Deep-Learning-Based Hemoglobin Concentration Prediction and Anemia Screening Using Ultra-Wide Field Fundus Images

Abstract

Background: Anemia is the most common hematological disorder. The purpose of this study was to establish and validate a deep-learning model to predict Hgb concentrations and screen anemia using ultra-wide-field (UWF) fundus images.Methods: The study was conducted at Peking Union Medical College Hospital. Optos color images taken between January 2017 and June 2021 were screened for building the dataset. ASModel_UWF using UWF images was developed. Mean absolute error (MAE) and area under the receiver operating characteristics curve (AUC) were used to evaluate its performance. Saliency maps were generated to make the visual explanation of the model.Results: ASModel_UWF acquired the MAE of the prediction task of 0.83 g/dl (95%CI: 0.81–0.85 g/dl) and the AUC of the screening task of 0.93 (95%CI: 0.92–0.95). Compared with other screening approaches, it achieved the best performance of AUC and sensitivity when the test dataset size was larger than 1000. The model tended to focus on the area around the optic disc, retinal vessels, and some regions located at the peripheral area of the retina, which were undetected by non-UWF imaging.Conclusion: The deep-learning model ASModel_UWF could both predict Hgb concentration and screen anemia in a non-invasive and accurate way with high efficiency.

Retinal image‐based artificial intelligence in detecting and predicting kidney diseases: Current advances and future perspectives

“…Later, Mitani et al 91 . and Zhao et al 92 . leverage retinal fundus images to predict hemoglobin concentration and detect the status of anemia accurately (MAE: 0.67, AUC: 0.87; MAE: 0.83, AUC: 0.93, respectively).…”

“…Chen et al 90 applied retinal vessel images from OCT to directly and non-evasively observe the hemoglobin concentration in the retina for the first time. Later, Mitani et al 91 and Zhao et al 92 leverage retinal fundus images to predict hemoglobin concentration and detect the status of anemia accurately (MAE: 0.67, AUC: 0.87; MAE: 0.83, AUC: 0.93, respectively). For the predictions of electrolyte disorders, sodium was predicted with an R 2 of 0.12 (95% CI: 0.10-0.13), while the performance was poor for potassium, calcium, and phosphorus (R 2 = 0.07, 0.07, and 0.02, respectively).…”

A self‐supervised causal feature reinforcement learning method for non‐invasive hemoglobin prediction

“…Hb concentration serves as an essential marker for diagnosing anemia. To estimate this crucial indicator non-invasively, deep learning techniques have become a significant area of study [9,12,21]. The goal of predicting Hb concentrations on mobile devices necessitates the development of lightweight and highly effective neural networks.…”

“…Mannino, Bauskar, and Kasiviswanathan [45,46] trained and tested their model on a small dataset, and the volatility of their experimental results was notably large. The input of fundus images from Cropped UWF and UWF [12,47] required professional instruments and complicated operation. The model developed from the fundus images required high-resolution images, but the prediction precision was only slightly higher than that of our model.…”

“…Current major methods of image processing include Sobel edge detection [7] and a clustering algorithm, based on K-means clustering [8], to predict the Hb level. Deep learning techniques mainly classified Hb levels using methods, such as fingertip and eye images [9][10][11][12]. A study [13] used fingertip videos as an input for the Artificial Neural Network to predict Hb level and obtained a 0.93 rank order of correlation between prediction and ground truth.…”