, AI integration may contribute positively to environmental sustainability by:

• reducing unnecessary imaging examinations;
• minimizing repeat studies caused by workflow inefficiencies;
• enhancing scheduling efficiency;
• supporting predictive maintenance of imaging equipment;
• optimizing resource utilization across imaging departments; and
• reducing waste associated with low-value imaging practices.

These benefits may partially offset the environmental costs associated with computational infrastructure. The findings of the present study suggest that healthcare institutions should adopt a “Green AI” approach that prioritizes efficient algorithms, responsible computing practices, and environmentally conscious implementation strategies. Such approaches may help maximize sustainability benefits while minimizing the environmental burden. Future research should move beyond perception-based assessments to directly evaluate the environmental impact of AI integration in radiology through objective measurements of energy consumption, carbon emissions, and resource utilization.

4.3. Study Limitations

This study has several limitations. First, its cross-sectional design precluded causal inferences about knowledge, attitudes, and perceived barriers related to AI in clinical practice. Longitudinal studies are necessary to track changing perceptions. Second, the reliance on self-reported data introduces potential biases. Third, convenience sampling may have introduced selection bias, potentially misrepresenting the views of all radiology professionals in Saudi Arabia. Fourth, it did not measure objective sustainability indicators. Finally, the Saudi Arabian healthcare system limit the applicability of our findings to other contexts. Despite these limitations, this study offers valuable insights into workforce readiness and challenges of AI integration in radiology.

Table 10. Strategic Recommendations for Sustainable AI Integration in Radiology.

Timeframe	Strategic Action	Responsible Stakeholders	Expected Outcome
Short-term (1–2 years)	Develop national AI literacy curriculum for radiology professionals	Ministry of Health, universities, professional societies	Improved workforce readiness
Short-term (1–2 years)	Establish national AI governance and ethics framework	Regulatory agencies and healthcare authorities	Enhanced trust, transparency, and governance
Medium-term (3–5 years)	Invest in interoperable PACS/RIS infrastructure	Healthcare organizations and government agencies	Improved implementation capacity
Medium-term (3–5 years)	Pilot AI-supported radiology workflows in high-volume centers	Tertiary hospitals and healthcare networks	Improved efficiency and service delivery
Long-term (5–10 years)	Establish national AI validation and monitoring center	National health innovation agencies	Continuous performance monitoring
Long-term (5–10 years)	Incorporate sustainability metrics into AI procurement and evaluation processes	Healthcare regulators and policymakers	Environmentally responsible implementation

Table 10. Strategic Recommendations for Sustainable AI Integration in Radiology.

Timeframe	Strategic Action	Responsible Stakeholders	Expected Outcome
Short-term (1–2 years)	Develop national AI literacy curriculum for radiology professionals	Ministry of Health, universities, professional societies	Improved workforce readiness
Short-term (1–2 years)	Establish national AI governance and ethics framework	Regulatory agencies and healthcare authorities	Enhanced trust, transparency, and governance
Medium-term (3–5 years)	Invest in interoperable PACS/RIS infrastructure	Healthcare organizations and government agencies	Improved implementation capacity
Medium-term (3–5 years)	Pilot AI-supported radiology workflows in high-volume centers	Tertiary hospitals and healthcare networks	Improved efficiency and service delivery
Long-term (5–10 years)	Establish national AI validation and monitoring center	National health innovation agencies	Continuous performance monitoring
Long-term (5–10 years)	Incorporate sustainability metrics into AI procurement and evaluation processes	Healthcare regulators and policymakers	Environmentally responsible implementation

The recommendations provide a practical roadmap for translating positive workforce perceptions into sustainable implementation strategies, balancing innovation with governance, infrastructure, workforce capability, and environmental responsibility.

5. Conclusions

This study examined workforce readiness and integrating AI into radiology. Radiology professionals exhibit positive attitudes toward AI and a moderate understanding of its clinical applications. However, readiness for implementation is hindered by organizational barriers such as high costs, insufficient training, and cybersecurity concerns. The results indicate that successful AI integration necessitates more than technology; coordinated organizational preparedness and infrastructure readiness are crucial. AI should be considered a strategic element in healthcare transformation. Aligning AI strategies with healthcare objectives is crucial for achieving SDGs. The research highlights the importance of a socio-technical approach to ensure stakeholder engagement and ongoing evaluation throughout the implementation process. Bridging the gap between positive attitudes and operational readiness is essential for enhancing healthcare quality, efficiency, and patient outcomes. Future research should concentrate on objective measures of implementation success and the long-term effects of AI integration.