Evaluating image quality in surgical photography: a multivariable analysis of cameras and shooting conditions.
Kourounis G., Elmahmudi AA., Thomson B., Tingle SJ., Glover EK., Mahendran B., Thompson E., Abbas SH., Al-Leswas D., Brown C., El Zawahry MAMES., Georgiades F., Knight SR., Kuzman M., Owen R., Psaltis E., Hunter J., Ugail H., Wilson C.
INTRODUCTION: Use of mobile devices with high-quality cameras has expanded medical photography. We investigate the impact of different devices and conditions on photograph quality in a surgical setting. METHODS: Fourteen surgeons across six centres scored photograph quality of kidneys donated for transplantation. Images were captured using an iPhone, iPad, or DSLR camera on automatic modes under varying lighting conditions. In blinded A/B testing, surgeons selected the image perceived more clinically useful for remote organ quality assessment and rated each on a 5-point Likert scale. Quality was objectively analysed using two computer vision referenceless quality assessment tools (BRISQUE & NIMA). RESULTS: Of 369 photographs, mobile device images were rated higher quality by surgeons (78.4%) compared to DSLR (9.4%, p < 0.001). Multilevel regression using BRISQUE showed higher quality for iPhones (β = -5.86, p < 0.001) and iPads (β = -3.90, p < 0.001) versus DSLR. Room lighting improved quality over direct overhead illumination with theatre lights (β = 17.87, p < 0.001). Inter-rater (Gwet AC = 0.78) and intra-rater (Cohen's κ = 0.86) agreements were high. DISCUSSION: Smartphones can produce high quality photographs. These findings should reassure clinicians that smartphone devices do not compromise photograph quality and support their use in clinical practice and image analysis research.