The Thai Journal of Radiological Technology

Development of a web application for stroke diagnosis assistance using deep learning artificial intelligence on computed tomography image

2024-01-26T15:08:45+07:00

Stroke presents a significant health risk for the elderly, demanding precise diagnosis through computed tomography (CT) imaging. Expert interpretation is crucial, and present artificial intelligence (AI) proves invaluable for radiologists, ensuring accuracy. Various accessible programs, but requiring payment, can be installed on devices. This study aims to develop a web application for stroke detection in brain CT images. The web application, compatible with computers, phones, and tablets, employs deep learning AI for stroke classification. Using a dataset of 1,636 images (1,111 normal, 525 stroke), about 70% (1,175 images) were used to train, 20% (329 images) for validation, and 10% (132 images) for test the AI model (Deep learning: VGG-16). Evaluation metrics, including accuracy, sensitivity, specificity, F1-score, and are under curve (AUC), gauged the web app's performance. Design and functionality were assessed through a 5-point Likert scale by three radiologists. Results show impressive accuracy, sensitivity, specificity, F1-score, and AUC (0.969, 0.952, 0.978, 0.952, 0.965). Design and performance scores were 4.13 ± 0.38 and 4.37 ± 0.33. In conclusion, the web application effectively diagnoses strokes in brain CT images, offering a user-friendly experience on the internet.

Optimizing the minimum detectable difference of chest protocol digital radiography system by V-shaped line gauge phantom and Taguchi analysis

2024-04-02T14:24:47+07:00

Background: The Minimum Detectable Difference (MDD) of Digital Radiography (DR) image systems was quantified and optimized using a self-developed V-shaped line gauge phantom and Taguchi optimization analysis. The increased clinical applications of the DR system are mainly due to its ability to provide instant and precise imaging for radiologists to diagnose. Objective: The purpose of this study was to optimize parameters for MDD values based on Taguchi's analysis and to verify a self-developed V-shaped line gauge phantom for the chest protocol of the DR system. Materials and methods: The chest protocol was exposed using the V-shaped line gauge and solid acrylic water phantom in various sizes with the DR system. Five factors were assigned in this study: (A) focal spot, (B) kVp, (C) mAs, (D) filter, and (E) phantom thickness. Since each factor could have two or three levels, eighteen groups of factor combinations were organized according to Taguchi’s algorithm. The V-shaped line gauge images were estimated. ANOVA was adopted to determine the significant factors and MDD values to evaluate the spatial resolution. Results: The optimal parameters for the highest Signal-to-Noise (S/N) ratio were as follows: (A) large focal spot, (B) 140 kVp, (C) 6.25 mAs, (D) 0.1 mmCu filter, and (E) phantom thickness of 16 cm. Conclusion: The optimal parameters in this study provided an MDD value of 0.87 mm compared to 1.66 mm for conventional parameters. The self-developed V-shaped line gauge phantom also proved to be suitable for the DR system. Furthermore, all factors affected the image quality with statistical significance following the ANOVA analysis.

Setup errors and CTV to PTV margin for upper abdominal cancer with intensity modulated radiotherapy technique using electronic portal imaging device in Sakon Nakhon Hospital

2023-12-07T15:25:16+07:00

Introduction: The systematic and random setup errors for patients with upper abdominal cancer are specific to each hospital. Objective: This study aimed to determine the systematic error, random error, and CTV to PTV margin for patient positioning in upper abdominal cancer with intensity-modulated radiotherapy (IMRT) using the electronic portal imaging device (EPID). Materials and Methods: Patient information and setup errors of 32 patients undergoing radiotherapy were collected. The data on setup errors in three directions (longitudinal, lateral, and vertical) were retrospectively collected from the Mosiaq Platform. Subsequently, individual and population systematic and random errors were calculated. The CTV to PTV margins were then determined using the Van Herk equation. Finally, correlations between setup error and BMI, age, and PTV volume were analyzed. Results: The results showed that the setup errors in the longitudinal, lateral, and vertical directions were 2.59±1.44, 2.17±0.95, and 1.66±0.59 mm, respectively. The population systematic errors in the longitudinal, lateral, and vertical directions were 1.44, 0.95, and 0.59 mm, respectively. The population random errors in the longitudinal, lateral, and vertical directions were 2.59, 2.17, and 1.66 mm, respectively. The determined CTV to PTV margins in the longitudinal, lateral, and vertical directions were 5.42, 3.89, and 2.62 mm, respectively. There was no correlation between setup error and BMI, age, or PTV volume, with p-values > 0.05. Conclusion: The CTV to PTV margins were less than 5 mm in all directions except the longitudinal direction. This finding may be useful for reviewing the PTV margin for Sakon Nakhon hospital.

Repeat film analysis and its implications for quality assurance in diagnostic radiology: An institutional case study

2023-12-07T15:34:36+07:00

The primary objective of this research was to analyze the rate of repeated chest examinations using the Reject and Usage Analysis tool program of the Samsung XGEO-GC80 X-ray machine. A retrospective study was conducted at King Chulalongkorn Memorial Hospital (KCMH) in the Department of Radiological Diagnostic. The study involved collecting statistics and closely examining the causes behind rejected images within the machine system. Among the key findings were Chest examinations accounted for the highest number of images, comprising 39.32% of the total 40,400 images. Repeated images were most prevalent in chest examinations, making up 32.21%of the 7,550 rejected images and 6.02% of all images. The primary reason for rejecting images was improper positioning, constituting 75.47% of all rejected images. In chest examinations, the most common reason for image rejection was improper breathing, contributing to 54.3% of all 1,321rejected images and 14.0% of all rejected images in the study. Furthermore, the study highlighted that a significant portion of X-ray image rejections resulted from human errors, primarily from radiological technologists. To reduce the rate of repeated imaging, the following recommendations were proposed for i) comprehensive evaluation of the patient's condition before the X-ray examination, ii) Ensuring proper patient positioning, iii) Implementation of various initiatives within the healthcare unit, including regular training for radiologists, and iv) Establishment of criteria for accepting radiographic images across all radiological technologists within the diagnostic department. In conclusion, these analyses are essential to enhance the quality of radiological services at KCMH, minimize image rejections, and ultimately improve patient care and radiation safety.