Computer-Aided Diagnosis of Pneumoconiosis X-ray Images Scanned with a Common CCD Scanner

Abstract

This paper presents a discrimination of pneumoconiosis X-ray images obtained with a common CCD scanner. Current computer-aided diagnosis systems of pneumoconiosis have been proposed to images obtained with a special scanner such as a drum scanner or a film scanner for X-ray pictures. However, since the special scanners need a large storage space and the scanners and commitment of the imaging need high-priced costs, the systems are not practical in small clinics. In this paper, we propose features for measuring abnormalities of pneumoconiosis as variables for the discrimination. Devices in the proposed system are only a tablet PC and a CCD scanner. In images obtained with CCD scanner, abnormal levels of pneumoconiosis could depend on density distribution in rib areas. Therefore, the proposed method measures the abnormalities by extracting characteristics of the distribution in the areas. Besides, using the abnormalities, the proposed method discriminates chest X-ray images into normal or abnormal cases of pneumoconiosis. Experimental results of the discriminations for 59 right-lung images have shown that the proposed abnormalities are well extracted for the discrimination.

Computer-Aided Diagnosis of Coal Workers’ Pneumoconiosis in Chest X-ray Radiographs Using Machine Learning: A Systematic Literature Review

“…The study found five major categories of texture feature analysis methods, where texture feature was extracted using Fourier spectrum [ 39 , 40 , 44 , 48 , 50 , 53 , 58 ], co-occurrence matrix analysis [ 42 , 48 , 50 , 53 , 55 , 57 , 58 , 59 , 61 , 64 , 79 ], histogram analysis [ 34 , 47 , 50 , 55 , 59 , 61 , 63 , 66 ], wavelet transform [ 52 , 56 ], and density distribution [ 42 , 45 , 46 , 51 , 54 , 60 , 62 ]. The details of the five methods are discussed in the following subsections and the texture features extracted from them, as described in Table 1 , are summarised.…”

“…An image’s texture varies depending on the scanner’s quality, which is very expensive in the clinical diagnosis system. To address this issue, Abe et al [ 51 , 54 ] and Nakamura et al [ 60 , 62 ] proposed a charge-coupled device (CCD) scanner for CAD of pneumoconiosis in CXR, where they computed the feature characteristics based on density distribution in a particular region or the areas between the ribs and its inter-costal. Their descriptions can be found in Table 1 .…”

Application of a Computer-aid Diagnosis of Pneumoconiosis for CR X-ray Images

“…In the CAD system for pneumoconiosis X-ray images digitalized by CCD scanner [6], abnormalities of pneumoconiosis are measured by extracting characteristics of the density distribution in rib areas. The rib areas are manually designated using a tablet PC.…”

“…Figure 2(c) shows an example of the curves drawn to Figure 2(a). Then, as shown in Figure 2(d), the rib areas R1-R4 are extracted by the drawn edges (Refer to [6]). As shown in Figure 1 and Figure 2, we can see that the images digitalized by CCD scanner are extremely unclear than CR images.…”

“…In order to enhance cost-performance of CAD for pneumoconiosis, a CAD system for pneumoconiosis using images digitalized by a common CCD scanner was reported [5,6]. This system is composed with a CCD scanner and a tablet PC, and discriminates pneumoconiosis with abnormalities in rib areas.…”