Spatial Features Terms for Describing Lung Nodule Location in Chest X-Ray Images

Spatial features have gained attention in CBIR researches as a mean to represent image properties currently. These features provide fine queries to locate object location as well as its relation with others within an image. In fact, spatial features which is portrayed by spatial features terms (SFT) such as left, right, on and in, have been applied in many research domains to denote image attributes. Although the features play an important role in representing image, yet, many researches still rely on visual features for that matter. This condition is also applied in medical image such as Chest X-ray (CXR) image. There is less effort done to identify the actual SFT that should be used to describe the anomalies like lung nodule in CXR image. To overcome this problem, collection of SFTs for describing CXR image must be identified. Hence, this paper presents the effort in identifying the type of SFT that should be used to describe lung nodule in CXR. In order to identify the term, ten radiologists were asked to describe lung nodules in ten CXR images. As a result, five SFTs that are frequently use to describe the image, i.e. left, right, upper, middle and lower were derived from the descriptions. These SFTs are able to visualize the lung region divisions vertically and horizontally within CXR image.