Absolute ground truth-based validation of computer-aided nodule detection and volumetry in low-dose CT imaging

Over the last few years, advancements in artificial intelligence (AI) and increased expertise in deep learning have proven to be proficient in learning patterns for large-scale data [1]. Convolutional neural networks have been successfully utilized in endeavors to develop algorithms for image interpretation tasks in radiology [1 –4]. A lot of interest has been raised for the development of AI tools for automatic detection of pulmonary nodules on chest Computed Tomography (CT)[1,5–7]. The potentially huge number of CT examinations, associated with the advent of population-based lung cancer screening (LCS) programs, could represent a substantial increase in the workload of radiologists as interpretation of these CT scans is a time-consuming, intensive task, prone to fatigue errors [4,7–11].
Source: Physica Medica: European Journal of Medical Physics - Category: General Medicine Authors: Source Type: research