Generation of chest CT pulmonary nodule images by latent diffusion models using the LIDC-IDRI dataset.
Authors
Affiliations (4)
Affiliations (4)
- Graduate School of Science and Engineering, Meijo University, 1-501 Shiogamaguchi, Tempaku-Ku, Nagoya City, Aichi, 468-8502, Japan.
- Graduate School of Science and Engineering, Meijo University, 1-501 Shiogamaguchi, Tempaku-Ku, Nagoya City, Aichi, 468-8502, Japan. [email protected].
- Fujita Health University, 1-98 Dengakugakubo, Kutsukake Cho, Toyoake City, Aichi, 470-1192, Japan.
- Gifu University, 1-1 Yanagido, Gifu, 501-1194, Japan.
Abstract
Lung cancer is the leading cause of cancer-related deaths. Recently, computer-aided diagnosis (CAD) systems have been developed to support diagnosis, but their performance depends heavily on the quality and quantity of training data. However, in clinical practice, it is difficult to collect a large amount of CT images for specific cases, such as small-cell carcinoma with low epidemiological incidence or benign tumors that are difficult to distinguish from malignant ones. This leads to the challenge of data imbalance. In this study, we propose a method to automatically generate chest CT nodule images that capture target features using latent diffusion models (LDM), and verify its effectiveness. Using the LIDC-IDRI dataset, we created pairs of nodule images and findings based on physician evaluations. For the image-generation models, we used Stable Diffusion version 1.5 (SDv1) and 2.0 (SDv2), which are types of LDM. Each model was fine-tuned using the created dataset. During the generation process, we adjusted the guidance scale (GS) to indicate the fidelity of the input text. Quantitative evaluation used metrics to measure image quality, diversity, and consistency with the text. Subjective evaluation consisted of visual assessments conducted by three radiological technologists. Quantitative and subjective evaluations showed that SDv2 (GS = 5) achieved the best performance in terms of image quality, diversity, and text consistency. In the subjective evaluation, no statistically significant differences were observed between the generated and real images, no statistically significant differences were detected between SDv2-generated and real ROI images across the evaluated categories. We propose a method for generating chest CT nodule images based on input text using LDM. The evaluation results demonstrate that the proposed method can generate high-quality images that can successfully capture specific medical features.