KFWC: A Knowledge-Driven Deep Learning Model for Fine-grained Classification of Wet-AMD.

BACKGROUND AND OBJECTIVES: Automated diagnosis using deep neural networks can help ophthalmologists detect the blinding eye disease wet Age-related Macular Degeneration (AMD). Wet-AMD has two similar subtypes, Neovascular AMD and Polypoidal Choroidal Vasculopathy (PCV). However, due to the difficulty in data collection and the similarity between images, most studies have only achieved the coarse-grained classification of wet-AMD rather than a fine-grained one of wet-AMD subtypes. Therefore, designing and building a deep learning model to diagnose neovascular AMD and PCV is a great challenge. METHODS: To solve this issue, in this paper, we propose a Knowledge-driven Fine-grained Wet-AMD Classification Model (KFWC) to enhance the model's accuracy in the fine-grained disease classification with insufficient data. We innovatively introduced a two-stage method. In the first stage, we present prior knowledge of 10 lesion signs through pre-training; in the second stage, the model implements the classification task with the help of human knowledge. With the pre-training of priori knowledge of 10 lesion signs from input images, KFWC locates the powerful image features in the fine-grained disease classification task and therefore achieves better classification. RESULTS: To demonstrate the effectiveness of KFWC, we conduct a series of experiments on a clinical dataset collected in cooperation with a Grade III Level A ophthalmology hospital in China. The AUC score of KFWC reaches 99.71%, with 6.69% over the best baseline and 4.14% over ophthalmologists. KFWC can also provide good interpretability and effectively alleviate the pressure of data collection and annotation in the field of fine-grained disease classification for wet-AMD. CONCLUSIONS: The model proposed in this paper effectively solves the difficulties of small data volume and high image similarity in the wet-AMD fine-grained classification task through a knowledge-driven approach. Besides, this method effectively relieves the pressure of data collection and annotation in the field of fine-grained classification. In the diagnosis of wet-AMD, KFWC is superior to previous work and human ophthalmologists.

Clinical decision support system for hypertension medication based on knowledge graph.

BACKGROUND: High prevalence of hypertension and complicated medication knowledge have presented challenges to hypertension clinicians and general practitioners. Clinical decision support systems (CDSSs) are developed to aid clinicians in decision making. Current clinical knowledge is stored in fixed templates, which are not intuitive for clinicians and limit the knowledge reusability. Knowledge graphs (KGs) store knowledge in a way that is not only intuitive to humans but also processable by computers directly. However, existing medical KGs such as UMLS and CMeKG are general purpose and thus lack enough knowledge to enable hypertension medication. METHODS: We first construct a KG specific to hypertension medication according to the Chinese hypertension guideline and then develop the corresponding CDSS to implement hypertension medication and knowledge management. Current advances in knowledge graph representation and modelling are researched and applied in the complex medical knowledge representation. Traditional knowledge representation and KG representation are innovatively combined in the storage of the KG to enable convenient knowledge management and easy application by the CDSS. Along a predefined reasoning path in the KG, the CDSS finally accomplishes the hypertension medication by applying knowledge stored in the KG. 124 health records of a hypertension Chief Physician from Beijing Anzhen Hospital, Capital Medical University, are collected to evaluate the system metrics on the single drug recommendation task. RESULTS AND CONCLUSION: The proposed CDSS has functions of medication knowledge graph management and hypertension medication decision support. With elaborate design on knowledge representation, knowledge management is intuitive and convenient. By virtue of the KG, medication recommendations are highly visualized and explainable. Experiments on 124 health records with 90% guideline compliance collected from hospitals in single class recommendation task achieve 91%, 83% and 77% on recall, hit@3 and MRR metrics respectively, which demonstrates the quality of the KG and effectiveness of the system.

MBNM: Multi-branch network based on memory features for long-tailed medical image recognition.

BACKGROUND AND OBJECTIVES: Deep learning algorithms show revolutionary potential in computer-aided diagnosis. These computer-aided diagnosis techniques often rely on large-scale, balanced standard datasets. However, there are many rare diseases in real clinical scenarios, which makes the medical datasets present a highly imbalanced long-tailed distribution, leading to the poor generalization ability of deep learning models. Currently, most algorithms to solve this problem involve more complex modules and loss functions. But for complicated tasks in the medical domain, usually suffer from issues such as increased inference time and unstable performance. Therefore, it is a great challenge to develop a computer-aided diagnosis algorithm for long-tailed medical data. METHODS: We proposed the Multi-Branch Network based on Memory Features (MBNM) for Long-Tailed Medical Image Recognition. MBNM includes three branches, where each branch focuses on a different learning task: 1) the regular learning branch learns the generalizable feature representations; 2) the tail learning branch gains extra intra-class diversity for the tail classes through the feature memory module and the improved reverse sampler to improve the classification performance of the tail classes; 3) the fusion balance branch integrates various decision-making advantages and introduces an adaptive loss function to re-balance the classification performance of easy and difficult samples. RESULTS: We conducted experiments on the multi-disease Ophthalmic OCT datasets with imbalance factors of 98.48 and skin image datasets Skin-7 with imbalance factors of 58.3. The Accuracy, MCR, F1-Score, Precision, and AUC of our model were significantly improved over the strong baselines in the auxiliary diagnosis scenario where the clinical medical data is extremely imbalanced. Furthermore, we demonstrated that MBNM outperforms the state-of-the-art models on the publicly available natural image datasets (CIFAR-10 and CIFAR-100). CONCLUSIONS: The proposed algorithm can solve the problem of imbalanced data distribution with little added cost. In addition, the memory module does not act in the inference phase, thereby saving inference time. And it shows outstanding performance on medical images and natural images with a variety of imbalance factors.