AI-AUGMENTED IMAGING FOR PRECISION DIAGNOSIS OF PULMONARY DISEASES

Abstract

Background: Medical imaging feels like it is going to get better and better with the introduction of artificial intelligence through quick, accurate, and precise diagnoses. Even so, AI-facilitated imaging, particularly in the diagnosis of pulmonary diseases is still on the adoption curve with conflicting feedback from clinicians on its promise and hurdles. Objective: This study aimed to evaluate the perception, confidence, and challenges facing healthcare professionals using AI-integrated imaging for the diagnosis of COPD, lung cancer, and other interstitial lung diseases. Methods: A survey cross-sectional quantitative study design was used whereby questionnaires were administered to 250 health workers, particularly radiologists, pulmonologists, AI scientists, and medical technologists. Data management and analysis were descriptive analysis, correlation analysis, reliability analysis, and ANOVA where relationships between variables such as confidence and the perceived impact of AI on the process of diagnosis were examined. Results: The findings showed that there was a considerable degree of confidence in AI being accurate as a diagnostician. There was a low yet positive relationship (r = 0.068, p > 0.05) between confidence in the accuracy of AI and the overall perceived impact. The Shapiro-Wilk test specified that confidence levels are not normally distributed (p < 0.0001). The Cronbach's Alpha for AI using the beneficiaries was negative meaning low internal consistency. One-way ANOVA showed no significant differences in perceptions of AI in the sample group concerning career (p > 0.05). Inadequate clinical validation and unavailability of sufficient training are the main barriers mentioned. Conclusion: There are AI-augmented imaging applications for the diagnosis of pulmonary disease; however, the medical community remains hesitant to completely adopt these tools. The major obstacles include a lack of adequate clinical validation as well as training. These barriers together with fostering trust in AI's capabilities to operate independently are important for its wider use within the clinical environment.