Purpose: Deep learning techniques have recently achieved impressive results in a variety of computer vision problems, raising expectations that they might be applied in other domains, such as medical image analysis. We propose and evaluate a convolutional neural network (CNN), designed for the classification of interstitial lung disease (ILD) patterns. Materials and methods: The proposed network consists of 5 convolutional layers with 2×2 kernels and LeakyReLU activations, followed by average

more »

... oling with size equal to the size of the final feature maps and three dense layers. The last dense layer has 7 outputs, equivalent to the classes considered: healthy, ground glass opacity (GGO), micronodules, consolidation, reticulation, honeycombing and a combination of GGO/reticulation (Fig. 1) . To train and evaluate the CNN, we used a dataset of 14696 image patches, derived by 120 CT scans from different scanners and hospitals (Fig. 2) . A comparative analysis proved the effectiveness of the proposed CNN against previous methods in a challenging dataset. Results: The classification performance (~85.5%) demonstrated the potential of CNNs in analyzing lung patterns. Pattern-sensitivities reached from 99% (consolidation) to 69% (honeycombing). The individual "true positive" and "false negative" results for each pattern is demonstrated in Fig. 3 . Conclusion: The CNN showed very promising results in lung pattern recognition outperforming many state-of-the-art methods. Future work includes, extending the CNN to three-dimensional data provided by CT volume scans. Clinical Relevance: Integrating the proposed method into a CAD system helps providing a differential diagnosis for ILDs as a supportive tool for radiologists. Fig 1. Examples of healthy tissue and typical ILD patterns from left to right: healthy, GGO, micronodules, consolidation, reticulation, honeycombing, combination of GGO and reticulation.