分级(工程)
睑裂
充血
分级比例尺
医学
人工智能
眼科
计算机科学
放射科
外科
工程类
土木工程
血流
作者
Rachael C. Peterson,James S. Wolffsohn
标识
DOI:10.1097/opx.0b013e3181981976
摘要
To convert objective image analysis of anterior ocular surfaces into recognisable clinical grades, in order to provide a more sensitive and reliable equivalent to current subjective grading methods; a prospective, randomized study correlating clinical grading with digital image assessment.The possible range of clinical presentations of bulbar and palpebral hyperaemia, palpebral roughness and corneal staining were represented by 4 sets of 10 images. The images were displayed in random order and graded by 50 clinicians using both subjective CCLRU and Efron grading scales. Previously validated objective image analysis was performed 3 times on each of the 40 images. Digital measures included edge-detection and relative-coloration components. Step-wise regression analysis determined correlations between the average subjective grade and the objective image analysis measures.Average subjective grades could be predicted by a combination of the objective image analysis components. These digital "grades" accounted for between 69% (for Efron scale-graded palpebral redness) and 98% (for Efron scale-graded bulbar hyperaemia) of the subjective variance.The results indicate that clinicians may use a combination of vessel areas and overall hue in their judgment of clinical severity for certain conditions. Objective grading can take these aspects into account, and be used to predict an average "objective grade" to be used by a clinician in describing the anterior eye. These measures are more sensitive and reliable than subjective grading while still utilizing familiar terminology, and can be applied in research or practice to improve the detection, and monitoring of ocular surface changes.
科研通智能强力驱动
Strongly Powered by AbleSci AI