Development of automatic generation system for lung nodule finding descriptions.

Worldwide, lung cancer is the leading cause of cancer-related deaths. To manage lung nodules, radiologists observe computed tomography images, review various imaging findings, and record these in radiology reports. The report contents should be of high quality and uniform regardless of the radiologist. Here, we propose an artificial intelligence system that automatically generates descriptions related to lung nodules in computed tomography images. Our system consists of an image recognition method for extracting contents-namely, bronchopulmonary segments and nodule characteristics from images-and a natural language processing method to generate fluent descriptions. To verify our system's clinical usefulness, we conducted an experiment in which two radiologists created nodule descriptions of findings using our system. Through our system, the similarity of the described contents between the two radiologists (p = 0.001) and the comprehensiveness of the contents (p = 0.025) improved, while the accuracy did not significantly deteriorate (p = 0.484).

Measurement of solid size in early-stage lung adenocarcinoma by virtual 3D thin-section CT applied artificial intelligence.

An artificial intelligence (AI) system that reconstructs virtual 3D thin-section CT (TSCT) images from conventional CT images by applying deep learning was developed. The aim of this study was to investigate whether virtual and real TSCT could measure the solid size of early-stage lung adenocarcinoma. The pair of original thin-CT and simulated thick-CT from the training data with TSCT images (thickness, 0.5-1.0 mm) of 2700 pulmonary nodules were used to train the thin-CT generator in the generative adversarial network (GAN) framework and develop a virtual TSCT AI system. For validation, CT images of 93 stage 0-I lung adenocarcinomas were collected, and virtual TSCTs were reconstructed from conventional 5-mm thick-CT images using the AI system. Two radiologists measured and compared the solid size of tumors on conventional CT and virtual and real TSCT. The agreement between the two observers showed an almost perfect agreement on the virtual TSCT for solid size measurements (intraclass correlation coefficient = 0.967, P < 0.001, respectively). The virtual TSCT had a significantly stronger correlation than that of conventional CT (P = 0.003 and P = 0.001, respectively). The degree of agreement between the clinical T stage determined by virtual TSCT and the clinical T stage determined by real TSCT was excellent in both observers (k = 0.882 and k = 0.881, respectively). The AI system developed in this study was able to measure the solid size of early-stage lung adenocarcinoma on virtual TSCT as well as on real TSCT.

Iodine-related attenuation in contrast-enhanced dual-energy computed tomography in small-sized solid-type lung cancers is associated with the postoperative prognosis.

BACKGROUND: To investigate the correlation between iodine-related attenuation in contrast-enhanced dual-energy computed tomography (DE-CT) and the postoperative prognosis of surgically resected solid-type small-sized lung cancers. METHODS: We retrospectively reviewed the DE-CT findings and postoperative course of solid-type lung cancers ≤3 cm in diameter. After injection of iodinated contrast media, arterial phases were scanned using 140-kVp and 80-kVp tube voltages. Three-dimensional iodine-related attenuation (3D-IRA) of primary tumors at the arterial phase was computed using the "lung nodule" application software. The corrected 3D-IRA normalized to the patient's body weight and contrast medium concentration was then calculated. RESULTS: A total of 120 resected solid-type lung cancers ≤3 cm in diameter were selected for analysis (82 males and 38 females; mean age, 67 years). During the observation period (median, 47 months), 32 patients showed postoperative recurrence. Recurrent tumors had significantly lower 3D-IRA and corrected 3D-IRA at early phase compared to non-recurrent tumors (p = 0.046 and p = 0.027, respectively). The area under the receiver operating characteristic curve for postoperative recurrence was 0.624 for the corrected 3D-IRA at early phase (p = 0.025), and the cutoff value was 5.88. Kaplan-Meier curves for disease-free survival indicated that patients showing tumors with 3D-IRA > 5.88 had a significantly better prognosis than those with tumors showing 3D-IRA < 5.88 (p = 0.017). CONCLUSIONS: The 3D-IRA of small-sized solid-type lung cancers on contrast-enhanced DE-CT was significantly associated with postoperative prognosis, and low 3D-IRA tumors showed a higher TNM stage and a significantly poorer prognosis.

Unexpected radioactive iodine accumulation on whole-body scan after I-131 ablation therapy for differentiated thyroid cancer.

We retrospectively evaluated the frequency of unexpected accumulation of radioactive iodine on the post-therapy whole-body scan (Rx-WBS) after radioactive iodine (RAI) ablation therapy in patients with differentiated thyroid cancer (DTC). We searched our institutional database for Rx-WBSs of DTC patients who underwent RAI ablation or adjuvant therapy between 2012 and 2019. Patients with distant metastasis diagnosed by CT or PET/CT before therapy, and those had previously received RAI therapy were excluded. In total, 293 patients (201 female and 92 male, median age 54 years) were selected. Two nuclear medicine physicians interpreted the Rx-WBS images by determining the visual intensity of radioiodine uptake by the thyroid bed, cervical and mediastinal lymph nodes, lungs, and bone. Clinical features of the patients with and without the metastatic accumulation were compared by chi-square test and median test. Logistic regression analyses were performed to compare the association between the presence of metastatic accumulation and these clinical factors. Eighty-four of 293 patients (28.7%) showed metastatic accumulation. Patients with metastatic RAI accumulation showed a significantly higher frequency of pathological N1 (pN1) and serum thyroglobulin (Tg) > 1.5 ng/ml under TSH stimulation (p = 0.035 and p = 0.031, respectively). Logistic regression analysis indicated that a serum Tg > 1.5 ng/ml was significantly correlated with the presence of metastatic accumulation (odds ratio = 1.985; p = 0.033). In conclusion, Patients with Tg > 1.5 ng/ml were more likely to show metastatic accumulation. In addition, the presence of lymph node metastasis at the initial thyroid surgery was also associated with this unexpected metastatic accumulation.

Postoperative recurrence of clinical early-stage non-small cell lung cancers: a comparison between solid and subsolid nodules.

BACKGROUND: For subsolid non-small cell lung cancers (NSCLCs), solid size (SS), which is the maximal diameter of the solid component, correlates more accurately with tumor prognosis than the total size, which is the maximal diameter of the entire tumor, including ground-glass opacity. We reviewed the propriety of the TNM staging based on the SS for early-stage NSCLCs. METHODS: We retrospectively reviewed the preoperative radiological reports, clinical records, and pathological reports of NSCLC cases in our hospital between 2010 and 2013, and clinical stage (c-Stage) 0 and I tumors were selected. Disease-free survival (DFS), based on survival analysis, was used to assess the tumor characteristics that predicted the prognosis. RESULTS: A total of 247 NSCLC diagnoses in 231 patients (88 women and 143 men; age, 67 ± 7 years) were included in our cohort. They were classified into solid (n = 131) and subsolid (n = 116) nodules. The DFS curves indicated that prognosis was significantly worse in the following order: c-Stage 0, c-Stage IA, and c-Stage IB tumors (p = 0.016). Patients with solid nodules showed a significantly worse prognosis than patients with subsolid nodules (p < 0.001). A multivariate Cox proportional hazards model showed that the significant predictive factors for DFS were c-Stage (hazard ratio, 1.600; p = 0.020) and solid nodules (hazard ratio, 3.077; p = 0.031). CONCLUSIONS: For early-stage NSCLCs, the c-Stage based on the SS in subsolid nodules was useful for predicting postoperative DFS. In addition, whether nodules were solid or subsolid was another independent prognostic factor.

Utility of Metabolic Parameters on FDG PET/CT in the Classification of Early-Stage Lung Adenocarcinoma: Prediction of Pathological Invasive Size.

PURPOSE: This paper aims to explore the role of a metabolic parameter on F-FDG-PET/CT for clinical T-classification in early-stage adenocarcinoma. PATIENTS AND METHODS: One hundred six surgically resected pathological TNM stage (p-stage) 0/I lung adenocarcinomas were retrospectively reviewed. The solid size (SS) measured on thin-section CT and the pathological invasive size (IS) of tumors were recorded. The SUVmax and metabolic tumor volume with SUV ≥1.0 (MTV1.0) derived from PET/CT data were measured on a workstation, and the metabolic tumor diameter with SUV ≥1.0 (MTD1.0) was calculated automatically from MTV1.0. For the correlations between the IS and the SS, MTD1.0, or SUVmax, Pearson's correlation coefficients were compared using the Meng-Rosenthal-Rubin method. Additionally, the reproducibility between the clinical TNM stage (c-stage), based on the SS or MTD1.0, and the p-stage was analyzed using the kappa coefficient (k). RESULTS: For the correlations between the IS and the other parameters, Pearson correlation coefficient was 0.630 for the SS, 0.600 for the SUVmax, and 0.725 for MTD1.0. MTD1.0 correlated significantly and more strongly with the IS than the SS and the SUVmax did (P = 0.040, and P = 0.008, respectively). The reproducibility between p-stage and c-stage based on the SS was moderate (k = 0.529, P < 0.001), whereas that between p-stage and c-stage based on MTD1.0 was substantial (k = 0.676, P < 0.001). CONCLUSIONS: MTD1.0 on FDG-PET/CT was correlated significantly and more strongly with the pathological IS in lung adenocarcinomas than with the SS on thin-section CT. FDG-PET/CT could classify more precisely early-stage lung adenocarcinoma than the presently used T-classification based on thin-section CT findings.

Computer-aided Volumetry of Part-Solid Lung Cancers by Using CT: Solid Component Size Predicts Prognosis.

Purpose To investigate the relationship between the postoperative prognosis of patients with part-solid non-small cell lung cancer and the solid component size acquired by using three-dimensional (3D) volumetry software on multidetector computed tomographic (CT) images. Materials and Methods A retrospective study by using preoperative multidetector CT data with 0.5-mm section thickness, clinical records, and pathologic reports of 96 patients with primary subsolid non-small cell lung cancer (47 men and 49 women; mean age ± standard deviation, 66 years ± 8) were reviewed. Two radiologists measured the two-dimensional (2D) maximal solid size of each nodule on an axial image (hereafter, 2D MSSA), the 3D maximal solid size on multiplanar reconstructed images (hereafter, 3D MSSMPR), and the 3D solid volume of greater than 0 HU (hereafter, 3D SV0HU) within each nodule. The correlations between the postoperative recurrence and the effects of clinical and pathologic characteristics, 2D MSSA, 3D MSSMPR, and 3D SV0HU as prognostic imaging biomarkers were assessed by using a Cox proportional hazards model. Results For the prediction of postoperative recurrence, the area under the receiver operating characteristics curve was 0.796 (95% confidence interval: 0.692, 0.900) for 2D MSSA, 0.776 (95% confidence interval: 0.667, 0.886) for 3D MSSMPR, and 0.835 (95% confidence interval: 0.749, 0.922) for 3D SV0HU. The optimal cutoff value for 3D SV0HU for predicting tumor recurrence was 0.54 cm3, with a sensitivity of 0.933 (95% confidence interval: 0.679, 0.998) and a specificity of 0.716 (95% confidence interval: 0.605, 0.811) for the recurrence. Significant predictive factors for disease-free survival were 3D SV0HU greater than or equal to 0.54 cm3 (hazard ratio, 6.61; P = .001) and lymphatic and/or vascular invasion derived from histopathologic analysis (hazard ratio, 2.96; P = .040). Conclusion The measurement of 3D SV0HU predicted the postoperative prognosis of patients with part-solid lung cancer more accurately than did 2D MSSA and 3D MSSMPR. © RSNA, 2018.