Gastric X-ray examination has been used as a major clinical inspection method for gastric cancer screening. An X-ray image has the advantage of being capable of not only clearly depicting the morphology of a gastric cancer lesion but also of widely showing the properties of the background mucosa. A diagnostic method for early gastric cancer that incorporates this characteristic is the "gastric cancer triangle. " The triangle of gastric cancer is obtained by classifying the background mucosa of gastric cancer into either fundic glands or intestinal metaplasia, histopathological type into either undifferentiated or differentiated type, and macroscopic type into either depressed or protruded type. The three elements are associated with each other. This diagnostic theory improves the accuracy of clinical diagnosis of gastric cancer. In addition, if the Helicobacter pylori infection status is determined by evaluating the degree of chronic active gastritis and atrophy based on the findings of X-ray imaging, it is possible to predict individual patients' risk of gastric cancer. In this article, the author reviewed the current status of gastric X-ray examination from the viewpoint of cancer screening and reported on advanced research that incorporated the computer technology of machine learning.

Purpose: We evaluated the accuracy of image interpretation by physicians depending on their experience in image reading using the "Categorized Criteria for X-Ray Image Reading of the Population-Based Gastric Cancer Screening" of the Japanese Society of Gastrointestinal Cancer Screening.

Subjects and Methods: Twelve gastroenterologists involved in the interpretation of X-ray images in gastric cancer screening participated in the study. They mean number of years of experience in interpreting gastric X-ray images was 13.3±10.1 (range, 1-30; median, 10.5). They were requested to categorize gastric X-ray images presented randomly from 20 malignant and 20 non-malignant cases that were endoscopically confirmed after abnormalities were noted on gastric X-ray screening.

Results: For the correlation between experience in and accuracy of interpretation of gastric X-ray images, R values of the accuracy, sensitivity, specificity, early cancer sensitivity, and advanced cancer sensitivity were 0.642 (Y = 0.556+0.011X, P < 0.05), 0.613 (Y = 0.383+0.012X, P < 0.05), 0.00364 (Y = 0.711+0.00007X, P = 0.991), 0.496 (Y = 0.366+0.017X, P < 0.01), and 0.776 (Y = 0.387+0.022X, P < 0.01), respectively.

Conclusion: Adequate experience is required for the accurate interpretation of X-ray images in gastric cancer screening. In addition, a long-term, well-designed education and training system is necessary to train radiologists.

Purpose: The Japanese Society of Gastrointestinal Cancer Screening implements a certification system for interpretation assistance by radiological technologists.

We assessed the category judgment of image reading support by radiological technologists and the category judgment of image interpretation by physicians using "Categorized Criteria for X-Ray Image Reading of the Population-Based Gastric Cancer Screening" of the Japanese Society of Gastrointestinal Cancer Screening.

Materials and Methods: We recruited 30 radiological technologists who were gastric cancer screening special technologists and 12 gastroenterologists engaged in the interpretation of gastric X-ray images for cancer screening. Gastric X-ray images of individuals who were later confirmed to have either malignant (n = 20) or non-malignant disease (n = 20) were used randomly to determine the accuracy of categorization.

In the case of malignancy, the interpretation sensitivity was judged to be correct based on the judgment of category 3b or higher. Then, in the case of non-malignant case, it was judged that the specificity of interpretation was correct in category 3a and below.

Results: Physicians' interpretation category judgment had a correct diagnosis rate of 66.3%, a sensitivity of 65.0% (61.5% for early-stage cancer and 71.4% for advanced cancer), and a specificity of 75.0%. Meanwhile, technologists' category judgment had an accuracy of 70.0%, a sensitivity of 85.0% (76.9% for early-stage cancer and 85.7% for advanced cancer), and a specificity of 60.0%.

Conclusion: There was no significant difference between physicians and radiological technologists in the judgment category classification.

Diffuse redness (DR) is an endoscopic feature strongly suggestive of Helicobacter pylori (H. pylori) infection. However, the assessment of this feature is dependent on the setting of the device and the subjectivity of the endoscopist, which is a major issue. We, therefore, developed a procedure for DR quantification using image analysis and assessed the usefulness of the procedure. A visual assessment of the endoscopic images of 115 patients who underwent upper gastrointestinal endoscopy and an H. pylori infection examination, was done by six endoscopists. The kappa value and diagnostic accuracy of H. pylori infection according to the visual assessment using all images (0.43 and 67.7%, respectively) were higher than those using only 2 images (antrum and greater curvature of body) (0.31 and 55.8%, respectively). The kappa values using the DR quantification (0.93), when the cut-off value calculated by the receiver operating characteristic curve analysis was defined as 0.975, were much higher than those by a visual assessment using either 2 or all endoscopic images (0.31 or 0.43, respectively).

The diagnostic accuracy of DR quantification to assess H. pylori infection was 64.1%, which was equivalent to that achieved by a visual assessment using all endoscopic images. The procedure of DR quantification is expected to be used to screen current H. pylori infection because of the high interobserver agreement and diagnostic accuracy.

Background: In gastric cancer screening, the upper gastrointestinal series has long been performed as the only method with evidence of a reduction effect on gastric cancer mortality. With the remarkable progress of endoscopic technology in the gastrointestinal field, endoscopy is now the first choice for gastrointestinal tract examination instead of X-ray examination. There has been a discrepancy between clinical practice and medical examination.

Subjects and methods: In Wakayama City, endoscopic examination has been available for population-based screening since 2012. Subsequently, in the revised 2015 guidelines, endoscopy was newly recommended for both population-based and opportunistic screening. In Wakayama City, endoscopic examination using a double-check system was introduced in 2015, one year ahead of the national policy.

Results: A significant increase in the cancer detection rate was observed with the introduction of endoscopy (2.9 cancers per 1,000 screening exams) and the double-check system (5.5 cancers per 1,000 screening exams) compared with the period of X-rays alone (0.6 cancers per 1,000 screening exams). In addition to gastric cancer, the detection rate of esophageal cancer has also increased. Most of the detected cancers were pointed out at the time of primary endoscopy and it was considered that introducing workshops or evaluation sheets could have the effect of improving the accuracy of endoscopic examinations. On the other hand, the participation rate showed a tendency to increase at the beginning of the endoscopic introduction but remained flat thereafter.

Conclusions: The gastric cancer detection rate significantly increased with the introduction of endoscopy with the double-check system; however, the screening participation rate is still low. It is necessary to strengthen the examination accuracy control and encourage medical screening.

Yamagata City government and the Yamagata City Medical Association initiated a gastric cancer/gastritis X-ray screening with stratified gastric cancer risk assessment (ABC classification) in fiscal year 2017. The total number of subjects was 13,663, which included all residents of Yamagata city older than 40 years old. They underwent conventional X-ray screening, and optional ABC classification was performed for subjects who did not receive the ABC classification assessment and those who did not undergo Helicobacter pylori treatment. Subjects in ABC classification groups B, C, or D were recommended to undergo gastric endoscopy. Five patients were diagnosed with early stage gastric cancer, which was not detected by X-ray screening. We confirmed that gastric X-ray screening combined with ABC classification was useful. When upper gastrointestinal endoscopy was initiated as the secondary diagnostic examination, we organized a secondary diagnostic examination committee to manage the quality of diagnostic accuracy. Establishment of the endoscopic examination system with a practical protocol, which included a double-check analysis of endoscopic image interpretation and feedback of the results to the examiners, enabled us to build a foundation for gastric endoscopic screening to detect early stage gastric cancer.

Background: The role of upper gastrointestinal endoscopy (UGIE) has been recognized in the early detection of laryngopharyngeal and oral cavity cancers; however, the use of UGIE in screening for these cancers has been insufficiently studied.

Subjects and Methods: A total of 51,628 UGIE examinations (n=15,433 patients) for opportunistic screening were performed. We retrospectively investigated the cases of confirmed diagnosis or the subsequent course of 123 cases with pharyngeal and oral findings that required additional investigation by the UGIE examinations.

Results: Additional inspection of suspicious laryngopharyngeal and oral findings was required for 0.24% (123/51,628) cases. In 77 cases, the diagnosis was confirmed by an otolaryngologist. Seven cases had lesion disappearance, which was confirmed by repeat UGIE. In two cases, repeat UGIE performed after five or more years showed no changes in the lesion. The lesions in these nine cases were considered non-malignant. The detection rate and positive predictive value (PPV) for laryngopharyngeal and oral cancers were 0% (0/51,628) and 0% (0/86), respectively. The detection rate and PPV for mild-to-moderate atypical lesions were 0.01% (6/51,628) and 7.8% (6/77), respectively. The most frequent diagnoses were normal, granuloma, papilloma, lymph follicle, and cyst.

Conclusions: UGIE for opportunistic screening without strict guidelines on the consequent observation of the laryngopharyngeal and oral cavity revealed low detection rates of laryngopharyngeal and oral cancers. In addition, utilizing UGIE in this manner led to false positives and overdiagnosis. To minimize these disadvantages, a consensus between laryngopharyngeal and esophageal experts is needed regarding the observation method, proper assessment, and risk stratification of oral and laryngopharyngeal areas using screening UGIE.