MedAL: Accurate and Robust Deep Active Learning for Medical Image Analysis

Abstract

Active Learning methods create an optimized and labeled training set from unlabeled data. We introduce a novel Online Active Deep Learning method for Medical Image Analysis. We extend our MedAL active learning framework to present new results in this paper. Experiments on three medical image datasets show that our novel online active learning model requires significantly less labelings, is more accurate, and is more robust to class imbalances than existing methods. Our method is also more accurate and computationally efficient than the baseline model. Compared to random sampling and uncertainty sampling, the method uses 275 and 200 (out of 768) fewer labeled examples, respectively. For Diabetic Retinopathy detection, our method attains a 5.88% accuracy improvement over the baseline model when 80% of the dataset is labeled, and the model reaches baseline accuracy when only 40% is labeled. F I G U R E 1 Proposed Active Learning pipeline. To solve a supervised classification task, we will use a deep network (DN), an initial labeled dataset D t r ai n , an unlabeled dataset D or acl e , and an oracle who can label data. We desire to label as few examples as possible. Each active learning iteration, we use the DN to compute a feature embedding for all labeled examples in D t r ai n and the top M unlabeled examples from D or acl e with highest predictive entropy. We select and label oracle examples furthest in feature space from the centroid of all labeled examples. The oracle examples are selected one at a time, and the centroid updated after each labeling. We train the model on the expanded training set and repeat the process. In the online setting, the model weights are not reset between iterations and we use only the newly labeled examples and a subset of previously labeled examples.

O‐MedAL: Online active deep learning for medical image analysis

“…Finally, we train the model on the new data and repeat the sampling process over again. The original MedAL paper (Smailagic et al, 2018) evaluated a variety of distance functions and found empirically that Euclidean distance and cosine distance yielded the highest and second highest entropy, respectively, when F I G U R E 1 Proposed active learning pipeline. To solve a supervised classification task, we will use a deep network (DN), an initial labeled dataset D train , an unlabeled dataset D oracle , and an oracle who can label data.…”

“…To configure the AL sampler, we use the Euclidean distance and obtain feature embeddings from the Mixed5 layer of Inception V3. These choices are a result of our prior empirical evidence that this combination of layer and distance function achieves the highest entropy (Smailagic et al, 2018).…”

“…We introduce MedAL to our prior work. MedAL is a novel AL approach to optimize the selection of unlabeled samples by combining predictive entropy-based uncertainty sampling (US) with a distance function on a learned feature space (Smailagic et al, 2018). MedAL's active sampling mechanism minimizes the required labels by selecting only those unlabeled images that are most informative to the model as it trains.…”

Knowledge Transfer Between Buildings for Post-Earthquake Damage Diagnosis Without Historical Data

“…By constructing neural networks with deep and special architectures, we can approximate a wide range of highly non-linear and complex mapping functions [59]. A variety of deep neural networks have been proved to be effective and powerful in many real-world tasks, including computer vision [60], natural language processing [61], medical imaging [62,63] and video game [64]. Some typical architectures of deep neural networks include deep convolutional neural network (CNN) [60] and recurrent neural network [65].…”

ARA: accurate, reliable and active histopathological image classification framework with Bayesian deep learning

“…This technique can be conceptualised by an analogy to a diligent student, who while taking a course actively asks the teacher for more examples on topics which are hard for them to understand. There are several attempts at active learning for histopathological image classification in the literature, using both traditional machine learning 37,[39][40][41][42] and deep learning [43][44][45][46] . However, none of these approaches utilised uncertainty for selection of new samples in active training.…”