A Series-Based Deep Learning Approach to Lung Nodule Image Classification

SIMPLE SUMMARY: Medical image classification is an important task in computer-aided diagnosis, medical image acquisition, and mining. Although deep learning has been shown to outperform traditional methods based on handcrafted features, it remains difficult due to significant intra-class variation and inter-class similarity caused by the diversity of imaging modalities and clinical pathologies. This study presents an innovative method that is an intersection between 3D image analysis and series classification problems. Therefore, the self-similarity features in medical images are captured by converting the regions of interest to series with a radial scan and these series are classified with U-shape convolutional neural networks. The findings of this study are expected to be used by researchers from various disciplines working on radial scanned images, as well as researchers working on artificial intelligence in health. ABSTRACT: Although many studies have shown that deep learning approaches yield better results than traditional methods based on manual features, CADs methods still have several limitations. These are due to the diversity in imaging modalities and clinical pathologies. This diversity creates difficulties because of variation and similarities between classes. In this context, the new approach from our study is a hybrid method that performs classifications using both medical image analysis and radial scanning series features. Hence, the areas of interest obtained from images are subjected to a radial scan, with their centers as poles, in order to obtain series. A U-shape convolutional neural network model is then used for the 4D data classification problem. We therefore present a novel approach to the classification of 4D data obtained from lung nodule images. With radial scanning, the eigenvalue of nodule images is captured, and a powerful classification is performed. According to our results, an accuracy of 92.84% was obtained and much more efficient classification scores resulted as compared to recent classifiers.