Respirology (Carlton, Vic.) · 2015
Grey scale texture analysis of endobronchial ultrasound mini probe images for prediction of benign or malignant aetiology.
СтатьяПоддержка исследования: вне правительства СШАНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Nguyen Phan, Bashirzadeh Farzad, Hundloe Justin, Salvado Olivier, Dowson Nicholas, Ware Robert, Masters Ian Brent, Ravi Kumar Aravind, Fielding David
Аннотация
BACKGROUND AND OBJECTIVE: Expert analysis of endobronchial ultrasound mini probe (EBUS-MP) images has established subjective criteria for discriminating benign and malignant disease. Minimal data are available for objective analysis of these images. The aim of this study was to determine if greyscale texture analysis could differentiate between benign and malignant lung lesions.
METHODS: Digital EBUS-MP images with a gain setting of 10/19 and contrast setting of 4/8 from 2007 until 2012 inclusive were included. These images had an expert-defined region of interest (ROI) mapped. ROI were analysed for the following greyscale texture features: mean pixel value, difference between maximum and minimum pixel value, standard deviation of the mean pixel value, entropy, correlation, energy and homogeneity. Significant greyscale texture features differentiating benign from malignant disease were used by two physicians to assess a validation set.
RESULTS: A total of 167 images were available. The first 85 lesions were used in the prediction set. Benign lesions had larger differences between maximum and minimum pixel values, larger standard deviations of the mean pixel values and higher entropy than malignant lesions (P < 0.0001 for all values). A total of 82 peripheral lesions were in the validation set. Physician 1 correctly classified 63/82 (76.8%) with a negative predictive value (NPV) for malignancy of 82% and positive predictive value (PPV) of 75%. Physician 2 correctly classified 62/82 (75.6%) with a NPV of 100% and PPV of 71.0%.
CONCLUSIONS: Greyscale texture analysis of EBUS-MP images can help establish aetiology with a high NPV for malignancy.
METHODS: Digital EBUS-MP images with a gain setting of 10/19 and contrast setting of 4/8 from 2007 until 2012 inclusive were included. These images had an expert-defined region of interest (ROI) mapped. ROI were analysed for the following greyscale texture features: mean pixel value, difference between maximum and minimum pixel value, standard deviation of the mean pixel value, entropy, correlation, energy and homogeneity. Significant greyscale texture features differentiating benign from malignant disease were used by two physicians to assess a validation set.
RESULTS: A total of 167 images were available. The first 85 lesions were used in the prediction set. Benign lesions had larger differences between maximum and minimum pixel values, larger standard deviations of the mean pixel values and higher entropy than malignant lesions (P < 0.0001 for all values). A total of 82 peripheral lesions were in the validation set. Physician 1 correctly classified 63/82 (76.8%) with a negative predictive value (NPV) for malignancy of 82% and positive predictive value (PPV) of 75%. Physician 2 correctly classified 62/82 (75.6%) with a NPV of 100% and PPV of 71.0%.
CONCLUSIONS: Greyscale texture analysis of EBUS-MP images can help establish aetiology with a high NPV for malignancy.